You are on page 1of 286

ERGONOMIC

DATA FOR
EQUIPMENT DESIGN
NATO CONFERENCE SERIES

I Ecology
II Systems Science
III Human Factors
IV Marine Sciences
V Air-Sea Interactions
VI Materials Science

III HUMAN FACTORS


Recent volumes ,n this series

Volume 15 Human Detection and Diagnosis of System Failures


Edited by Jens Rasmussen and William B. Rouse

Volume 16 Anthropometry and Biomechanics: Theory and Application


Edited by Ronald Easterby, K. H. E. Kroemer, and Don B. Chaffin

Volume 17 Manned Systems Design: Methods, Equipment, and Applications


Edited by J. Moraal and K.-F. Kraiss

Volume 18 Women and the World of Work


Edited by Anne Hoiberg

Volume 19 Behavioral Treatment of Disease


Edited by Richard S. Surwit, Redford B. Williams, Jr.,
Andrew Steptoe, and Robert Biersner
Volume 20 Tutorials on Motion Perception
Edited by Alexander H. Wertheim, Willem AWagenaar,
and Herschel W. Leibowitz

Volume 21 Human Assessment and Cultural Factors


Edited by S. H. Irvine and John W. Berry

Volume 22 The Acquisition of Symbolic Skills


Edited by Don Rogers and John A. Sloboda

Volume 23 Role Transitions: Explorations and Explanations


Edited by Vernon L. Allen and Evert van de Vliert

Volume 24 Aging and Technological Advances


Edited by Pauline K. Robinson, Judy Livingston, and James E. Birren
Volume 25 Ergonomic Data for Equipment Design
Edited by Heinz Schmidtke
ERGONOMIC
DATA FOR
EQUIPMENT DESIGN
Edited by
Heinz Schmidtke
Institute of Ergonomics
Technical University of Munich
Munich, Federal Republic of Germany

Published in cooperation with NATO Scientific Affain Division

PLENUM PRESS· NEW YORK AND LONDON


Ubrary 01 Congress Cataloging in Publication Data

Main entry under title:

Ergonomic data for equipment design.

(NATO conference series. III, Human factors: v. 25)


"Proceedings 01 8 NATO ARion ergonomic data for equipment design, held MarCh
22-26,1962, in Munich , Federal Republic 01 Germany"-T.p. versp.
Bibliography: p.
Includes indexes.
1. Human engineering-Congresses . 2 . MSrl-ITI8c hine systems-Congresses. 3 .
Engineering design-Congresses. I. Schmidtll:e, Heir"lZ. N. Series.
TA166.E695 1965 620.6 ' 2 84-22894
ISBN 978-1-4684-4906-8 ISBN 978-1-4684-4904-4 (eBook)
DOl 10. 10071978-1-4684-4904-4

Proceedings 01 a NATO ARion Ergonomic Data for EQUipment Design,


held March 22-26, 1982, in Munich, Federal Republic 01 Germany

© 1964 Plenum Press, New YOf1(


A DivisiOn 01 Plenum Publishing Corporation
233 Spring Street, New YOf1(, N.Y. 10013

AI rights reserved

No part of this book may be reproduced , stored in a retrieval system, or transmitted


in any form or by any means, electronic, mechanical . photocopying, microfilming.
recording, or otherwise, without written permiSsion from the Publisher
PREFACE

For many years ergonomists and human engineering specialists


have made significant contributions to the solution of problems
faced in the area of human labour and to the introduction of their
research results and field experience into the process of equipment
design. However, the rapid increase in complexity of equipment in
use as well as in development demonstrates the necessity of broaden-
ing the point of view continuously. The workshop held in Munich
from March 22nd to March 26th, 1982, was an excellent opportunity
for the participants to discuss their respective interests and
their interpretation of needs for future research.

The workshop was sponsored by the Human Factors Special


Programme Panel of the Scientific Affairs Division of NATO. This
sponsorship, together with the helpful assistance rendered by
Drs. Bayraktar, Kroemer, and Sanders, is gratefully acknowledged.

This volume contains the papers presented during the workshop.


All these papers are directly related to the general aim: the ex-
change of experience collected in the field of ergonomic data for
equipment design on the one hand and the definition of unexplored
areas on the other. It is hoped that this presentation will help
to define future research methods in the area of ergonomic data
and set into motion fruitful discussions on the validity of the
data in use today.

No meeting of this nature could be organized without the help


of many people. In particular I would like to thank the members of
the Institute or Ergonomics of the Technical University Munich and
Dr. Robert Andrews of Plenum Publishing Company, London, for
agreeing to publish the proceedings of the workshop.

Heinz Schmidtke

v
CONTENTS

Ergonomics and Equipment Design 1


H. Schmidtke

System Ergonomics as a Basic Approach to


Man-Machine Systems Design 15
B. DOring

Task Analysis in System Ergonomics 31


H. Bubb

Generation of Ergonomic Data and Their


Application to Equipment Design 57
R. Bernotat

Some Limitations to Ergonomic Design 77


J. Moraal

Anthropometric Reference Systems 93


H. W. Jiirgens

Engineering Anthropometry 101


K.H.E. Kroemer

Basic Data for the Design of Consoles 115


H. -P. Riihmann

Ergonomic Data for Console Design 145


W.H. Cushman

Ergonomic Data for the Design of Body


Support . • • . • • • • • . 159
H. Schmidtke

The Design and Location of Controls: A Brief Review


and an Introduction to New Problems 179
W.W. Wierwille

vii
viii CONTENTS

The Hand-Machine Interface 195


C. G. Drury

Ergonomic Database for Visual Displays and VDUs • • . • • 219


H.L. Snyder

Ergonomics of Software for Visual Communication • • • • • 235


A. van Meeteren

Towards an Ergonomic Design of Software Tools • . . • • . 257


w. Dzida
Munich Theses of Ergonomics 275
K.H.E. Kroemer

Name Index 277

Subject Index 281


ERGONOMICS AND EQUIPMENT DESIGN

Heinz Schmidtke

Institut fur Ergonomie

Technische Universitat MUnchen, MUnchen/FRG

As far as is known, the term "Ergonomics" was for the first


time applied by the Polish scientist W. Jastrzebowski dur,ing 1857.
Although ergonomics may thus be regarded as a relatively new scient-
ific discipline, it has undergone an astonishing development over
the past few decades. A rapid rise in the standard of education,
which in turn increased the general level of expectation of the
population in industrialized countries, gave rise to a growing real-
ization for the need to introduce more humane working conditions.
The words: "Humane working conditions" are used here in the sense
of excluding psychophysical over-extension, on the one hand, and,
on the other hand, as a preventitive against the misuse of human
labour as a "uni-purpose machine".

Efforts to ease over-extension in the sense of excluding the


external health - hazardous influences on work may be achieved by
means of varied precautions. The introduction of an eight-hour shift
after the First World War whereby working hours were reduced, was
one of the first precautions to be effectively applied. A develop-
ment running parallel to the shortening of working hours was the
rapid proce~s of mechanization, resulting in the substitution of
human power by the machine. It nevertheless resulted in the mechan-
ization of production, so that continuous assembly line production,
as distinguished from the intense division of labour, gradually do-
minated over the hitherto applied method of single-part production.
A consequence of division of labour is however, that an exclusive
physical demand on a person may lead to a progressive meaningless-
ness of his occupation. It was especially a growing meaninglessness,
in the sense of a dimunition of the job content in question to a few
hand movements, which led to dissatisfaction in recent times amongst
employees and to a restructuring of human labour in a number of
coun trie s •
2 H. SCHMIDTKE

"Overload by means of underload" in assembly line work however


merely constitutes the one side of the coin. The reverse side re-
flects progressive mechanization and automatization, which took place
escpecially since the great advancement of microelectronics in the
fields of production and management which, in turn, led to the use
of new production processes operated automatically~ the people, so
to speak, as monitors of a higher order, merely fulfilling a con-
trolling function. This form of emptiness of work often goes hand
in hand with interpersonal isolation at the place of work. Less and
less people working in exceedingly larger working areas, produce
more and more end-products. The psychological strain factor, result-
ing from isolation, leads, similarly to physical strain limited only
to a few muscles, to increasing dissatisfaction amongst people with
their work.

In the production- and administration processes, mechanization


and automatization have lightened human labour on a broader basis.
Rationalization measures frequently result in an increase in the
pressure of performance. Time pressure and new equipment and ma-
chines, being frequently complicated to understand and to operate,
lead to psychological stress. In addition to these already-mentioned
adverse circumstances looms a fear amongst many people of loosing
their jobs as a result of rapidly encroaching technology.

Where is the place for ergonomics in this scenario? May ergo-


nomics rightly be defined as a science of human occupation within
fixed, cultural, social and physical surroundings? Is it in a po-
sition to - or should it be its task to make work more bearable?
An occupation may be regarded as bearable if it does not, also not
on a long-term basis, result in an impairment or danger to health,
regardless whether such impairment is of a rhysical or psychologic-
al nature. For the prevention of an impairment to health it is im-
portant that the authorities, responsible for defining the required
standards of achievement, have sufficient information on the limits
of continuous performance at their disposal. Practical marginal
standards of bearableness have evolved in the science of ergonomics
for the area of physical stress suffered as a result of muscular
exertion and environmental influences (e.g. climate, noise, vibra-
tion etc.). The aforesaid standards have found successsful appli-
cation in practise as well as recognition in various countries by
means of legislation, regulations and agreements entered into bet-
ween employer- and employee organizations. No similar standards of
bearableness have to date evolved in the joint area of informatory
or psycho-mental stress. There remains, as before, a wide scope
for future research -in this field.

In addition to its contribution towards solving the degree of


bearableness of work, it is the task of ergonomics to define such
parameters, through scientifically substantiated research, influenc-
ing the acceptability of work by groups as well as individuals.
ERGONOMICS AND EQUIPMENT DESIGN 3

Injury-free work should not necessarily be regarded as being reason-


able.

From the exclusion of pathological operational sequences in no


way follows that certain work is acceptable to employee groups or
that it is regarded by individuals as being beneficial to the de-
velopment of their personality. The muted dissatisfaction of many
people over their working conditions present an indication of our
lack of knowledge, even at this present age, on the type of factors
determining reasonableness and satisfaction. A field of exploration
is opened to sociological-orientated ergonomists, the importance of
which can only grow.

Whilst primarily ergonomists with broad experience in physiology


and occupational health are called for in order to determine degrees
of bearableness and whilst ergonomists with a social-scientific
background are normally involved in determining reasonableness and
individual satisfaction, engineering-orientated ergonomists are
burdened with a task of combining various sciences. Working places,
production factors and machines, the physical working environment
and the organization of labour are thus designed and organized by
engineers. As a rule, the design criteria taken into account include
economy in the sense of cost effectiveness, and degree of system
effectiveness, safety in production and operation and adaptability
to varied production aims. These design criteria, which are undoubt-
edly of importance, do however not always coincide with the op-
portunities and needs of those using the working places and machines.
user~optimization and degree of system effectiveness do however not
necessarily have to stand in exclusion of each other. The effective-
ness of a man-machine-system does not only result from the re-
alized technical standard of the machine. The extent to which the
realized or realizable technical standard integrates in the psycho-
physical constellation of the subsystem man, is equally decisive.

The most important task of a hardware-orientated ergonomy re-


volves around the integration of data on human beings into the
equipment design process. It is in this sense that ergonomics pro-
vide a service function for the design engineer. It cannot reason-
ably be expected from a designer to have mastered wide areas of the
available research and information on the combined field of human-
istics. The fulfillment of service functions does however not always
present an easy task to the ergonomist as new questions or problem
areas may arise during the development of new technical systems
for the treatment of which exist as yet no special data. A trans-
fer of findings from system A to system B, which was constructed
either by means of differing technology or in a different environ-
ment, is however only possible if it takes place under certain re-
strictions. This may be elucidated by means of the following ex-
ample: The cooperation between engineers and ergonomists may have
led to the development of a driver's seat of a motor vehicle with
4 H. SCHMIDTKE

an optimal capability of damping the critical natural frequency of


the human body in relation to the vibrational impulse emanating
from the road. By now transferring the aforementioned seat, opti-
malized for a motor vehicle, to a vehicle operating on a construction
site or to a tractor, one could achieve the reverse result of what
is originally intended namely, that resulting from differing environ-
mental factors, the user of the seat may achieve an amplification of
vibration instead of a damping in the critical frequency range. In
the long run this could impair the health of the user in the form
of signsof abrasion of the inter-vertebral plates of the spine.

The effectiveness of advice on ergonomical design criteria is


largely dependant on the advice being sought at a relatively early
stage of the development of a technical system. Corrective ergo-
nomics, applied with the purpose of eliminating deficiencies in ex-
isting systems, normally remain ineffective insofar as corrections
prove to be expensive and the results achieved often being complete-
ly out of proportion to the financial involvement required. The ef-
fectiveness of design advice is furthermore dependant on the ergo-
nomist being in a position to transform his know-how und experience
into approximate values and areas of tolerance in such a way that
they can have direct application in the design process. It is of
little use to the designer to receive information on what is for-
bidden, whilst being denied access to what is possible.

Approximate values and areas of tolerance, purporting to be


more than mere expressions of opinion, should be scientifically
tested. The Federal Republic of Germany has for instance introduced
legislation, regulating the cooperation between employers and
employees (Betriebsverfassungsgesetz) and requiring proven ergo-
nomical principles to be considered in the formulation of labour
agreement clauses dealing with the place and conditions of employ-
ment. The German legislator has however failed to define the cir-
cumstances under which ergonomical findings are deemed to have been
scientifically tested. If this discussion is not to be restricted
to the lawmaker then a clear concept is called for between ergono-
mics as a science, on the one hand, and employers/employees on the
other hand. The wider the basis for such ergonomical design re-
commendations the easier i t is to achieve the required consensus.

However, before discussing the basis for ergonomical design


principles it is necessary to determine the criteria against which
such principles can be judged. Generally a number of criteria are
conceivable. A considerably restrictive criterion is that the task
to be performed by a human being must be executionable. Those tasks
that can occasionally (i.e. not continuously) be performed, without
impairment to health, may be regarded as being executionable. The
already-mentioned criterion, i.e. bearableness, appears to be less
restrictive as i t also takes into account, on a long-term basis,
the duration of the daily working hours as well as total amount of
ERGONOMICS AND EQUIPMENT DESIGN 5

work performed during a lifetime. The criterion of reasonableness,


being even less restrictive, would impute the acceptance of the po-
tential user of the system. The advancement of personality appears
to be that criterion which goes furthest of all. Design recommen-
dations based on the aforementioned criterion do not merely con-
sider the health and productivity of a person, but, in addition,
take full cognizance of such person's needs.

In placing these four criteria next to each other, a dilemna in


the ergonomical design recommendations becomes apparent. There is
principally no possibility to decide, with the aid of scientifically-
founded information, which of the mentioned criteria to apply. I want
to attempt to substantiate this assertion by means of the following
parable. In his book "The Philosophy of Physical Science" published
in 1939, the English astrophysician Eddington describes the relation-
ship between scientific findings and the reality of life by means of
an impressive example.

In this example the scientist is compared with an ichtyologist


studying underwater creatures. This ichtyologist throws out a net,
subsequently examining the catch according to scientific, methodical
rules: measuring, weighting and describing. Two basic findings (rules)
are reached after numerous hauls and after an evaluation of the out-
come:

all the fish exceed 5 cm in length;


- all fish have gills;

he then proceeds to draw the conclusion that the aforesaid findings


would also materialize in future hauls. A critical colleague, whom
we may, in analogy to a metaphysicist, call a meta-ichtyologist, may
very well accept the second basic finding on the premise that there
are indeed numerous fish in the sea being smaller than 5 cm. The
latter fish can obviously not be caught with a net which meshes have
a circumference of 5 cm. This argument needs not impress our ichtyo-
logist,who may retort: "Those objects which I am unable to catch
with my net, in principle lie outside the sphere of ichtyology, not
constituting those objects defined in ichtyology as such. As far as
~y research is concerned, those objects that I am unable to catch
cannot be regarded as fish."

This example may prove to be of invaluable assistance in our


discussion on the determination of appropriate criteria. In applica-
tion to ergonomics, our example may, by analogy, equate the ichtyo-
logist's net to the intellectual and methodological tools of the
ergonomist in his search ·for scientific information. The casting
and hauling of the net is comparable to scientific research and
testing in the laboratory or in the field. The difference in opinion
of the two ichtyologists is just as uncontradictory as that of two
ergonomists using differing criteria.
6 H. SCHMIDTKE

The first ergonomist (restrictive-ergonomist) limits his scope


of research to what he can scientifically prove. He needs not base
his findings on mere speculations on the direction in which the needs
of future system users would develop. The accuracy of his findings
rests on his own judgement. The bearbleness criterion formulated by
him should not have adverse results in its practical application. The
second ergonomist whom, in analogy to our aforementioned example, we
may call a meta-ergonomist, uses as a basis a working environment
devoid (or, as yet devoid) of any measurable, influence-bearing
factors, proof of which he may believe to have derived from group
behaviour. His problem is to concretely express this proof. Accord-
ing to the first ergonomist, the object has to be measurable. The
meta-ergonomist necessarily views the requirement of measurability
as an inadmissable restriction of the objective reality, namely real-
ity, as seen by him subjectively.

This discrepancy between the conception of our two ichtyologists


and ergonomists characterizes an essential state of facts, namely the
polarity between science and reality. As is the case with all other
sciences, ergonomics is not concerned with the actual reality but
merely with a projection of this reality, namely that aspect which
is determinable through efficient research methods and adequate in-
terpretation of results. Everyone adhering to this methodicism is
able to verify this aspect of reality by means of re-examination.
This knowledge is derived restrictively from the actual reality by
reason of its inherent character of projection. This fact must be
born in mind when attempting to comprehend, with the aid of scient-
ific methods, such a complex phenomenon as human labour.

Depending on the type of methodical tool, i.e. the width of


the net used to catch the required research objects, it will become
possible to deal with narrower or wider aspects of the situation
corresponding to the reality of work. Research methods are not tools
which are determinable independently of the reality a priori, but
have proven to be suitable to constant interaction with the reality.
It follows that the actual reality has a retrospective effect on the
paradigm by means of which it is strived to comprehend, describe and
assess this reality scientifically. The aforementioned differing
criteria may be regarded as paradigmae of ergonomics since the deter-
mination of a paradigm necessarily influences exten~of the area of
the statement as well as the possible depth of knowledge.

What now is the relevance of the present discussion to the area


of recommendations for ergonomical design? It may be learnt that
every recommended value is dependent on a scientifically-based para-
digm. It follows that differing scientists may arrive at different
recommended values. Since the choice of paradigm permits neither
strict reglementation nor the ability to reconstruct in each case,
the dispersion of existing recommended values in literature should
not come as a surprise. It is however of no assistance to a person
ERGONOMICS AND EQUIPMENT DESIGN 7

applying recommended values if he obtains a number of alternative


replies to a concrete question. The only solution to the afore-
mentioned dilemna in the field of ergonomics is to indicate an area
of tolerance instead of seeking to find a recommended value for each
individual case. The lower tolerance limit could then, for instance,
be based on a very restrictive, the upper limit however on a far less
restrictive paradigm. Care must however be taken that the lower to-
lerance limit is based on a paradigm orientated towards the bearable-
ness criterion.

A wide range of design recommendations are based on data de-


rived from human body mass or strength. Even the transfer of body
measurements and strength into hardware originates from a certain
paradigm. In the field of ergonomics it is common, albeit on a
broader basis, to apply specified percentile restrictions of the
frequency distribution of the measuring data. A vast amount of data,
in the form of percentile presentations, has been collected on body
measurements. A comparison has merely served to confirm everyday
experience of major differences on regional, racial and social
grounds which exist concurrently with differences between the sexes.
It follows that critical problems of transferability are bound to
arise in the case of products with a, geographically-speaking, wide-
ly-spread market. The criterion, being unrestrictively applicable
to a predetermined geographical region, i.e. 5 % or 95 %-criterion
(or, for safety-orientated purposes, the 1 % or 99 %-criterion), can
as a result no longer be applied. A switch-over to a general 50 %-
criterion, i.e. to a mean value, brings the matter hardly any fur-
ther. In order to circumvent the outlined problem, it becomes ne-
cessary to devise a new paradigm. However, there are not even signs
of any recommendations to date.

The situation regarding data on body strength is even more crit-


ical. Whilst there is ample and reliable information on special po-
pulation groups (e.g. pilots in the US Air Force), the transmission
of power by the latter-mentioned group can only be exercised re-
strictively, form and direction-wise, in an enclosed area. Apart from
a few recent cases there exist no percentilized data of strength for
the population in a defined area.

The present state of ergonomics, as service institution for the


benefit of designers and production engineers, is characterized by
uncertainty in the choice of decision criteria constituting the bases
of design recommendations as well as possible workload. There is no
doubt that it is out of the reach of one or a few ergonomists to
breach this gap. Even the availability of textbooks, reflecting ap-
proximate values, does not bring us any further, for they are normal-
ly silent on the mode of origin of values, their reliability and
validity. The presumption that some of the data collected constitute
operationally - defined data (which need thus not necessarily be non-
sensical but are not verifiable) hence being rebuttable. Pragmatic
8 H. SCHMIDTKE

definitions, based on expert experience, may well prove to be help-


ful and necessary for a narrowly defined field of application or for
a technically completely new development. Definitions, as aforesaid,
developed by means of copied data being taken over from an existing
set of data, may however result in illusionary validity which, in
turn, may give rise to false hopes. The situation becomes especially
critical if decisions, based on illusionary validity, are made to
integrate such data into national or international standards. En-
deavours in this direction are common. It is true that standards are
not created to last eternally. However, the selfdynamics should not
be underestimated, once publicized, expecially not if it becomes ne-
cessary to attempt to effect changes within international bodies
(e.g. ISO).

The past few years have seen efforts on the part of national as
well as international institutions to develop ergonomical standards.
These efforts underly various philosophies. In the Federal Republic
of Germany the prevalent view is to distinguish between appliances
or object-related standards, on the one hand, and basic standards,
on the other hand. According to this view, ergonomical standards can
only be categorized under basic standards. The expert institutions,
dealing with the creation of standards, consequently take great care
not to follow a collision course with the standards of equipment
but rather to advise them thoroughly. The validity of the data basis
of ergonomic standards constitutes the main problem area, notwith-
standing the kind of philosophy applied, on which the task of stand-
ardization is based. It requires only little imagination to suspect
that users of pseudo-valid data undertake everything possible to
convert their data records into standards, particularly under such
circumstances where this data has already been used in the manufact-
ure of equipment. In view of the fact that a modification of basic
standards will necessarily lead to a redesign of equipment and tools
it is no wonder to find manufacturers defending their data basis ve-
hemently.

The contributions to follow are aimed towards highlighting the


more important, if not always most spectacular, problem areas found
in the design of production equipment and of introducing available
information on the subject matter. Special problems, especially
those being meaningful to a great number of people, were furthermore
deliberately drawn into the discussion.

In discussing single problem areas occurring in the technical


organization of labour, care should be taken not to create the im-
pression that all problems in the humanization of labour can be
solved merely through the application of ergonomical data to the
organizational process of machinery and installations. The appli-
cation of ergonomical data to the construction of production factors
plays only a relatively marginal role in the sphere of human work-
ing conditions.
ERGONOMICS AND EQUIPMENT DESIGN 9

According to the classical findings by Roethlisberger and Mayo,


an ergonomist, even if considering such a marginal role to be im-
portant, should not overestimate its function. From an ergonomical
point of view many designs may be regarded by the user of production
factors to be highly energy-consuming to a person, and to result in
a great strain on the sense organs or on his concentration. Optimal-
ly designed production factors will prove hardly successful, if
utilized in business with poor interhuman relationships, massive
achievement- and work rate pressures, employment structures that
discourage even the slightest tendencies of a free development of
personalities, extreme divided labour and intellectually under ex-
tended employees. Whilst it was initially stated that the rise in
educational standards and in general expectations resulted in a
modification of the demand structures of the working person, it should
constantly be kept in mind that ergonomical design recommendations
can only play a minor role in the adaptation of general working con-
ditions to such modified demand structures.

The functional rating of ergonomical design recommendations


however becomes a more important factor if applied towards the im-
provement of safety precautions at work. The conclusion can be
drawn from depressing, available statistics on accidents that a large
number of deaths or serious accidents could have been avoided, if
potentially faulty operations, whether anticipated or not, would
initially have been considered. The ergonomists' recommendations on
safety precautions in the construction field however often remain
unheaded, as he can normally only offer general rules instead of
concrete data. Future research in this field should be aimed towards
replacing general rules for defined fields of application with con-
crete and technically realizable data. It is correct to state that
as a general rule external forces (vibrations) may not result in the
unintended switching on of machinery. It would however be of greater
assistance to a designer if the following demand would appear in re-
lation to a concrete situation: "Switching devices for the switching
on of electric, hydraulic, or pneumatic energy of of movable parts
have to remain safe up to x cm/s 2 in the case of acceler.ation or
up to y cm/s 3 , in the case of acceleration changes, whilst remaining
in an off position." This demand enables the designer to choose those
control systems which operate safely under field conditions.

A considerable rise in the technical standard of a number of


places of employment has resulted in the increasing fulfilment of
part-functions by accessory appliances. Such a rise in the standard
of eqUipment necessarily results in a greater utilization of controls
and displays. This is evidenced by commercial airplanes or control
rooms, e.g. of nuclear power plants which are crammed with displays.
By comparing the radar equipment normally in use during the sixties
with those on the market at present, it is found that. for the marital
surveillance and position finding functions of the radar equipment
of the sixties merely required 20 controls. TOday 60 push-buttons,
10 H. SCHMIDTKE

switches and press keys are required for equipment fulfilling a si-
milar function, ARPA-radar even requiring controls in excess there-
of. This development is not only less beneficial to users but, in ad-
dition, promotes control errors with possible critical, security-
orientated consequences.

An increase in the use of highly sophisticated equipment need


not necessarily go hand in hand with increasing safety at work, but
may actually result in a dangerous decrease in the degree of safety.
The nature of technical equipment is frequently suggestive of the
user being able to apply such equipment within clearly defined
limits. In practise such limits are usually not set by the technical
equipment. The necessity arises for the ergonomists to counter the
endless addition of new subsystems in his construction proposals.
For safety sake he should endeavour to attain a higher degree of the
technical integration of subsystems. The compulsion towards achieving
mental integration, being still common today, promotes errors, there-
by jeopordizing the safety of the user.

The task of achieving a technical integration of subsystems is


today mainly fulfilled by computers. It is, by their very nature,
only possible for computers to receive instructions via an artificial
language. Present computer language is essentially different from
colloquial language despite developments over the past few decades
whereby the application of computer language was greatly simplified.
The efforts being made in the development of user-orientated lan-
guages in the field of software-engineering should not be under-
estimated. This new area, for which the term "software-ergonomics"
has been adopted, has nevertheless opened a new field of application
for ergonomics. Despite the considerable amount of information gained
through research into artificial intelligence and the psychology of
cognition and problem solving, ergonomics is by and large faced with
virgin territory. In order to fulfill their function efficiently and
safely, programming languages as mental equipment should be optimally
adaptable to the user, as is the case with hardware elements. It is
no easy to detect programming errors resulting from highly compli-
cated programme languages and highly structured computer programmes.
Such errors may however have catastrophical consequences in the case
of dynamic systems, more especially in the case of weaponry systems.
Hence a much more intensive research effort should be encouraged and
supported in the field of software-ergonomics than has been the case
so far.

We finally wish to consider the principles underlying ergonomic-


al design recommendations. Although a number of these principles have
already been mentioned, it is necessary to discuss them systematically.

Ergonomical design guidelines cannot function independantly.


They must consequently and as matter of principle be orientated to-
wards a person as the user of equ~pment.
ERGONOMICS AND EQUIPMENT DESIGN 11

This orientation should strive towards the following aims:


- to safeguard the health of a person;
- prevent him from overload;
- to encourage the lifelong development of his personality;
- to facilitate safe employment conditions, thereby protecting him
and other people from injury.
Running parallel to the aforementioned aims and relating to
single persons only, is a further aim, namely to warrant a display
of performance meeting economically based requirements, i.e. both
quantitatively as well as qualitatively.
At a first glance it may appear that both classes of functions
oppose or even exclude each other. Upon closer scrutiny, it however
becomes clear that this is by no means the case. Both the costs
arising out of impairments to health and accidents are born by the
general public and, more especially, financed out of the national
economy, thereby representing a considerable cost factor. Whilst
employees with higher qualifications and broader training constitute
admittedly a higher cost factor than less qualified ones, thereby
placing a burden on the money supply, it is likewise true that per-
sonal qualifications represent underestimated potential capital.
The wider the scope of training of the employees, the easier it be-
comes for them to adapt to changing production conditions, hence
of surviving the economic battle for a share of the market. The
failure of having found suitable means of applying the factor "per-
sonnel qualification" to a system of cost calculation and of assess-
ing it realistically, should be regarded as a serious shortcoming
of economical science.

Using the mentioned aims as a premise, the integration of er-


gonomical findings into the design process, based on known findings,
may be achieved by conSidering the following principles:

1. Equipment, tools and machinery should be optimally adjusted to


the body-measurements of the potential user-population. This
principle is only realizable if all equipment can be operated by
users of the 5 th and 95 th percentile without having to assume a
non-phYSiological forced posture. The measurements of equipment,
which may constitute an impairment to life and safety, should be
suited to users between the 1st and 99 th percentile and, as far
as necessary, be adjustable to individual body measurements. Re-
gional and sexual differences should be taken into consideration.
2. Equipment, requiring for its operation or use that resistance be
overcome through muscular power, should during occasional use not
exceed the positional strength of the 5 th percentile of the user-
population. In the case of security-orientated equipment, the ad-
missable strength should be reduced to such an extent that it
could be exercised by persons with maximum positional strength
corresponding to that of the 1 st percentile.
12 H. SCHMIDTKE'

3. If a specific task requires physical work to be frequently carried


out by large muscular groups of the body for a long duration, the
technical equipment should be designed in such a manner that the
physical performance sought to be achieved by males does not ex-
ceed the following values:
Pull and push task:
approx. 70 watts (n approx. 24%, biol. performance 300 watts).
Carrying Loads:
approx. 50 watts (n approx. 17%, biol. performance 300 watts).
Heaving loads:
approx. 25 watts (n approx. 9%, biol. performance 300 watts).
Spadework:
approx. 15 watts (n approx. 5%, biol. performance 300 watts).
In the case of females the aforementioned values should in each
case be decreased by approximately 30%.

4. Technical design should as far as possible be applied in such a


manner as to avoid the subject having to hold something (static
muscular work) as this results in rapid fatigue and in dispro-
portionately long fatigue allowances.

5. In designing places of work for monitoring and control functions


it is important to take into consideration that human capacity
to take in and process information is somewhat limited. In combin-
ing subsystems technical integration should, as far as possible,
be given preference over mental integration.

6. As far as a person is required to perform control functions,


cognizance should be taken that his upper limiting frequency is
substantially lower than is the case of technical control systems.
Control performances of ~ 1 Hz generally already constitute an
overstrain.

7. Apart from the upper limiting frequency, the quality of manual


control performance largely depends on the degree of order of
the system which is sought to be controled. Raising the order of
the system above a first order system, in general adversely af-
fects performance and thereby also security.

8. In designing technical systems it is important to note that the


transition from a linear to a non-linear system behaviour is ex-
tremely critical for the user of the system as he generally finds
it impossible to create internal models out of non-linear system
behaviour. Information on a transition in the non-linear area may
lead to a considerable improvement in the safety of work.

9. Technical equipment being adaptable to varying working conditions


should be designed in such a way that the switching sequences, af-
fected by the working conditions, may be regulated by means of
separate controls. This is the only way of preventing technical
ERGONOMICS AND EQUIPMENT DESIGN 13

malfunctioning resulting from memory lapses. The development of


systems for user guidance plays an important role in this regard.
As a rule neither operating guide nor instruction manual con-
stitute the optimal solution.

10. Highly automized technical systems no longer have the operator in


a closed loop. The misuse of a person in his position as monitor
of a higher order, invariably leads to the breakdown of his vigil-
ance.

The list of the basic principles of ergonomics is most probably


incomplete. Every properly qualified and experienced ergonomist is in
a position to supplement the said list. A considerable contribution
could however be made towards the improvement of working conditions
if the embodyment cf the aforegoing principles into the process of
equipment design has taken place. In reality the meshes of the con-
struction network are however in many places so wide that even the
few mentioned principles are threatened of falling through. It should
be the permanent task of ergonomical conferences, workshops and
symposia to point out this state of facts to those responsible for the
development of equipment.
SYSTEM ERGONOMICS AS A BASIC APPROACH
TO MAN-MACHINE SYSTEMS DESIGN

Bernhard DOring

Forschungsinstitut fur Anthropotechnik

Wachtberg-Werthhoven/FRG

INTRODUCTION

With respect to machine components of current industrial and


military systems, e.g. with nuclear power plants, modern ships and
aircrafts, or modern weapon systems, two quite recent trends have
become noticable: increasing system complexity and increasing use
of automation, usually resulting from the application of computers.
Consequently, the computer is sharing key functions with the man.
This in turn changes the nature of tasks which have to be performed
by the human operators of those systems. Instead of predominantly
perceptual-motor tasks which personnel had to perform previously,
operators now are increasingly involved with monitoring, management
and decision tasks (Vossel et al., 1977).

Generally, systems in which personnel and technical equipment


are working together to fulfill common goals or objectives are
called man-machine systems (MMS) (Hoyos, 1974) or socio-technical
systems (De Greene, 1973). With these increasingly complex systems,
design problems become rather complex. On the other hand, more com-
plex MMSs are able to perform their required functions effectively
only, if both machine component characteristics and the capabilities
and limitations of personnel have been properly matched with each
other as a result of the whole system development process. In order
to determine and match those characteristics it will be necessary
to properly allocate design responsibilities among system engineer-
ing, systems ergonomics, and other disciplines contributing to the
design of MMS. In this presentation attention will be focused on
the contributions of systems ergonomics. But before describing ergo-
nomic contributions it will be helpful to consider the structure of
a MMS in some detail.

15
16 B. DORING

A MMS can be defined as a functionally organized assemblage of


machines, man, and the processes by which they interact within an
environment to produce some desired system outputs (Meister, 1971).
Generally, a MMS can be considered as a combination of one or more
human elements and one or more machine elements whereby these ele-
ments are regarded as system components or subsystems (Fig. 1). In
this frame of reference the concept of "machine" is considered to
consist of virtually any type of physical object, device, equipment,
facility, thing, etc., that people use in carrying out some activity
directed towards achieving some desired purpose or in performing some
function (Mc Cormick, 1976). Relationships between machine and human
system components can be material, energetic, or informational in
nature. Relationships also exist between the system and its environ-
ment. With an aircraft, for example, the external environment con-
sists not only of the geographical and meteorological surroundings,
such as weather, terrain, atmospheric conditions, etc., but also
other associated MMSs such as air traffic control, airport systems,
etc. Usually the boundaries between a system and its environment
depend on the problem to be solved and therefore can be defined only
arbitrarily. One property of system components is that they may be
regarded as belonging either to the system, or to the systen environ-
ment. Another important property of system components is that they
can be regarded as systems themselves and hence as subsystems of a

Environment

Fig. 1. Scheme of a man-machine system


SYSTEM ERGONOMICS IN MAN-MACHINE SYSTEMS DESIGN 17

larger system. This leads to a hierarchical structure of the problem


area.

SYSTEMlHRGONOMICS AND SYSTEMS LIFE CYCLE

The matching and integration of machines and personnel into an


effective MMS is the most important contribution of systems ergono-
mics. Systems ergonomics is concerned especially with the develop-
ment of ergonomic analysis and design methods and their application
to the solving of MMS problems which arise during systems design. To
accomplish that, knowledge of diverse disCiplines have to be taken
into consideration. Already in 1967, Singleton and his colleguages
(1967) depicted a scheme in which the diverse disciplines that con-
tribute to systems ergonomics and thus to systems design are shown
(Fig. 2). They stated that systeusergonomics is comprised of "Ergo-
nomics and Systems Design". By ergonomics they meant the parts of
psychology and anatomy which are relevant to the study and design of
human tasks, workplaces, machines and environments. By system design
they meant an organized approach to the business of decision-making
in any design context with what they regarded as a proper emphasis
on human factors. This combination of disciplines is characteristic
of the systems ergonomist who must in fact be knowledgeable about on
both fields. On the one hand he must know enough to communicate ef-
fectively with other design engineers and decision-makers, and on
the other hand he must stay current with his special ergonomist know-
ledge by being able to understand the work of other life scientists.

But his relationship to this latter scientific group is not a


dependent, but rather an interdependent one; he has an invaluable
part to play in the stimulation and orientation of academic research
on human performance. The most important consequence of this perfor-
mance orientation is that the speCification of system performance
objectives become the starting-point for any human factors problems
to be solved in the context of a given MMS.

Another characteristic of the systems ergonomist is that he is


concerned with the whole system, not just the man-machine interface.
Earlier, ergonomic activity was confined to man-machine compatibility
in terms of display and control design and environmental factors such
as lighting, noise, temperature, humidity, acceleration, vibration,
ect. Today, system ergonomics is involved earlier in the system ac-
quisition process with questions of level of automation, allocation
of functions among men and machines and basic manning. Also, they
are now involved with the task of "fitting of the man to the job",
i.e. selection and training which previously was the exclusive domain
of occupational or personnel psychologists.

Still another characteristic of system ergonomists is that they


are not restricted to operational functions. They must allocate and
00

SYSTEMS DESIGN

Other
Ph,.ical ANALYTICAL
Science. TECHNIQUES

To all To all
Ph,.ical Scienc .. Hu.an Science.

To all anal,tical technique.

APPLIED MATHEMATICS !JI


c
0-
Fig. 2. Disciplines contributing to the design of oman-machine systems (Singleton et al., 1967) :::u
Z
G)
SYSTEM ERGONOMICS IN MAN-MACHINE SYSTEMS DESIGN 19

design for maintenance, logisticsand other support functions. During


systems design, systems ergonomists have to look at all human factors
problems from an overall systems viewpoin~ keeping in mind the possi-
bility of optimizing trade-offs which can be achieved among all func-
tions, components and system criteria.

The development process of a MMS can be formally partitioned


into separate time phases. This partitioning, although arbitrary and
general, is very useful for associating systems ergonomics activities
with the systems life cycle. Similar schemes of partitioning can be
found in the literature, e.g. AFSCM 80-3 (1966), Chapanis (1970),
Christensen (1971), Coburn (1973), Daenzer (1979), Meister (1971),
Shackel (1971). All these schemes show more or less the following
division of the system life cycle:

- development phase
o concept
o definition
o design

- realisation phase
o production
o introduction

- operational phase.

The development phase can be divided further into concept, de-


finition, and design subphases. During these subphases a preliminary
study, a main study and detailed design usually are performed (Haber-
fellner, 1979) (Fig. 3). During the preliminary study the overall
problem is identified and analyzed to decide whether a new system is
required or not. If a new system is required, general system goals
and constraints have to be determined. Based on this information,
fundamental alternate solution concepts and their practicality are
developed and compared. Then the best approach for solving the pro-
blem is selected. For example, if the overall problem is to counter
large enemy tank forces, does this call for a new tank, tank de-
stroyer, anti-tank helicopter, attack aircraft, or portable anti-
tank missile system? In the main study alternative system concepts
for the selected solution approach are developed. For example, if
a portable anti-tank missile system is selected, alternatives could
be a self-guided or a man-guided missile (via wire) based on IR,
LLLTV, radar or optical sensors. During the detailed design sub-
phase, analysis of critical components and detailed hardware and
software design of subsystems are accomplished. Alternative system
designs are compared with each other and the design which best meets
established system goals will be selected for realisation.

The realisation phase comprises the production of the system


and its introduction or deployment to the user. System production
20 B. DORING

------

> __
------
~"- TERMINATE
STUDIES ...z
w
I. ____ _ ~
w
I <!)
«
I z
«
...u
~

-- w
.....
o
0::
a..

REALISATION
------
PHASE
------
OPERATIONAL
PHASE

Fig. 3. System life cycle (Haberfellner, 1979)

means e.g. the manufacturing of the system hardware, machine compo-


nent integration, testing and evaluation as well as the generation
of manuals and job aids for system operation and maintenance. Addi-
tionally, during this time system personnel required can be selected
and trained. Thus, when the system is introduced towards the end of
this phase, all system components, i.e. machine components and per-
sonnel, are available, so that final system test and evaluation can
be accomplished.

Sometime during the operational phase of a system, its growing


obsolescence may indicate a need for system update or retirement,
whereas the latter usually leads to development of a new system. How-
SYSTEM ERGONOMICS IN MAN-MACHINE SYSTEMS DESIGN 21

ever, in both cases the system life cycle just described starts again
with a new development phase in which new problem situations and
technical solution principles have to be considered. The system ac-
quisition process described is controlled bythe.project management
indicated by the box on the right side of Fig. 3.

Because the largest impact of ergonomics on system performance


and life cycle costs occurs during the essential system design
efforts in the development phase, systems ergonomists should partici-
pate as early as possible in that process. On the other hand, if er-
gonomists are first brought in during the realisation or operational
phase, the best which can be expected is identification of critical
problems and only minor improvements of the system at relatively
high cost and deployment delays. Consequently, the effectiveness of
ergonomics is severely restricted when system redesign has to be
accomplished.

A GENERAL APPROACH APPLICABLE FOR SOLVING DESIGN PROBLEMS

As system development proceeds through the development phase


(Fig. 3), design problems have to be treated in increasing detail.
Also, additional details about system requirements and constraints
become known and affect the development process. Characteristic of
the system development process is a general problem. This problem
solving approach can be applied throughout systems design on in-
creasingly detailed levels, during each sub-phase, and for every
subsystem and component design effort. This approach can and should
be used for both engineering and ergonomic efforts. This general
problem solving approach (Fig. 4) consists of the following three
major steps (Haberfellner, 1979):

- goal elaboration,
- solution generation, and
- solution selection.

During the first step (GOAL ELABORATION) the problem area is


structured, i.e. the boundaries between the system, subsystems,
components, etc. and their environments are established. Here at
the level of detail is depending on the problem level under con-
sideration. This structuring is done by analyzing the problem si-
tuation, i.e. the missions, functions, tasks, task elements, etc.
that have to be fulfilled. Again the level of detail depends on the
problem level. With this analYSis, requirements and constraints
relative to missions, system performance, costs, type and number of
personnel, time aspects, applicable technologies, physical and in-
formational system environment, etc. are determined. These data con-
stitute the informational frame for decomposing and formulating de-
sign goals down to the level of functional requirements which are
the basi s for the next step.
22 B. DORING

GOAL
ELABORATION

SOLUTION
GENERATION

SOLUTION
SELECTION

Fig. 4. General problem solving approach (Haberfellner, 1979)

The next step is SOLUTION GENERATION. It consists of solutions


synthesis and analysis. During synthesis a sufficiently broad spec-
trum of alternative design concepts for eac~ engineering and ergo-
nomical design problem specified should be generated, out of which
the optimal solution can later be selected. Analysis is hard to
separate from synthesis because normally each synthesis activity
is accompanied intuitively, and perhaps unconsciously, by some ana-
lysis activity. But here we are talking instead about a formal and
conscious analysis performed after major and important design results
have been synthesized. During this analysis each concept generated
is examined with regard to technical and ergonomic design goals in
order to eliminate inadequate solutions and identify unsatisfactory
ones which may later be improved. From the ergonomical viewpoint it
is especially important to evaluate the performance demands on system
personnel associated with alternative design concepts during this
analysis. Concepts for which demands exceed personnel capabilities
and limitations are rejected or redesigned.
SYSTEM ERGONOMICS IN MAN-MACHINE SYSTEMS DESIGN 23

The last step in the problem solving approach is SOLUTION


SELECTION. In this step, previously generated design concepts are
evaluated with regard to their ability to fulfill design goals sti-
pulated in the first step. In this connection detailed design goals
are regarded as evaluation criteria and assigned weighting factors
that establish their relative importance. The design goals regarded
comprise performances and costs, for both machine components and
personnel of the system. In order to have an objective means of
evaluating functions, it is desirable to specify as many criteria as
possible in quantitative terms. Assuming that all quantitative data
have been gathered for each criterion, it is possible to compare
alternative concepts with every other on the basis of each criterion.
Finally, the concept with the highest total criteria value is select-
ed. Detailed information on how to determine criterion values and
handle weighting factors are available, for instance, from Dathe
(1971), McCrimmon (1968), and Zangemeister (1971).

BASIC DESIGN STEPS OF SYSTEMS ERGONOMICS

During MMS development, ergonomists have to solve various pro-


blems at different levels of detail. Extensive descriptions of the
activities ergonomists have to perform during the development pro-
cess can be found in the literature, e.g. at De Greene (1970),
DOring (1976), Kirchner et al. (1974), Meister (1971), Schmidtke
(1981), Shackel (1974), Singleton (1974), Singleton et al. (1967).
In this presentation only a short overview of the main ergonomic
design steps during the design of manned systems will be given for
showing their interdependency.

Given that system performance requirements and system functions


have already been determined, e.g. by systems analysts, the first
problem ergonomists are confronted with are the determination of
which functions have human factors impact and the allocation of
system functions to personnel and machine elements of the system
(Fig. 5).

The FUNCTION ALLOCATION process occurs with increasing level


of detail at all stages of the development phase, i.e. from pre-
liminary planning in the conceptual subphase to final design at the
end of the system's design subphase. The desired goal with function
allocation is to produce a system configuration which can meet all
system requirements through the correct assignment of functions to
human and machine elements. Cost effectiveness criteria play an
important role during this allocation. In order to arrive at the
best system concept, different alternative allocation configurations
would have to be developed; trade-off s.tudies to assess them are
performed, and the best alternative concept selected. The alter-
natives should comprise the range between manual, semi-automatic
and automatic system configurations as well as different system
24 B. DORING

function allocation {
man - machine

l
ma'::I:"!'=~ {
estimation

1
5~~~~:!~ { ~ ...
interfae.s ~

1
workstation
{ (operOtor"""
______ --' ...... mOc:hlM-]
IL _____ _

T
INTERFACE

TEM~TURE

,
HUMIDITY ILLUMINATION
\ , ;.,NOISE
_k
environment (,.-opera....:.-:t-or~) ~ I-m"a-'ch~ine-"

T
'\
TOXIC
. . . RADIATION

VIBRATION SUBSTANCES

"I machine II
"I machine 21
.. I machine 3 1

personnel selection
and training

Fig. 5. Ergonomic design steps during manned systems design


SYSTEM ERGONOMICS IN MAN-MACHINE SYSTEMS DESIGN 25

sizes and cost ranges. Further analysis of those functions which


have been assigned to machines and the design of machine components
(with the exception of man-machine interfaces) are accomplished pri-
marily by the engineering staff. The ergonomists are mainly respon-
sible for the analysis and synthesis of those tasks that are assigned
to personnel. The analysis results are the basis of further design
tasks ergonomists have to perform during MMS development.

One of the earliest ergonomic design tasks is to make a pre-


liminary ESTIMATE OF MANNING REQUIREMENTS, i . e ., the number and types
of personnel required for system operation and maintenance are rough-
ly determined. Because systems differ considerably among themselves
the scope of such manning studies has to be appropriately scaled to
the system. If a system is rather simple, or if it is only a minor
variant of some existing system, a manning study probably is not
necessary. But with larger or more novel systems such as a nuclear
power plant, a ship, or an aircraft, a manning study should be carried
out. Again the degree of detail which has to be determined increases
as the development phase proceeds. After preliminary manning require-
ments are determined, preliminary tasks, interface requirements and
workstation concepts will be generated which may be changed if detail-
ed results of subsequent ergonomic design steps become available.
Thus, proceeding from the preliminary study where manning concepts
developed are rather elementary and crude, they are specified with
successively greater detail and accuracy until final personnel re-
quirements are established towards the end of detailed design.

In DETERMINING THE REQUIREMENTS FOR TASKS AND INTERFACES in


detail, the temporal, logical, and functional relations of personnel
tasks are analyzed and arranged. The tasks analyzed are those which
will have to be performed later on by the personnel for operating
and maintaining the machine components when the system is operational.
The results of this task analysis are on the one hand, the behavioral
task requirements for the operators, on the othe-r hand the infor-
mation and control requirements for the different man-machine inter-
faces in the system. In this context, an interface can be defined
as that part of an eqUipment which the operator manipulates or ob-
serves whenever he has to perform a task involving direct contact
with equipment. The man-machine interface in many cases turns out
to be a control panel or a console in which one or more control
panels are to be located. Other man-machine interfaces in a system,
for example, for the maintenance man, are internal components, test
points, and test equipment. All requirements mentioned have to con-
sider the mission dependent variations in the course of events and
activities that go on in the machine components as well as in the
system environment. Thus, task analysis is basic to the development
of all personnel subsystem products including human engineering,
training, personnel requirements, and manuals. Data derived in task
analysis do not generate these products directly, but they provide
the basis for subsequent ergonomic activities.
26 B. DORING

The next problem which arises in system development is the


DESIGN OF WORKSTATIONS which are places where one or more operators
work. Ideally, workstation design starts with consideration of the
requirements of the operator then the workplace and the equipment
are arranged around the man. In so doing, workstation and operator
should be regarded as a functional unit. Ergonomic design efforts
include design of man-machine interfaces, workspaces, or consoles,
job aids, and the arrangement of equipment and personnel in work
areas. Basic data for these design efforts are the behavioral re-
quirements of the personnel and the information and control require-
ments of the interfaces which were determined previously. Starting
with this data, interface components such as displays and controls
are selected or designed~ dimensions of consoles are determined
after consideration of numbers and sizes of displays, controls, and
other console components and anthropometric data of the personnel
projected~ displays and controls are grouped and arranged on the
console according to operator visual and reach areas, task sequences
and frequencies, and other human engineering layout principles. When
applicable, multiple consoles, other equipment and personnel are
arranged within workareas or facilities on the basis of mission de-
pendent links between them, maintenance concepts, necessary equip-
ment items with their dimensions, functions, and support require-
ments. Further data required for arranging ar~ equipment inter-
connections, and the number of personnel and their task.

Men and machines have to work in various environments, and the


conditions of those environments may have a pronounced effect on
the efficiency of a MMS. Essentiai ENVIRONMENTAL FACTORS that affect
task performance of personnel are, for instance, lighting, tempera-
ture, humidity, pollutants, noise~ vibration, acceleration, radia-
tion, etc. Out of tolerance values of these factors tend to reduce
personnel efficiency or worse, can lead to health problems. There-
fore, ergonomists have to consider and control these environmental
influences in their design.

Basic to this design task are, again, the behavioral task re-
quirementsestablished. Because not only must the critical values
of environmental factors recommended in the relevant literature be
considered but also with some factors task specific requirements
have to be specified and met for effective task performance by per-
sonnel.

Knowledge of tasks, work stations, and the environment are


necessary to determine in detail the number and types of personnel
required for operation of system work stations and maintenance of
system equipment which in turn is a prerequisite to detailed JOB
DESIGN. Job design consists of grouping tasks, functions, duties,
and responsibilities into jobs or positions to be performed by
different individuals.
SYSTEM ERGONOMICS IN MAN-MACHINE SYSTEMS DESIGN 27

For completeness it should be mentioned that, after the system


is designed, ergonomists are frequently involved in SELECTION AND
TRAINING OF PERSONNEL. Although the major amount of matching between
man and machine is accomplished by design of machine elements, final
matching and adjustments of individual differences in personnel are
effected by selection and training.

In explaining the systems ergonomics activities for designing


a MMS it is easier to present them in a rather logical, sequential,
step-by-step fashion as shown in Fig. 5. But I do not want to leave
the impression that the process proceeds exactly in this way. The
sequence is only approximately correct. Actually the process differs
from the given step-by-step presentation because of its iterative
nature. The general problem-solving approach (Fig. 4) must be applied
at each functional level so that the situation has to be analyzed
and the design goals have to be formulated again. This effectively
means that mission and function analyses are performed over and over
again as you descend throught functional levels and as it is applied
across the large number of system functions. Similary steps for
allocating functions and determinating task and interface require-
ments are repeated many times in the course of synthesis and analysis
or design of the system. That means, for each work station, we are
starting with the mission of the system under consideration (Fig. 6).
~J. analyzing the mission, system functions are determined. The
analysis of system functions leads to functional requirements which
are the basis for allocating the functions to men and machines. The
detailed function analysis results in the required task performance
of the operator and the required machine processes. Finally the
analysis of the operator tasks and the machine processes gives the
data for work station design, work environment design, workload
evaluation, and personnel selection and training.

CLOSING REMARKS

The ergonomic design steps described should be performed through-


out the development phase and for all system development projects.
But it is frequently not necessary to perform all activities in
every detail. Ergonomists will almost always be involved in some
detail studies at the end of the development phase. If they have not
been involved in the earlier stages of that phase then the infor-
mation which, e.g., the human engineer requires for the work station
design will either not be available at all or will not be available
in a form useful to him. In that case some of the ergonomic design
tasks from the earlier development phase, e.g. the function alloca-
tion, will have to be performed during the detailed design. This
usually means early phase tasks cannot be done as thoroughly as they
should be nor for all system areas.
28 B. DORING

~
.Work Station Design
.Work Environment Design
• Workload Evaluation
• Persorv.! Selection and
Training

Fig. 6. The systems ergonomics approach

Furthermore, the level of ergonomics involvement also differs


from system to system because of different levels of automation,
technical risk, etc. For those various reasons the size and cost
estimates of the systems ergonomists group in the system develop-
ment team must be adjusted to the requirements of the project.
Other ergonomical management aspects essential to a successful' system
development program which should be identified are the required types
of human factors specialists, their task responsibilities, program
schedules, end products, reports and deadlines. The importance of
planning, funding, and scheduling of ergonomical activities in detail
for a development project cannot be overemphasized. This is because
if the program is not planned there is no basis for determining and
allocating funds or for scheduling the program. If funds are not
allocated there will be no ergonomical program on the project and
if the program is not scheduled, or scheduled properly, ergonomi-
cal inputs into the design will be untimely and will have little
practical effect.

Of course, system design costs will be slightly higher because


of systems ergonomics activities during systems development. But
these additional costs are relatively small and produce large system
gains; in some cases even system cost reductions. For example,
Coburn (1973) declares that the additional expenditure necessary
for ergonomical activities during the development of systems is only
SYSTEM ERGONOMICS IN MAN-MACHINE SYSTEMS DESIGN 29

1 % to 2 % of the total system development budget. And these costs


are compensated several times by the resulting system effectiveness
which may increase considerably. Some typical types of improvements
which result from the use of ergonomics during the system develop-
ment process are e.g.:

- Improved system performance.

- Reduced personnel requirements.

- Improved manpower utilization.

- Increased safety and reduced losses from accidents and errors.

- Improved standardization and maintenance.

- Improved user acceptance.

It is fairly certain that systems of the future will have to


be designed using the systems ergonomics approach. Most of the
successful advanced systems which have been recently designed were
developed according to this approach. As systems become more sophi-
sticated, complex and expensive it becomes more and more difficult
to be sure of meeting mission requirements without considering ergo-
nomical data systematically. Furthermore, only with this approach
can the system design be made cost-effective which is an essential
system goal in today's environment of diminishing economic resources.

Systems ergonomists involvement in the development process is


important because the impact of men in the system on system cost
and performance is very large. As systems become more complex new
functions are being performed so that the workload on the man in-
creases in spite of the fact that many functions' are being auto-
mated. Automation has increased the number of maintenance functions
allocated to the man and has tended to leave the more difficult and
cognitive tasks, both operational and maintenance, to the man.
Furthermore, increased system capabilities have required increased
flexibility in system responses thereby placing more responsibility
on the man in the system and complicating the design of man-machine
interfaces. TO ignore such human factors problems in system design
today can only lead to ineffective systems.

REFERENCES

AFSCM 80-3, 1966,"Handbook of Instructions for Aerospace Personnel


Subsystems Design," Air Force Systems Command, Andrew Air
Force Base, Washington, D.C.
Chapanis, A., 1970, Human Factors in Systems Engineering, in:
"Systems Psychology," K. B. De Greene, ed., McGraw-Hill
Book Company, New York.
30 B. DORING

Christensen, J. M., 1971, Human Factors Engineering Considerations


in Systems Development, in: "Measurement of Man at Work,"
W. T. Singleton, J. G. Fox, and D. Whitfield, eds., Taylor
and Francis, London.
Coburn, R., 1973, Human Engineering Guide to Ship Systems Develop-
ment, Naval Electronics Laboratory Center, San Diego.
Daenzer, W. F., ed., 1978/79, "Systems Engineering," Hanstein Verlag
GmbH, Koln, and Verlag Industrielle Organisation, Zurich.
Dathe, M., 1971, "Moderne Projektplanung in Technik und Wissenschaft,"
Hanser, MUnchen.
De Greene, K. B., ed., 1970, "System Psychology," McGraw-Hill, New
York.
DOring, B., 1976, Analytical Methods in Man-Machine Systems Develop-
ment, in: "IntroductJ-on to Human Engineering," K.-F. Kraiss,
and J.lMoraal, eds., Verlag TOV Rheinland, Koln.
Haberfellner, R., 1978/79, Lebensphasen eines Systems, in: "Systems
Engineering," W. F. Daenzer, ed., Hanstein-verlagGmbH, Koln,
and Verlag Industrielle Organisation, Zurich.
Hoyos, C. Graf, 1974, "Arbeitspsychologie," Kohlhammer, Stuttgart.
Kirchner, J. H., and Rohmert, W., 1974, "Ergonomische Leitregeln zur
menschengerechten Arbeitsgestaltung," Hanser, Munchen.
McCrimmon, K. R., 1968, Decision Making Among Multiple-Attribute
Alternatives: A Servey and Consolidated Approach, Memorandum
RM-4823-APRA, RAND-Corporation.
McCormick, E. J., 1976, "Human Factors in Engineering and Design,"
McGraw-Hill, New York.
Meister, D., 1971, "Human Factors: Theory and Practice," Wiley, New
York.
Schmidtke, H., ed., 1981, "Lehrbuch der Ergonomie," Hanser, Munchen.
Shackel, B., 1971, Criteria in Relation to Large-Scale Systems and
Design, in: "Measurement of Man at Work," W. T. Singleton,
J. G. Fox, and D. S. Whitfield, eds., Taylor and Francis,
London.
Shackel, B., ed., 1971, "Applied Ergonomics Handbook," IPC Science
and Technology Press, Guilford (Surrey).
Singleton, W. T., Easterby, R. S., and Whitfield, D. S., eds., 1967,
"The Human Operator in Complex Systems," Taylor and Francis,
London.
Singleton, W. T., 1974, "Man-Machine Systems," Penguin Books, Har-
mondsworth (Middlesex).
Vossel, G., and Frohlich, W. D., 1977, Aufmerksamkeit bei tiberwa-
chungs- und SteuerungstAtigkeiten, Wehrpsychologische Unter-
suchungen 4: 1.
Zangemeister, C., 1971, "Nutzwertanalyse in der Systemtechnik,"
Wittmannsche Buchhandlung, MUnchen.
TASK ANALYSIS IN SYSTEM ERGONOMICS

Heiner Bubb

Institut fur Ergonomie der Technischen Universitat

Mlinchen/FRG

In the preceding paper it has been more generally shown, how


system ergonomics is especially concerned with the development of
ergonomic analysis and design methods and their application to the
solution of man-machine-system problems which arise during system
design. The steps have been pointed out, through which a system
ergonomic application has to run (development phase, realisation
phase and finally operation phase) and it has been demonstrated
that various disciplines have to contribute to system ergonomics.
Moreover different solution proposals are generated by system ergo-
nomics. It should also be a task of this discipline to select the
best solution by the cost criterion. Assuming that system perfor-
mance requirements and system function have already been determined,
the "function allocation" distributes system functions to personal
and machine elements.

The intention of this paper is to show in more detail:

- What is the object of system ergonomics?


- What rules exist for the application of system ergonomics?
- What is the starting point of a system ergonomic analysis?

THE OBJECT OF SYSTEM ERGONOMICS

The object of ergonomics is to adapt work and work environment


to the properties of man by a rational consideration of the inter-
action between man and his work. One method is the stress and strain
analysis. Stress analysis essentially develops recommendations
about how stress caused by work and work environments can be re-
duced to avoid unreasonable strain. This method is in particular

31
32 H.BUBB

appropriate to physical load and environment influences, like noise,


vibration, climate and so on.

A different method is the analysis of performance. If a change


in system design leads to an increase of human performance with un-
changed task and specifications, one may assume that stress and ac-
cordingly strain will have decreased. Nowadays however physical ac-
tivities are replaced more and more by mental activities according
to the possibilities of mechanization. Both methods however give
little support in the case of mental load, because no general method
to measure mental work is available. Therefore neither mental load
nor strain can be measured. For the same reason it is not possible
to determine generally the performance of human mental work. And
therefore both methods are unfit to develo~ methods for a reduction
of strain or an improvement of performance. That can only be done
by the methods of system techniques.

System analysis represents the study of the principle structure


of man's involvement in a complex machine system in order to get
hints for possible improvements. In order to get a rough impression
of the structure of man-machine-system human work is to be defined
with regard to information and information processing. According to
Hilf (1976) every aimed and intended human activity may be called
"work". Hence for every activity there is an ideal imaginary prob-
lem, called "task" and a realisation, called "result". The connection
between task and result is given by the man-machine-system (MMS; see
Fig. 1).

By the use of a machine man tries to cope with a task, set by


himself or by others. Machine in this sense may be every tool from
the simple pencil to the console of a highly automated technical
process. The principle structure of the man-machine-interaction re-
mains unchanged. The feed back path in Fig. 1 closes the loop and
shows that the operator is able to compare the input vs. the output
i.e. task and result. This comparison may be carried out either
while working or after finishing. This depends on the task or on
the machine used.

Environmental influences may disturb this interaction. One of


the topics of a system ergonomic approach is the determination of
elements and their interrelation. The result of this work may be
drawn in a block diagram. Furthermore the interrelation found should
be described mathematically as far as possible. Finally, a comparison
with experimental results leads to improving measures. The experi-
ments ought to be universally valid. Of course, these measures may
optimize the system's performance as well as the reduction of human
load. Moreover system techniques enables predictions of the system's
output already in the developmental phase.
TASK ANALYSIS IN SV.STEM ERGONOMICS 33

/'"
envi ronment

task =
{man machine}----. result
~~------------------------------~~

Fig. 1. General structure diagram of human work

APPLICATION OF SYSTEM ERGONOMICS

The drawing of a block diagram may be explained by a simple


example, which seems to be quite unsimilar to a man-machine-system
(see Fig. 2): two people talking to each other. We can regard each
person as an element of the system. In a block diagram each element
Is drawn as a rectangular. Let us assume that for the moment only
one person is talking and the other one is listening. In other words
information is transferred from one element to the other. Graphical-
ly this is represented by an arrow pointing from the speaker to the
listener. In a good conversation the second man may also take over
the speaker's function. Therefore in our diagram a further arrow
should be drawn from the second element to the first. TO make a clear
interpretation we have to distinguish exactly between the input side
and the output side of each element. This is done in Fig. 2.

The discri~nation of input and output supports the possibi-


lity of an analysis in more detail: At the input side of the man,
sense organs like eyes, ears and so on are to be mentioned. The in-
formation which has been received in this manner will be converted
by man into noticeable information which reaches the environment
only by the innervation of the muscular system, i.e. the output
side.

The simple picture of Fig. 2 supports two important viewpoints:

- Information runs in certain channels, always directed from


the output of an element to the input of another element.
- Elements are defined by their property to change the infor-
mation in a certain manner given by the element's working
principle. When we are able to define the working principle
of information changing mathematically, we have done one
important step to describe the system scientifically.

But first let us replace the second man in the block diagram
of Fig. 2 by a machine, especially a drilling machine. Of course
input and output of the man remains the same. At the input side of
the machine there are the controls, by which the operator "tells"
34 H.BUBB

speaker listener

,.. listener speaker r---

Fig. 2. Structure diagram of a conversation

his intention to the machine. It is not possible to describe the


output of the machine as generally as the output of the man. The
output of the machine as a technical element is given by the pur-
pose for which the machine has been constructed. But it is possible
to interpret the output information of the machine generally. It is
the result of the man-machine-system.

When we now complete the task in Fig. 3 we get a more detailed


block diagram of the man-mach~e-system than in Fig. 1. By the fore-
going di$cussion the fundamental proceedings of system analysis has
become obvious: The physical nature of elements and their interaction
are disregarded and only the formal structure of interaction is ex-
amined.

We call such areas of the interaction, which change the infor-


mation in any way, "structure elements". This change may be depending
TASK ANALYSIS IN SYSTEM ERGONOMICS 35

2!
'"c:0 c:
'"E

-[
~

task ~
E result
r---- Q;'"
:::>
0 man u machine r--r-
'"'"c: '"E
:::> e
c:
~
'"'" 0
u

Fig. 3. Structure diagram of a man~machine-system

on time: A certain time passes between the input movement of the


control elements and the output reaction of the control system. This
dependence may be described by differential equations. The repre-
sentation of the frequency response is most suitable for a mathemat-
ical description of complex interactions. This dependence on time,
gives the single element and therefore the whole system a dynamic.
We call this "system dynamic". You will understand the importance
of the dynamic very well if you consider the following act: There
is quite a difference in the time reaction of the control element
whether you turn the steering wheel of a motor car or that of a big
ocean steamer.

Anonther changing property of system elements is the so called


nonlinearity. The simplest forms of nanlinearities are:

- saturation i.e. there are boundaries for the amplification


- thresholds i.e. there is no arbitrary small amplification
- reversing of sign i.e. for example there is no amplification
on the negative axis.

All experiments concerning nonlinearities show decrements in


manual control performance. An example for this is the braking of a
car: in this case the input is given by the pressure on the brake
pedal; the information on the output side is the deceleration rate
of the vehicle. If the input related to the output was perfectly
linear, any attainable deceleration could be achieved by a corre-
sponding pedal pressure. The driver then could react successfully
36 H.BUBB

in (almost) every situation. Unfortunately there is a nonlinearity


in the transfer function, i.e. the deceleration cannot exeed boun-
daries determined by the road-tyre-contact.

THE STARTING POINT OF A SYSTEM ERGONOMIC ANALYSIS

The first step to develop ergonomic solution proposals is the


analysis of task. By the observation of many experiments we may
extract that mental tasks created by the interaction of man and
machine possess the following five independent properties:

- the temporal sequence of partial task, called operation


- the spacial relatively temporal restriction of task, called
manner of control
- the number of degrees of freedom to be controlled, called
dimensionality
- the active or passive involvement of man in the working pro-
cess, called manner of task
- the principle of displaying task and result, called display.

Each of these properties bears a certain degree of difficulty.


For the first three properties the difficulties are determined by
the task itself. In this case ergonomic measures may only reduce
the difficulty as far as allowed by the task. For the last two pro-
perties the degree of difficulty may be influenced by ergonomic de-
sign to a large exten t.

Operation

Almost every general task may be organized in several partial


tasks. As we have already described, operation means the temporal
organization of these. partial tasks. Fig. 4 shows a simple example
in the form of a cartoon. In most cases the production of anything
may be only done by sequential working with different machines. We
call that sequential operation. The task of operating a typewriter
is an example: In this case every key and the type related to it is
a machine. These machines must be operated in a sequence given by
the text to be written·, which is the overall task. Similar but much
more complex operations must be done by the operator of a large scale
apparatus. The sequence of partial tasks given by the overall task
and by the constellation of machines may be demonstrated by a flow
diagram. Fig. 5 shows the flow diagram of changing gears in a motor
car. We may observe, how complex this simple operation is in detail.
But the human operator is able to do such complex things and to talk
simultaneously with his passengers. A simple judgement of mental load
by sequential operation is the so called time-budget. It means the
average relation between the time tT required by the task and the
available time tA in percent:
TASK ANALYSIS IN SYSTEM ERGONOMICS 37

Fig. 4. Operation: sequential operation

1 N
N Ln=1 ) • 100 %. (1)

WL is the workload. This judgement procedure is especially applied


during the development of civil and military aeroplanes and increas-
ingly in other complex MMS. Experience and many experiments show the
boundary of human permanent performance, given by workload of 55 to
65 % (Seifert, 1980). If this boundary is taken over by the given
task one solution could be a partial automation of some tasks. An
example is the motor car with automatic transmission.

A different organization of partial task may be the simultaneous


operation: this is, when several different tasks must be fulfilled
at the same time (see Fig. 6). For example the driver of a shovel
dredger has to control simultaneously the movements of his vehicle
and,in additio~ the position of the shovel in time accordance, to the
first task. There are two possible difficulties in simultaneous op-
erations: first, interference between tasks may arise. The conse-
quence is decreasing performance in the task considered to be less
important, while the requirements in the main task increase. Most
investigations are concerned with the human strategies of devoting
attention to the different tasks (for example: Poulton, 1974;
Carbonell, 1966; Smallwood, 1967; Kelly, 1968). In many cases how-
ever a sequential operation becomes necessary, as simultaneous op-
eration is impossible for technical and organizational reasons. For
example, this is the case when there are more controls than fingers,
arms and feet. Such situations may be described by the methods of
operation theory (Kampe et al., 1973; Gnedenkow and Kovalenko, 1971).
The main issue of the operation theory is the so called "waiting
room" represented by the short time memory in the case of human op-
erator (for example: Schumacher, 1976).

If the man is overtaxed by simultaneous operations, ergonomic mea-


sures may be on the one hand the combination of several informations
w
CD

no no

no

right toot slowly press ~ ~ right loot slowly press


down the occelerator down the accelerator
::r:
aJ
Fig. 5. Flowing diagram of changing gears in a motor car C
aJ
aJ
TASK ANALYSIS IN SYSTEM ERGONOMICS 39

Fig. 6. Operation: simultaneous operation

to one - so called synthetic - display or on the other hand the


controlling of several partial machines by only one control element.
An example of the synthetic display is the DME (Distance-Measuring-
Equipment) in an aeroplane. In certain time intervalls the distance
to ground stations is measured and then the true movement can be
calculated by a bord computer. Fig. 7 shows such a display. An ex-
amPle of combined controlling is given by modern electronic organs.
The organist may play only the melody (a sequential operation!) and
the computer in the organ produces the rhythm and the accompaniment.

Manner of control

In the preceding discussion it was always assumed that the


task is set by others and is strictly determined. However, in most
practical cases the operator has to derivate the necessary operation
sequences from the stimulus configuration of the environments. By
the keyword "manner of control" the question "How restricted is the
correlation between stimulus configuration and task?" is investi-
gated.

Strictly determined tasks are called dynamic tasks. Fig. 8


shows an example in the form of a cartoon. Dynamic tasks are cha-
racterized by a forcing function, i.e. a function depending on time.
The operator always has to keep the output o! the control system in
congruence with this function. Examples are the strictly ordered
sequences of assembly-line work, the position finding of aeroplanes
on the radar screen using the cursor and the exact flying on the
glide path.

Most dynamic tasks however are given in so-called tracking


experiments, by which for example the frequency response of man
could be investigated (McRuer and Krendel, 1957; McRuer and Jex,
1967; Elkind, 1956 and 1964). Such experiments make it possible to
define the range of frequency, wherein human operation is possible.
40 H.BUBB

Fig. 7. Synthetic display by the example of DME (here in the form


of a pictural display)

The upper cutoff-frequency is given by the average detecting time


of the human sense organs (about 100 ms) or else his simple reaction
time (about 200 ms). Therefore the upper cutoff-frequency lies near
2.5 cps or 1.25 cps (mostly assumed as 1 cps). In most tracking ex-
periments however the cutoff-frequency is chosen much lower (for
example: 0.15 cps by P. Bubb, 1978, and H. Ruhmann, 1978).

The lower cutoff-frequency is given by the observed threshold


of movement. It lies near 1 angular minute/second to 2 angular
minutes/second for vision (Johannsen, 1965). As in this case the
amplitude of the movement plays an important role, the lower cutoff-
frequency depends on the magnitude of the visual angle under which
the movement is observed. Fig. 9 demonstrates this correlation. The
TASK ANALYSIS IN SYSTEM ERGONOMICS 41

Fig. 8. Dynamic task

1iBBB lL
,
100' 90'
80'
4000 I'I" 1 II Ii
40· 45' Ii
2000 manCl!uvrable aero
20'
1000 ~ ~
800
...
C
10'
8' 400 I' Ii
...> I
I'
E
4'
200
Ii
0
E
'0 2' ~
... 19B
=
"0
l'
40 I' I!
I!
Q.
E
0
20 ~
10 ~ I!
8
I' Li
I' ~ Ii
li
81 81 81 81 8
10-4 10-3 10- 1 1 Hz
manCl!uvroble cutoff - frequency
Fig. 9. Lower and upper cutoff-frequency of man

lower cutoff-frequency corresponds to the frequency of task and


result. Therefore large slow-speed-ships cannot be controlled as
intuitively as a motor car but only by an analytical calculation.

Also dynamic tasks enable us to define their difficulty if the


time dependence remains within the manoeuvrable frequency range.
The simple fundamental idea is: The machine must be able to ac-
complish the task, that means it must be always a little "faster"
42 H.BUBB

than required by the task. This may be expressed as degree of dif-


ficulty p" described by the relation between the maximal possible
output signal of machine Xmax and the maximal input signal of the
forcing function wmax ' or should the occasion arise by the cor-
responding derivatives.

w
max
p
.
x
w
max
max
respectively: ;;-- (2)
max
w
max
respectively: =-;r-
max

The greatest value of the equation (2) determines the degree


of difficulty. Maximum difficulty is given by p = 1. In reality p
should be much smaller to make it possible to compensate for mis-
takes and for the influence of reaction time. By the observation of
many experiments and by the experience of practical control systems
verbal description may be related to different steps of degree of
difficulty (Schmidtke, 1976). This is done in Table 1 •. It is very
interesting that the boundary of human permanent performance also
here is about 50 % of the maximal demand as in the calculation by
the time budget.

In most cases tasks are not given in such narrow bounds as the
dynamic task. Most tasks are characterized by a time independent
instruction related to the desired final product. We call that
static tasks (see Fig. 10). An example is drilling with a machine
or working with a turning-lathe. Only the final properties are pre-
scribed by the blueprint and not the way to them.

In practice there are many tasks, which combine the properties


of static and dynamic tasks. They are characterized by a static aim.
The operator is not altogether free to determine the way to this
aim and therefore the carrying out has the properties of a dynamic
task. An example is driving with a car, but let us return to this
fact later. First let us consider a remarkable difference between
static and dynamic tasks.

Ip the case of dynamic tasks the bounds of controlability a~e


given essentially by the upper and lower cutoff-frequencies of the
man and by the dynamic properties of the machine. In case of static
tasks the human operator may additionally determine the degree of
difficulty to a large extent by his choice of the working expedition.
Therefore it often happens that he overtakes the bounds of linearity
of the control system. Thus he usually is no longer able to control
the system. An example is skidding with the car due to excessive
TASK ANALYSIS IN SYSTEM ERGONOMICS 43

Table 1. Task performance depending on the degree of diffic~lty

degree of difficulty description

a < p EO; 0,25 easy

0,25 < P EO; 0,5 difficult

0,5 < p EO; 0,75 very difficult

0,75 < p EO; 1 extremely difficult

Fig. 10. Static task

velocity. An ergonomic measure to avoid such dangers is displaying


the bounds of linearity to the operator. Often such a display is
only possible in the form of a synthetic display, as different pieces
of information determine the bound of linearity. For instance, the
indication of braking distance would improve the choice of a secure
velocity (H. Bubb, 1981). Fig. 11 shows the braking distance indi-
cation in form of a head-up-display. In this case the driver should
choose the velocity in such a manner that a pedestrian could cross
the street in front of the light bar which indicates the braking
distance. Moreove~also the bounds of a sideward.manoeuvre could be
indicated.

With more careful viewing we may observe that in the case of


a static task there are some additional partial tasks. subordinated
hierarchically. This fact is given in controlling vehicles of near-
ly all kinds. Fig. 12 shows .the corresponding structure diagram
(Bernotat, 1964~ Kelley, 1968). Hierarchically seen the navigation
44 H.BUBB

Fig. 11. Breaking distance indicator in head-up-display technique

- - r-- r--
~
0 posi-
.e
~

transport - course and path and -0


VI tioning '"u result
0
'"
quan- =
01 "C
~
task > overage ::::J
exact :c
'"c:: velocity
01 0
velocity Vi titles
'">

i- i- r'- -

Fig. 12. Hierarchically subordinated loops while controlling a


vehicle

task is the highest level. It contains the essential transport task,


for example flying from London to Paris with 150 passengers at a
certain time. This task is static. The route and the average velocity
TASK ANALYSIS IN SYSTEM ERGONOMICS 45

may be derivated. These dates represent the task of the next lower
level, the guiding task. The exact way and the necessary changes of
velocity or altitude are derived depending on the momentary circum-
stances. They represent the task of the lowest level, the stabilizing
task. These demands are realized by the control system, i.e. the
vehicle. In the case of a motor car that means that on the level of
the guiding task the driver places an individual course on the road
ahead depending on the bounds of the road, road conditions, other
vehicles and pedestrians, and he intends to keep the course on the
level of the stabilizing task. As you see this lowest level now is
a dynamic task.

If the operator is taken over by these tasks an ergonomic aid


is given by automation. The automation of the lowest level is re-
latively easy. In aeroplane technology this is comprehended by
"primary flight control systems" (Edwards, 1976). It means for ex-
ample the automatic stabilization of the position and the velocity
of the aeroplane by the autopilot. In motor cars the same level is
automized by the "cruise control". We would certainly improve the
traffic safety, when we would also automate the control of the steer-
ing wheel. In this case the driver would give the desired course by
the steering wheel and even under the condition of cross wind the
car would not derivate from its path.

In aeroplanes the next higher level is already automated: the


so called "flight director" calculates the necessary course and
velocity which is taken as the forcing function by the autopilot.

Even the level of navigation has to be automated in modern


concepts of traffic control (per example: ALS from VW and Bosch and
the "scoutsystem" from BMW and Siemens). In this case the driver
would give his desired destination to the bord computer. This com-
puter calculates the best way and displays when necessary a turning
at road junctions.

Dimensionality

In order to judge the complexity of controlling the number of


degrees of freedom to be controlled is most important. We called it
dimensionality. According to physics, the movement of a free body
in space has six degrees of freedom. In most cases only a few of them
can be controlled (see Fig. 13). For example the jig of a drilling
machine can only be moved in one direction. Movement in all the other
dimensions of space is not possible by a forced guide. The same is
given in the control of a railway engine. Whereas this machine makes
movements on the threedimensional surface, the engine-driver only is
able to determine the velocity. It is a onedimensional task.

Therefore driving a car is a twodimensional task, as the driver


is able to determine sideward and foreward movements. This is correct
46 H.BUBB

>

Fig. 13. Dimensionality

although by its technical realization the car is not able to make


a sidewards movement without a simultaneous foreward movement. Both
movements are coupled. Only the aim, the idea which should be real-
ized, not the given technical realization is important in deter-
mining the dimensionality. In our case it is the twodimensional
movement on the surface by a machine.

According to similar considerations flying an aeroplane may be


characterized as a threedimensional task. In addition to the car
there is the controlled movement in vertical direction. Therefore
the example of Fig. 13, controlling a portal crane, is a fourdimen-
sional task. Of course steering a space craft is a sixdimensional
task.
TASK ANALYSIS IN SYSTEM ERGONOMICS 47

The property of task "dimensionality" belongs not only to ve-


hicles but also to other machines. In order to determine the dimen-
sionality it is of great importance whether the variation of the
controlled object is observed as a spacial position variation (for
example: a diagram on a screen) or not. In the latter case the task
would be one of simultaneous operation.

Some ergonomic rules are given by the dimensionality. The con-


trol of one- and twodimensional tasks is rather easy for man because
of his own possibility of movement. Man has also experience in three-
dimensional tasks. But the control of four- and moredimensional tasks
afford a very high learning effort.

The difficulty based on dimensionality is directly determined


by the task. However, a further difficulty is added by the dynamic
properties of machine and also by the so-called coupling of the dif-
ferent dimensions. An example is given by the coupling of aside and
alongside movement of a motor car. According to the experience of
the operator such coupled dimensions may be controlled rather well.
But the manner of coupling should not change while working. The ex-
ample of motor car shows that a car near the bounds of skidding may
not be controlled sufficiently by the driver. A measure to improve
the control may be to disconnect the coupling. For the case of a
motor car Schulze (1981) proposed a velocity-dependend steering gear
system that ensures uniform vehicles response to steering input at
all velocities.

As machines are traditionally constructed from a technical view


point, often several control elements of different partial systems
are to be manipulated in order to influence only one or two dimen-
sions. Old steam engines are an example. In a manual operated motor
car you have to use five control elements for this twodimensional
task. An essential improvement in control could be achieved, if the
control of machines would be constructed by the dimensionality. Nowa-
days this is easily possible, as microprocessors are available. The
optimal ergonomic aim should be: The number of control elements
should not be greater than the dimensionality of task •

.Manner of task

We called the active or passive involvement of man in the work-


ing process manner of task (see also Fig. 14). We always substituted
the active involvement in the preceding discussion. It means the in-
formation of the task first reaches the operator, then gets into the
machine in a changed form and leaves the system as result. It is
therefore a seriell man-machine-Ioop (see Fig. 15). In most cases
the operator is able to observe the difference between task and re-
sult. If the difference exceeds a threshold often determined by the
operator himself, the operator accordingly changes the information
transmitted to the control elements. That is the working principle
48 H.BUBB

Fig. 14. Manner of task: active and monitive task

task resul t
------------~-( man machine l

open man - machine loop

task result
-----------I-~-I man
I
closed man - machl ne loop

Fig. 15. Active system or seriell man-machine-loop

of a closed loop system. An example is the controlling of vehicles


of all kinds. In some cases an operation by man is only possible
after finishing the current work. We call that an open loop system.
An example is the choice of drilling speed depending on the kind of
material. While running usually no change is possible. In both cases
the man ist actively involved in the system. We call them active
systems. Fig. 15 shows the structure diagrams of such systems.
TASK ANALYSIS IN SYSTEM ERGONOMICS 49

Another kind of task is given by the parallel man-machine-loop


(see Fig. 16). This is always given by the use of automatons (see
also Fig. 14). The man is connected in parallel to the machine in
the sense that he observes only from time to time the difference
between task and result but is not active. Only in the case of gross
defficiencies he changes some parameters of the machine or switches
to manual operation. Because of the observing task we call this
monitive system.

The decision to take an active or monitive system must be made


by the system planner and is essentially not given by the task. In
the preceding discussion we had found out some situations which re-
commend automation. But, technically seen, automation is only pos-
sible, when the task i.e. the forcing function, is measureable or
physically describeable. More generally we may state: automatons
are preferred, if exactly defined information which stands out
distinctly from the background noise, must be processed safely and
fast. But the man must be an active element of the system, when
correct but not exactly programmable action must be derivated from
different information channels.

But also some human properties are important for the decision
man or machine: the control and supervisory task of a monitive
system are characterized by monotony. And monitive systems rather
require good abilities in decision making. In contrast to that,
active systems stress the man by the specific pressure to permanent
attention. Active systems require a high level of practice. Even by
the loss of the level of practice there arise difficulties, when
switching from automatic to manual operation (Edwards, 1976; Johann-
sen, 1976).

Display

In many cases task and result is represented to man by the


natural surroundings. But often it is necessary to display this in-
formation. In the case of technical displays one principle is to
ask whether to display task and result separetely or already the
difference of both (see Fig. 17). When displaying separately we call
that pursuit task or also pursuit display (Poulton, 1974). The point
of subtraction is inside the human operator. The corresponding struc-
ture diagram is demonstrated in Fig. lB. When displaying the dif-
ference of task and result we call that compensatory task or ~
pensatory display, because the human operator has to compensate
for the displayed difference. In this case the point of substraction
is made by a technical operation. Fig. 19 points out the correspond-
ing structure diagram.

The pursuit display is often used in observation wards not


linked to a vehicle. A typical example is the radar station of air
traffic controllers. The aeroplanes are displayed on the screen
50 H.BUBB

task t man ).
j
machi ne
resul t

parallel man - machine loop

Fig. 16. Monitive system or parallel man-machine-loop

PurSUI lOS

(ompfnsolory los

Fig. 17. Display: pursuit and compensatory task


TASK ANALYSIS IN SYSTEM ERGONOMICS 51

task
result

Fig. 18. Structure diagram of a pursuit display

task
+-c display ~ machine If result

Fig. 19. Structure diagram of a compensatory display

congruent with the signal of the aeroplane in order to fix its po-
sition. Therefore he is able to observe separately the motion of
the forcing function, i.e. the task respectively the aeroplane, and
the motion of the result, i.e. the position of the cursor. He may
get experience of the motion of task and result independent of each
other. Thus he is able to make a good short time prediction of the
future motion and so to react in time. But by pursuit displaying
the whole area of variation of task and result is to be displayed.
That means, that possibly both appear very small.

Therefore in technical systems often the compensatory display


is preferred, because the display amplification may be chosen free-
ly. A U-form dependence there may be suggested between display
amplification and error. Therefore the best amplification can be
chosen. But there will not be developed a good "feeling" of the
future difference or of the dynamic properties of the control system
from the compensatory display. Experiments, however, demonstrate no
difference of performance between pursuit and compensatory display,
if there is time enough to get experience.

Beside these two manners of displays the so called preview


display is often used (Sheridan and Ferrell, 1974). Fig. 20 shows
the structure diagram in combination with a compensatory display~
Preview display means that the operator may observe the future de-
velopment of the forcing function (i.e. task) within a certain time.
The best preview time depends on the cutoff-frequency of the forcing
function. The man must be able to look ahead to the next turning
52 H.BUBB

task resu II

Fig. 20. Structure diagram of a preview display

point of the forcing function in order to create an optimal program


of reaction (Poulton, 1974). In practice a preview time of 2 to 3
sec is sufficient (Poulton, 1966; McLean and Hoffman, 1973).

A simple example of a preview display is the task on a sewing-


machine. A similar task is given to the driver who views the road
ahead. Many investigators have dealt with driving (for example:
Weir and McRuer, 1968, 1973; Fiala, 1966; Crossman and Szostak,
1969). The common result is: driving is a compensatory task, as the
driver judges all distances from his own position; driving is si-
multaneously a pursuit task, as the driver has always an idea of
his motion in space. And driving is a task with preview display, as
already said above. This counts certainly also for other kinds of
vehicles. Out of that we see that controlling a vehicle generally
constitutes a compensatory task. In the case of a pure compensatory
display the reference point for measuring the differer.ces may be
chosen in such a manner that no difficulties of compatibility arise.
But in the case of compensatory display in a vehicle there is a
specific problem. This problem easily becomes clear by Fig. 21: As-
sumed in the real situation the driver of a car has to turn right
because the road bends to right. Then we may relate the road to the
forcing function (F) and the motion of the car to the result (R).
Of course, the driver has to turn his steering wheel to the right
in order to keep right. If we transfer this situation to a simple
simulator, we get the picture below. But now the sense of rotation
of the steering wheel is never compatible to the desired direction
of motion of the manoeuvrable pointer, although the structure dia-
gramm, in Fig. 21 right, remains the same. We could get a compatible
situation by inverting the sign of the subtraction of task and re-
sult (see in Fig. 21 the lowest line). Now a right turn of the steer-
ing wheel introduces a movement of the pointer to the right. What is
the difference? In the second case the operator tries to draw the
manoeuvrable pointer into the neutral position. The reference p-oint
for measuring the difference between task and result is in the forc-
ing function, i.e. the task. Whereas in the first case the operator
tries to influence the movement of the manoeuvrable vehicle. Now
the reference point is in the vehicle, i.e. the result. This dif-
ference causes difficulties when displaying external information
artificially in a vehicle. The observation of a gyro horizon in an
aircraft makes this clear (see Fig. 22). While flying with external
TASK ANALYSIS IN SYSTEM ERGONOMICS 53

real situation
F

cOllman structure diagram

~

simulated situation 'R-~
R F

structur. diagram
compatible situation
F

Fig. 21. Compensatory display compatible to reality and compatible


to the movement of the control element

sight the movement of surroundings and of a customary gyro horizon


are compatible. The corresponding sense of turn of the control ele-
ment is reversed to it, but the pilot has learned that he influences
the movement of his aircraft. On the other hand,in the situation of
no external sight,this can be confusing: the pilot may in error take
the attitude of the gyro horizon to be the attitude of the aircraft
(Fitts and Jones, 1961). A compatible situation would exist, if not
the external sight but the aircraft itself would be displayed symbol-
ically. This, however, is incompatible with external sight situation.
There is only one way to avoid this difficulty: a true motion head-
up-display is reflected off the windscreen of the aircraft into the
pilot's eyes (Naish, 1964, 1970). It is collimated at infinity to
ensure that it is always in focus. Fig. 23 shows an example. With
the head-up-display the ambiguity of customary gyro horizon is
loosened. The pilot reacts on the artificial information in the same
manner as on the natural information.

We can generally say that if external information is to be dis-


played in a vehicle by the head-up-display technique difficulties
of interpretation may be avoided. The head-up-display of braking
distance in Fig. 11 has been an example for that. Here the light bar
is not collimated at infinity but at the distance to be displayed.
Also in this case the driver has no difficulties in interpreting
the information correctly.
54 H.BUBB

real situation

(~

IVlsual fllghtl Iinstrument flight I

IIY lng Wllh exlernal sighl flyong wllhaul pxlernal Sighl IIYlng wllhaul pxlernal Sighl
ccmpaloble SIIuol,on

"I ns ide - oul · concept"' "oulslde ·I n· concepl


Fig. 22. The problem of the gyro horizon

Fig. 23. Gyro horizon by the head-up display


TASK ANALYSIS IN SYSTEM ERGONOMICS 55

The aim of this chapter was to show, how machines might be ad-
apted to the properties of the man. Therefore the performance of
the system may be improved and simultaneously the load on the man
be decreased. This becomes possible, if you do not start from the
possibilities of technology but start from the today's knowledge of
information processing by man. By such a process man becomes able
again to master the machine, even under the condition of highest
complexity, and for that reason modern apprehensions that man could
be controlled by technology, can be prevented.

REFERENCES

Bernotat, R., 1964, Der Mensch als Element des Flugflihrungssystems,


Luftfahrt-Raumfahrttechnik, 10:66.
Bubb, H., 1981, The Influence of Braking Distance Indication on the
Dr':'v~r's Behaviour, in: "Human Factors in Transport Research,"
D. J. Oborne, and J. A. Levis, eds.,
Bubb, P., 1978, Untersuchung Uber den EinfluB stochastischer Roll-
schwingungen auf die Steuerleistung des Menschen bei Regel-
strecken unterschiedlichen Ordnungsgrades, Ph.D. Thesis,
Technische Universitat Munchen.
Carbonell, J. R., 1966, A Queueing Model of Many-Instrument Visual
Sampling, IEEE Trans. Human Factors in Electronic, HEE-7:157.
Crossman, E. R. F. W., and Szostak, H., 1969, Man-Machine-Models for
Car-Steering, Forth Annual NASA-University Conference on
Manual Control, National Aeronautics and Space Administration,
Washington, D. C.
Edwards, E., 1976, Some Aspects. of Automation in Civil Transport Air-
craft, in: "Monitoring Behavior and Supervisory Control,"
T. B. Sheridan, and G. Johannsen, eds., Plenum, New York
(N. Y.).
Elkind, J. I., 1956, Characteristics of Simple Manual Control Systems,
Technical Report 111, MIT, Lexington (Mass.).
Elkind, J. 1., 1964, A Survey of the Development of Models for the
Human Controller, in: "Guidance and Control II," R. C. Long-
ford, ed., AcademiC-Press, New York (N. Y.).
Fiala, E., 1966, Lenken von Fahrzeugen als kybernetische Aufgabe,
Automob~ltechnische Zeitschrift, 68:156.
GnedenkoW',B. W., and Kovalenko, 1. N., 1971, "Einfuhrung in die Be-
dienungs theorie ," Oldenbourg, MUnchen.
Hilf, H., 1976, "Einfuhrung in die Arbeitswissenschaft," de Gruyter,
Berlin, New York.
Johannsen, G., 1965, Geschehenswahrnehmung, in: "Handbuch der Psycho-
logie," W. MetZger, ed., Hograefe, Gc5ttingen.
Kelley, C. R., 1968, "Manual and Automatic Control," Wiley, New York
(N. Y.).
Krampe, H., Kubat, J., and Runge, W., 1973, "Bedienungsmodelle. Ein
Leitfaden fur die praktische Anwendung," Oldenbourg, MUnchen,
Wien.
56 H.BUBB

McLean, J. F., and Hoffmann, E. R., 1973, The Effects of Restrict


of Preview on Driver Steering Control and Performance,
Human Factors, 15:421.
McRuer, D. T., Jex, H. R., Clement, W. F., and Graham, D., 1967,
Development of a Systems Analysis Theory of Manual Control
Displays, System Technology Inc., Report No. TR-163-1.
McRuer, D. T., and Krendel, E. S., 1957, Dynamic Response of Human
Operators, WADC TR-56-524, Wright Patterson Air Force Base,
Dayton (Ohio).
Naish, J. M., 1964, Combination of Information of Superimposed Vi-
sual Fields, Nature, 202:641.
Naish, J. M., 1970, Control Gains in Head-Up Presentation, Proceed-
ings of the Sixth Annual Conference on Manual Control, Dayton
(Ohio) .
Poulton, E. C., 1964, Postpreview and Preview in Tracking with
Complex and Simple Inputs, Ergonomics, 7:257.
Poulton, E. C., 1966, Tracking, in: "Acquisition of Skill,"
E. A. Bilodeau, ed., Academic Press, London.
Poul ton, E. C., 1974, "Tracking Skill and Manual Con trol ," Academic
Press, New York, San Francisco, London.
Ruhmann, H. P., 1978, Untersuchung tiber den EinfluB der mechanischen
Eigenschaften von Bedienungselementen auf die Steuerleistung
des Menschen bei stochastischen Rollschwingungen, Ph.D. Thesis,
Technische Universitat Munchen.
Schmidtke, H., 1976, "Ergonomische Bewertung von Arbeitssystemen.
En twurf eines Verfahrens," Hanser, Munchen, Wien.
Schulze, B. G., 1981,Application of Ergonomic Principles in Driving
Control Design, in: "Human Factors in Transport Research,"
D.J. Oborne, and J. A. Levis, eds., Academic Press, London.
Schumacher, W., 1976, Untersuchung der Strategien des Menschen bei
der Bedienung mehrerer Anforderungen, AbschluBbericht zum
Forschungsvorhaben T/RF 36/RF 360/410, Karlsruhe.
GENERATION OF ERGONOMIC DATA AND

THEIR APPLICATION TO EQUIPMENT DESIGN

R. Bernotat

Forschungsinstitut fur Anthropotechnik

Wachtberg-Werthoven/FRG

INTRODUCTION

The purpose of this meeting, as given by the organizers, is to


discuss the data available today for design and to indicate existing
gaps in Ergonomic knowledge. The purpose of this paper is to provide
some insight into the application of Ergonomics and what the require-
ments of the designer are. In more detail the status of research in
Ergonomics as well as the problems in transferring the data to ap-
plication is discussed. The paper is far from being a comprehensive
treatment of the topic, but rather a collection of observations and
experiences over the last years brought into some order. In fact,
for the purpose of this meeting it is planned as an incentive and
stimulator for discussion.

The title, at a first glance, appears to be simple, but a more


closer look shows that it deals with the two faces of Ergonomics:
Science and Technology. Without raising the old struggle whether
Ergonomic is one or the other we accept the viewpoint of the Inter-
national Ergonomic Association that it is both at the same time.
However, it is well known that the mutual understanding of re-
searcher and applier of Ergonomics needs to be improved urgently
(Rutenfranz et al., 1980). In the professional journals the transfer
problems, therefore, have for years been one of the most often dis-
cussed topics. Even in the University Curricula in Ergonomics the
Janus-faced nature of Ergonomics should be reflected if they are
well structured (Bernotat and Hunt, 1977).

57
58 R. BERNOTAT

APPLICATION OF ERGONOMICS TO EQUIPMENT DESIGN

In general, we assume as the ergonomic design goal a high per-


formance of the man-machine system but keeping the workload of the
human operator within tolerable limits, in 9ther words avoiding over-
load as well as underload. If we look at todays equipment, we will
notice that we are far from reaching this goal.

The main reasons for this are:

1. Only very few designers in industry have some education in


Ergonomics. In fact, most of them have never heard of it.

2. Those designers who have some reasonable amount of education


in Ergonomics, experience severe problems in getting this
knowledge into the design process. Probable reasons:

o
Application of Ergonomics sometimes delays the design
process, at least if it comes into the design process
late.

o
It can often not be shown what the quantitative benefit
in selling the product will be.

3. The language barrier. Ergonomics is by its nature inter-


disciplinary with main contributions from Physiology, Psy-
chology and the Engineering Sciences. The professional lan-
guage reflects these contributions. Publications, there-
fore, are difficult to understand by most designers.

4. Standards. Designers are used to applying standards. How-


ever, in Ergonomics, the complexity of processes within the
man-machine systems hampers the definition of standards in
numerical form and limits the usability to terminology, func-
tional requirements etc. We shall come back to this question
a little bit later.

5. Transfer from research to application. Research results are


often written in a scientific manner, that means in a fo~
so that the researcher hopes to get accepted by his re-
searcher colleagues as a "good scientist". This, on the
other hand, often makes the paper nearly unreadable for
those designers who are more interested in having a kind
of cooking recipe directly applicable to their daily ergo-
nomic problems.

6. Design theory. Most of the reasons mentioned are more or


less correlated. A closer look shows the real problem: The
nonexistence of an "Ergonomic Design Theory".
GENERATION OF ERGONOMIC DATA 59

Three main reasons can be stated:


o The human being is too complex a "system" to be fully
understood or describable in all of his properties,
limits, tolerances and performance capabilities.
o No comprehensive mathematical tool has been available
up to now to describe and integrate all the above men-
tioned measures and findings about human behavior.
o Ergonomics is too young a science to have had a real
chance to develop the required theory.

However, it should be said that especially during the last


years a rapid development of describing and integrating
methods can be detected in the professional journals and
conferences.

But let's assume now a very favorable job situation for an


ergonomic designer:
o The manager has a positive attitude.
o There is a designer team including an ergonomist or
deSigners with a good education in Ergonomics.

What are the different levels of application of ergonomic data


and the respective requirements concerning research? Most of the
handbook data are given in the form of relationships A = f(B) or
A = f(B,C) etc. The designer with experience in Ergonomics is able
to use this knowledge in a reasonable manner, avoiding traps like
sticking too close to absolute values because he knows the variabi-
lity of human characteristics or an apparent discrepancy between the
data in two handbooks because he knows that the data are generated
under different experimental conditions. Quantitative standards are
normally based on the above mentioned data and often given as single
values. Application of these values by unexperienced designers fre-
quently leads to sup-optimal solutions and sometimes may even be
dangerous. If, for example, the 50th percentile of the eye height
of sitting male operators is used in the design of the central con-
sole for a nuclear power station, then as a consequence 50 % of the
users will have a limited view on part of the display field if they
accept the designed sitting position.

Often the design resulting from the application of standards


is checked finally by nonexperts. They sometimes reject the equip-
ment because parts of the design deviate from the standards. I can-
not agree with Chaikin who, in a paper about the value of ergonomic
standards states, that "the days of rigid adherence to all human
engineering design criteria to the nearest millimeter irrespective
60 R. BERNOTAT

r •••• rch go ••• : .qulpment .... Ign


ba.ed on:

Istandards!---+ I machine! single YQlues

.J: It.~]
U
ao
- -... Imachine!
27"1
192H
relationships
8:a ~
u
~
]a ;-~~-~~~-------------------------------------------
partial models
of
human operator

experiments

Isubjects 1.....-:.....l"!U!'!!'!_~
.J: if data and models
U
ao
. ---+ ~------,
I machine •
are insufficient
Q. or
g.
if aiming at
S optimal solution
i
E
";j
..
~

Fig. 1. Ergonomic Research and Application

of application validity and practical considerations are over"


(Chaikin, 1978).

In the Federal Republic of Germany, strong efforts have been


made in recent years to promote the establishment of Ergonomic stan-
dards. Good results have been reached as far as terminology is con-
cerned. Slow progress has been made where attempts have been made
to set up standards for a variety of users and human performance
in complex tasks (Lehmannand Saelzer, 1980).

On behalf of the International Ergonomics Association Shackel


organized in 1973 an international symposium on "Ergonomics and
Standards" (Shackel, 1973). One of the conclusions was that dimen-
sional standards are relatively well developed but that more emphasis
should be put on functional requirements and on test criteria.
GENERATION OF ERGONOMIC DATA 61

This is in agreement with German experience in using military


standards. For example, a military requirement concerning "Illumi-
nation of Operational Centers on Ships" does not request fixed illu-
mination levels but instead requires the application of design pro-
cedures given in the form of a flow diagram of decisions which can
lead to completely different solutions in different operational cen-
ters, depending on the tasks and the type of equipment (Bauvor-
schrift 35 fur elektrische Anlagen, Beleuchtungsanlagen usw. fur
Schiffe der Bundeswehr - in print -). This type of ergonomic stan-
dardization in fact contributes in a flexible manner to an ergonomic
layout of equipment. However, the applicability of standards and
data is limited because they are based on the erroneous assumption
that a combined application of apparently independent influences on
the human operator is permitted. In reality, there are strong inter-
acting relationships among the different influencing factors.

Research, therefore, has begun more than a decade ago to con-


struct "mathematical" models of human characteristics and performance
capabilities. These models integrate a larger amount of knowledge
about the human operator. And its mathematical form permits a better
prediction of the future performance of a planned man-machine sys-
stem.

In those cases where a very complex work station has to be de-


signed, for example, an aircraft cockpit or the control center of
a power station, simulation, including human subjects, is a neces-
sary tool if existing data are insufficient and/or a solution closer
to the optimum is requested. The researcher for this case has to
develop the appropriate experimental approach, especially the ap-
propriate evaluation methods.

RESEARCH IN ERGONOMIC TOWARDS EQUIPMENT DESIGN

In this section, a short survey of some factors in Ergonomic


research is given which are crucial for the usability of ergonomic
data under operational conditions. The main point is that, in ge-
neral, the operational conditions differ from experimental condi-
tions which inevitably limits the value of the ergonomic data when
applied to design. On the other hand, the multiplicity of working
conditions and type of equipment is so large that not all configu-
rations can be measured representatively. Therefore, human per-
formance is usually measured on a number of typical tasks and task
situations.

Types of experiment

The highest prediction value can be gained from field experi-


~ because they have the highest degree of realism. But the con-
ducting of such experiments in praxis has some constraints:
62 R. BERNOTAT
o The machine has to be already available. Experimental re-
sults can only lead to a few corrections in the equipment
layout. However, Ergonomics is most effective if it is ap-
plied already in early design phases, maybe even before
the first drawings of the equipment are begun.
o The physical environment is usually not under control of the
researcher in the field. Example: The performance of a manual
steering system for a ship has to be evaluated. But condi-
tions such as weather and currents, change from day to day.
o Another serious problem for the researcher is measurement
and the data collection. Many measuring methods concerning
the human being are sensitive to the physical and even soc-
ial environments.
o High costs. Performing an experiment by definition means to
change in a controlled manner one or more conditions and
see what the result is. With complex equipment such as a
large control console as the interface to a plant etc.,
using a number of subjects, one experiment may last for
weeks and months.

Because of the enormous effort and high costs the field


experiment normally is done only for simpler equipment or
not aiming at research but at improving an existing man-
machine system which has obvious ergonomic weaknesses.

The prevailing method for these reasons is the laboratory


experiment with the following advantages:
o Most of the experimental conditions can be controlled by
the researcher,
o
more measuring methods are available,
o
experiments can easily be repeated.
o It is not necessary to have the real machine. Complex
equipment in its "dynamics" and "behavior" can be simulated
on a computer in realtime having only the interface to the
human operator in hardware.

The resulting effort und costs depend on the type of equipment


and the research object. It can be a highly complex system where
the experimental setup is used over some years, as for example, the
simulator of a nuclear power plant in the Halden Project (Nesland,
1979).
GENERATION OF ERGONOMIC DATA 63

field experiment

laboratory experiment

Fig. 2. Measurements in Man-Machine Systems

Leaving specifiq equipment and focussing more on human per-


formance in classes of human tasks, such as decision making, moni-
toring etc. we find experimental setups on a much smaller scale.

But most experiments, at least in the past, have been per-


formed not with a specific type of equipment, but aimed at getting
data about more basic properties and capabilities and using only
relatively simple technical setups. Most of the data in Ergonomic
handbooks today stem from this type of research.
o Anthropometric data such as human body dimensions, reach
capabilities, muscular strength etc. (Roebuck et al.,
1975),
o human sensing capabilities such as sight, hearing, touch
etc. and the related psychophysical laws,
o
human motor activities,
o learning and motivation,
o human reactions to the physical environment such as heat,
humidity, vibration, noise, barometric air pressure etc.

Another type of data in the handbooks concerns the layout of


basic components of eqUipment such as displays and controls based
on a large number of experiments. These are the data frequently
used by designers in industry, when they try to adapt the design
64 R. BERNOTAT

" --;ocicii infl~;;ce; -',


" physical stimuli ,)
,----- -------

performance

Fig. 3. Measurement of Human Properties and Capabilities

to the human operator by ergonomic means. Some of the data are given
in the form of numbers or tables, but most data are presented in
diagrams A = feB) etc.

Measuring problems

As shown in the figure describing the field experiment quite


a number of measures have to be taken simultaneously to get an in-
sight and description of the status and the processes in the man-
machine-system.

The tasks of the man-machine-system have to be known at


any time in order to define this part of the input load
on the human operator and to be able to compare perfor-
mance results with task objectives. In complex tasks the
varying task load may not be easy to define and often we
have no technical methods to measure its changes over
time. The task of an air traffic controller, for example,
in a dangerous traffic situation may best be determined
by an experienced controller observing the whole situa-
tion.

m The social environment also is very difficult to define.


s
The structure of a team, the interrelationship between
any 2 crew members, or the presence of a high ranking
superior can have strong effects on the performance of
the human operator. However, it is difficult to define the
input load ms quantitatively and as a function of time.

m The state of the physical environment can be measured suf-


p
ficiently with no principal problems. Only some technical
effort is needed.

The measurement of human workload is one of the basic prob-


lems in Ergonomics. Up to now there is no unique and ge-
nerally accepted definition of workload (Moray, 1979~
Jahns, 1973). However, widely used today is the concept
that workload is the reaction of the individual human op-
GENERATION OF ERGONOMIC OAT A 65

erator to the loads from tasks, social and physical environ-


ment and the interaction with the machine.

It should be mentioned that physical workload can be mea-


sured relatively well whereas methods to determine mental
load need urgently to be improved. The reason for that
need is that modern technology increasingly is changing
the character of our work from physical to mental.

m.1 An important part of ergonomic design is the layout of


information flow between man and equipment. For this reason
much research effort is directed towards the measurement.
of this interaction.

Human information output via hands and feet, pushing buttons


or pedals, turning knobs or even via speech can be mea-
sured today but needs some technical equipment.

Human information input on the other hand is difficult to


measure. Although we can measure what the ·machine output
to the human operator is, it is difficult to say exactly
what information he takes from the offer. For example, ex-
pensive eye tracking equipment, which restrains head move-
ment of the subject, makes it possible to measure human eye
movement with an accuracy of about 1 degree. However, this
measure is insufficient because he may be "looking without
seeing". Comparable devices for other human senses are not
on the market.

m In general the measurement of the status of the machine


m
or equipment is not a problem.

m Also the result normally can be measured in a manner ac-


r
curate enough to compare it with the tasks given to the
man-machine-system.

Factorial Experiments

Another question about the usefulness of existing ergonomic


data is related to the single value experiments. Much of the hand-
book data stems from experiments where only one independent variable
was changed systematically by the researcher. Even if the experiment
was performed in a correct scientific manner, that means type and
number of subjects were correct and interfering variables were eli-
minated, compensated or kept constant, and the experimental design
was carefully done, the applicability of the results is limited.
The reason is that normally in practise more than one variable acts
on the human operator. What quantitative effect this superposition
has on human performance is known only in a very few cases. The
66 R. BERNOTAT

----------------,
I
HUMAN OPERATOR
I
Visual scanning
I
I Audio discrimination I
I I
l
Learning rote Probability of Hror
I Physiological chonges Time to respond
I
Environlnent I Muscle e.ertion I ResponH consistency
I I RnponH rang.
~ I Control movement

rI
Proc~ur.sl fr~u.ncy and I R.sponH accuracy

I amplituM I
I
I
~F :
I INPUT LOAD ,:
I
I
I
-:OPERATOR EFFORT:
I
I
I
I
-, PERFORMANCE

Sensory inputs which


-------------_ ....I
I

Proportion of processing Dolo outputs generat~ through


must be transfor~ capacity which must be the effort e.ert~ by the
by the operotor to sotisfy (or is) used to meet the human operator which serw as
a g.wn performance r~uirements imposed inputs to other components of
criteroon and lor to by the input load the man - mach.ne -envroonrnent
maintain a homeostatic system and proviM feedback on
operator state effort adequacy.

Fig. 4. Attributes of Operator Workload

INDEPENCENT 'tMIABlE

j
I IUotBER AN) TYPE

j
EXTRANEOUS ~BlES -----t.~ SUBJECTS -+----OF SlBJECTS
• ELIMINATE _ I REPRESENTATIVE SAMPlE)
• HOlD CONSTANT
• COMPENSATE

MEASURNENTS.OEPEfClENT VARIABlES
- SUBJECTIVE
- OBJECTIVE

Fig. 5. Experimental Factors

reason is that the manifold of possible combinations is very great


and the research in that direction as a consequence is time-con-
suming and expensive.

But designers, especially in this field, require here more


useful data. The German Research Society (DFG) has put this topic
in a high rank order in the list of urgent research in Ergonomics
(Rutenfranz et al., 1980).
GENERATION OF ERGONOMIC DATA 67

independent variables

\\1//__
Extraneous variables _ _ _.....1 subject ..,I~ ~_ n subjects
• • (representative sample I

11\
dependent variables

Fig. 6. Multifactorial Experimental Design

Number of subjects

It is a well accepted requirement among researchers that the


number of subjects should be large enough to get valid data and
draw conclusions, with a confidence that can be estimated by stat-
istical means. This is not a scientific problem. But in practice
many researchers do have a problem in getting the right subjects
in .sufficient numbers for the necessary duration. Moreover, avail-
ability of experimental apparatus, money restrictions or other
reasons may force them to reduce the number of subjects.

An error which can be estimated occurs when empirical data is


transformed by a smoothing process into an assumed distribution.
Another error occurs if a conclusion is drawn concerning the dis-
tribution of the population of future users. The researcher normal-
ly in his publication makes an estimate by "defining an "interval of
confidence" •

The real problem arises if this final distribution is taken as


fully exact and applied to design or even to standards. Chaikin has
found in his search for original sources of standards that recommen-
dations based on observations rather than on scientific results some-
times later on got into the form of standards (Chaikin, 1978) or
"their origin is not iunnersed in research involving legions of test
subjects". The same author noticed, on the other hand, that the data
have withstood the test of time.

Types of subjects

A much more serious effect may occur if subjects used are not
"representative", that means that they are not typical for the
future user.
68 A. BERNOTAT

Ix;plctancy 11
standard deviation C1

pix)
population

normal distribution

11-0 11+0

assumed values
for
mean and S.D.

sample
d
assumed distribution
10

Fig. 7. Typical Sample and Population

Two examples may illustrate the effect:

a) Many ergonomic data are collected in research institutes and


universities. The usual subject there, available in large
numbers and at no cost for the researcher, are students.
Little can be said against that approach from the scientific
viewpoint. But in the migration to the handbooks often a lot
of important information is filtered out. At best, the size
of the sample and the age of the subject is given. Without
knowing the origin, the average designer will use the data.

A fictitious example: The pulse rate of a number of students


in Psychology was measured as a function of loudness of
white noise. The students were 20-22 years old, knew the
scientific question, were paid for participation, and for
GENERATION OF ERGONOMIC DATA 69

B
.B performance degradation population
.;: due to fatigue end/or performance
"til stress distribution
:.0 , .... - . . ,
of users
r:-
~ I" "
o I
.c
[ ~
I
)(

sample
performance of
subjects abcNe average
e.g. RI D staff

Fig. 8. Misleading Inference from Experiments with Atypical Subjects

both reasons were highly motivated. There are serious doubts


whether the results can be used to estimate the annoyance
of elderly people living near a military airfield and having
a negative attitude against the jet noise of landing air-
craft.

b) Another example of atypical subjects. In experiments often


the research staff or, during the design process, the engi-
neers themselves are used as subjects.

In many ways this sample is by selection, education and moti-


vation above average. In other words, in applying these data
to the general population, 50 % of the later users may have
a performance lower than expected for the design.

I do not say that test pilots or test drivers should not be


used. There are some gOOd reasons for doing this, for example,
70 R. BERNOTAT

the smaller time effort. But the designer should be very


careful not to use the results 1 : 1 compared to other
future users.

Experimental conditions adequate to the working conditions

If under normal operational conditions a performance degradation


due to fatigue can occur this should be taken into account in the
design. If the necessary data are not available in the literature,
special experiments may have to be done for critical design aspects
of the equipment with fatigued sUbjects.

Another reason for severe performance degradation is stress.


This normally is not taken into account especially in the develop-
ment of weapon systems and often leads to a high rate of human-
induced system failures under operational conditions. The "General
iAccounting Office" of the Congress vf the United States about a year
ago, appropriately had in its report to the Congress of the United
States a strong recommendation tu foster human factors .research and
application so that "specifications and handbooks" address adequate-
ly human factors, such as skill levels, proficiency, availability,
environmental stress and fatigue. The research problem is that battle
stress and fear and the resulting performance degradation cannot be
stimulated, at least not in the usual laboratory experiment, because
subjects know that they are not taking part in a dangerous experi-
ment.

Learning time

Even for apparently simple tasks, such as reading displays


human performance improves ~onsiderably with learning (Murrel, 1969).
It makes a difference whether the subject learns to read one type
of display or has to report the momentary values of different dis-
plays in a comparison experiment.

For more complicated tasks with, for example, integrated elec-


tronic displays giving a large number of values necessary for con-
trol of complex equipment, the learning time may be weeks or even
months. The question for the researcher and user is, at what level
of learning does it make sense to compare performance levels?

The necessary time required for this type of experiment, again,


accounts for the small amount of such data in the handbooks and pos-
sibly is one reason for apparent differences in published data.

Research goals in Ergonomics

Where are we, where do we go in research? The simplified scheme


in the figure shows the main levels.
GENERATION OF ERGONOMIC DATA 71
comparison reasonable? subject 1

error ~ subject 2
subject 3

I
A

perf-l
statIonary J
l'
-
'",.,,,

2 4 6 8 10 12 14 16 18 20
NPUcatilllll of....,....
~

.....
\ear...... time "I
Fig. 9. Learning Curves and MMS-Performance

In the past, effort had been directed towards collection of


data and their analysis. Today research is aiming at the integration
of knowledge by setting-up human operator models which describe cer-
tain human properties and capabilities in the form of partial models.
An extensive survey of this approach can be found in the report of
Pew et ale (1977) and in the conference proceedings of Johannsen
and Sheridan (1976). These models describe human control behavior,
for example, in a stabilization task (Johannsen et al., 1976), or
human guidance behavior, such as a driver with preview of the road
ahead (Donges, 1978). During recent years a growing interest can be
noticed towards models describing human monitoring, decision making
and even planning behavior (Sheridan and Ferrell, 1974; Johannsen et
al., 1981).
Another research approach uses network models where the human
task performance is part of a network of system functions (Wortman
et al., 1977; Seifert and DOring, 1981). Network models as developed
and used in System-Ergonomics are a very promising tool with a real
good chance of being applied to the design of complex MMS.
An important long term goal for the future is the development
and testing of a comprehensive design theory integrating the dif-
ferent contributions. There are some signs that a design theory may
develop faster in the next decade (Rouse, 1981; Spillers, 1974).

SOME RECOMMENDATIONS FOR RESEARCH AND APPLICATION


Drawing conclusions from the observations and statements in
this paper we can set up the following list~ I want to say again
that this list is not at all complete, and only gives some of the
important aspects in the equipment design facet of Ergonomics.
BtzC
72 R. BERNOTAT

• COLLECTION OF DATA
-data Ax A
-diagrams

d~
-distri but ions

Ax
• ANALYSIS
determination of el""ents
and their relationship

• SYNTHESIS
integration of knowledge
about
influences
al the human ,l?roperties
and capabilities in form
J
of partial models

bl human task performance


as part of networks

• DESIGN THEORIES
de'lelopment of higher order design-theories based
on the synthesis-models leading to ergonomic design
procedures for man-machine-systems which are
teachable and applicable.
Evaluation methods are an important part of these theories

Fig. 10. Research Goals in Ergonomics

Research
o Measuring methods have to be improved especially concerning
mental workload and influences from the social environment.
o
Some agreement or even "standardization" of basic measuring
procedures would help to make data comparable.
o
Moxomultifactorial experiments should be started to get per-
formance predictions closer to real World performance. How-
ever, because such research requires long term efforts,
funding may be difficult.
o Software Ergonomics with increasing automation and computer
application becomes an increasingly interesting and impor-
tant topic.
o
More research should be directed towards the assessment of
human failures, their type and probability, so that such
GENERATION OF ERGONOMIC DATA 73

,,'-- ....... , ....


,
planning t::____::::~
"
olternative pions

guidance
4----1 .rror
pion execution

stabilisation ~~ control of subsystems

system

Fig. 11. Hierarchy of System Management and Control

knowledge can be incorporated into system designs (Swain and


Guttman, 1980).

o More consequently than in the past researchers in Ergonomics


should be made to cooperate with researchers in system-theo-
ries and design-theories in order to foster an integration
of Ergonomics (Swain and Guttman, 1980).

Application

o Improve the education of designers in Ergonomics (Bernotat


and Hunt, 1977). Because Ergonomics is absorbing and inte-
grating contributions from many disciplines, it is unavoid-
able that terms from many areas of knowledge come together.
Education of users is one way to overcome the language barrier
between researcher and user.

o Ergonomic Standards in numerical form are possible only in a


very few cases. Standards for terminology, functional re-
quirements and test criteria have a far better chance of be-
ing applied.

Another possible approach used by some German authorities is


to officially require the organization and participation of
Ergonomists instead of increasing the number of ergonomic
standards and requirements. This solution in the long run is
probably the more flexible and least expensive solution.
74 R. BERNOTAT

o Setup of ergonomic data bases on a national or international


level. Until now most useful data bases - if they existed at
all - have been in the development of military systems.

REFERENCES

Anonymus, 1981, Effectiveness of U.S. Forces can be increased through


improved weapon system design, GAO Report to the Congress of
the United States, PS~81-17.
Anonymus, Bauvorschrift 35 fur elektrische Anlagen, Beleuchtungsan-
lagen fur Schiffe der Bundeswehr, Bundesamt fur Wehrtechnik
und Beschaffung, Koblenz (in print).
Bernotat, R., and Hunt, D., 1977, "University Curricula in Ergonom-
ics," Forschungsinstitut fur Anthropotechnik, Meckenheim.
Chaikin, G., 1978, Human Engineering Design Criteria - The Value of
Obsolete Standards and Guides, U.S. Army Human Engineering
Laboratories, Technical Note 13-78.
Donges, E., 1977, A Control Theoretic Model of Driver Streering Be-
havior, in: "Proceedings of the 13th Annual Conference on
Manual Control," Cambridge (Mass.).
Donges, E., 1978, A two-level Model of Driver Steering Behavior,
Human Factors, 20:691.
Jahns, D., 1973, A Concept of Operator Workload in Manual Vehicle
Operations, Rep. No. 14, Forschungsinstitut fur Anthropotech-
nik, Meckenheim.
Johannsen, G., and Sheridan, T., eds., 1976, "Monitoring Bevhavior
and Supervisory Control," Plenum, New York (N. Y.).
Johannsen, G., Boller, H., Donges, E., and Stein, W., 1977, "Der
Mensch im Regelkreis, Lineare Mode lIe ," Oldenbourg, MUnchen.
Johannsen, G., Rouse, W., and Hillmann, K., 1981, Studies of Planning
Behavior of Aircraft Pilots in Normal, Abnormal and Emergency
Situations, Forschungsinstitut fur Anthropotechnik, Werth-
hoven.
Lehmann, K., and Saelzer, H., 1980, Zukunftige Normungsschwerpunkte
am Beispiel der Sicherheitstechnik und Ergonomie, Z. f. Ra-
tionalisierung, 31:6.
Mor ay , N., ed., 1979, "Men tal Workload, its Theory and Meas ure-
ment, II Plenum, New York (N. Y.).
Murre 11, K., 1969, "Ergonomics," Chapman and Halt, London.
Nesland, K., 1979, Measurement of.operator performance - an experi-
mental setup. Halden Reactor Project, Halden, Norway, in:
"Proceedings of the Specialist Meeting on Procedures and
Systems for Assisting an Operator During Normal and Anomalous
Nuclear Power Plant Operator Situations," Mfinchen.
Pew, R., Baron, S., Fechner, C., and Miller, D., 1977, Critical Re-
view and Analysis of Performance Models applicable to Man-
Machine Systems Evaluation, Bolt, Beranek and Newman, Rep. No.
3446.
GENERATION OF ERGONOMIC DATA 75

Roebuck, J., Kraemer, K., and Thomson, W., 1975, "Engineering Anthro-
pometry Methods," Wiley, New York (N. Y.).
Rouse, W., 1981, Human-computer interaction in the control of dynamic
systems, Computing Surveys, 13:72.
Rutenfranz, J., Luczak, H., Lehnert, G., Rohmert, W., and
Szadowski, D., 1980, "Denkschrift zur Lage der Arbeitsmedizin
und der Ergonomie in der Bundesrepublik Deutschland," Boldt,
Boppard.
Seifert, D., and Doering, B., 1981, SAINT - Ein Verfahren zur Model-
lierung, Simulation und Analyse von Mensch-Maschine-Systemen,
z. f. Angewandte Systemanalyse, 2:127.
Shackel, B., ed., 1973, "International Symposium on Ergonomics and
Standards," Proceedings University of Technology, Loughborough.
Sheridan, T., and Ferrell, W., 1974, "Man-Machine Systems: Infor-
mation, Control and Decision Models of Human Performance,"
MIT Press, Cambridge (Mass.).
Spillers, W., 1974, "Basic Questions of Design Theory," North-Holland,
Amsterdam.
SWain, A., and Guttman, H., 1980, Handbook of Human Reliability Ana-
lysis with Emphasis on Nuclear Power Plant Applications,
Sandia Laboratories, Albuquerque, prepared for Office of
Nuclear Regulatory Research, Washington, D. C., NRC FIN, No.
A 1188.
Wortman, D., Duket, S., and Seifert, D., 1977, Modeling and Analysis
using SAINT, a combined discrete/continuous Network Simula-
tion Language, in: Proceedings, Winter Simulation Conference,
U.S. Nat. Bur. of Standards.
SOME LIMITATIONS TO ERGONOMIC DESIGN

J. Moraal

Institute for Perception TNO

Soesterberg / The Netherlands

INTRODUcrION

In coping with his environment man has to perform numerous and


diversified tasks, be it at home, in his work or in recreational or
social acitivities. Tasks vary from very simple to highly complex,
in sort and amount of required experience and skills, like precision,
speed and strength, and in whether or not tools or aids are used to
enlarge human capabilities. These tools and aids - usually called
"machines" - also vary from simple to highly complex, e.g. a spoon
versus a space capsule. Tasks have to be performed efficiently, com-
fortably and safely, at lea~t in the opinion of the ergonomist or
human engineer. In order to meet these criteria, it is not very ad-
equate to merely tell people how to behave because, although they may
be well motivated, it mostly is not known what exactly efficient
and safe behaviour is, how it should be defined and what kind of
capabilities are required to behave accordingly.

Man's capabilities are limited. Viewed as a system that pro-


cesses information from the environment, he cannot avoid errors of
perception, decision making and action. The simple reason for this
is, that the total amount of information impinging on man's sensory
system is somewhat in the order of 5000 to 10,000 times the amount
the human brain is capable of handling. Therefore, the division of
attention and the optimal allocation of mental effort is a problem
of utmost importance.

Whatever as the definition of human (factors) engineering or


ergonomics is given, the basic principle seems clear: the adapta-
tion of the "(technical and physical) environment and machines to
man, in order to make optimal use of his capabilities when perfo~

77
-.J
(X)
Table 1. Performance Shaping Factors

EXTERNAL STRESSORS INTERNAL

Task and Equipment


Situational Characteristics Characteristics Psychological Stressors Organismic Factors

Architectural Features Perceptual Requirements Suddenness of Onset Previous Training/


Quality of Environment: Motor Requirements Duration of Stress Experience
Temperature, Humidity, (Speed, Strength, Task Speed State of Current
and Air Quality Precision) Task Load Practice or Skill
Lighting Control-Display High Jeopardy Risk Personality and
Noise and Vibration Relationships Threats (of Failure, Intelligence
Degree of General Anticipatory Requirements Loss of Job) Variables
Cleanliness Interpretation Monotonous, Degrading, Motivation and
Work Hours/Work Breaks Decision-Making or Meaningless Work Attitudes
Availability/Adequacy of Complexity (Information Long, Uneventful Vigi- Knowledge of Required
Special Equipment, Load) lance Periods Performance
Tools, and Supplies Narrowness of Task Conflicts of Motives Standards
Manning Parameters Frequency and Repeti- about Job Performance Physical Condition
Organizational Structure tiveness Reinforcement Absent Attitudes Based on
(e.g., Authority, Re- Task Criticality or Negative Influence of
sponsibility, Communi- Long- and Short-Term Sensory Deprivation Familiy and
cation Channels) Memory Distractions (Noise, Other OUtside
Actions by Supervisors, Co- Calculational Require- Glare, Movement, Persons or
workers, Union Represent- ments Flicker, Color) Agencies
atives, and Regulatory Feedback (Knowledge of Inconsistent Cueing Group Identifications
Personnel Results)
Rewards, Recognition, Continuity (Discrete Physiological Stressors c..
Benefits vs Continuous) 3:
Team Structure Duration of Stress
o::zJ
Fatigue ~
Pain or Discomfort r-
Job and Task Instructions Man-Machine Interface Hunger or Thirst
~
i:
Factors: Design of Temperature Extremes m
r-
Procedures Required Prime Equipment, Radiation ~
(Written or not written) Test Equipment, G-Force Extremes :::j
Written or Oral Communica- Manufacturing Equip- Atmospheric Pressure ~
tions ment, Job Aids, Extremes o
Cautions and Warnings Tools, Fixtures Oxyge? Insufficiency z
(J)
Work Methods Vibration
Plant Policies (Shop Movement Constriction d
m
Practices) Lack of Physical ~
Exercise o
z
o
i:
n
cm
(J)
C5
z

-.oJ
CD
80 J.MORAAL

ing tasks. The recipe is simple. Firstly, find out what has to be
done by an analysis of the aims of a system and of the task de-
mands. Secondly, find out what man can do, i.e. investigate his per-
formance capabilities. Thirdly, fill the gap between both sets of
data by choosing the right design of equipment and environment and
by the right choice of task aids. Given, that man or the human op-
erator is well-selected, well-trained and well-motivated, the final
result will be optimal task performance. However, this recipe is too
simple. Man is not a given entity, always operating in the same way,
stable under different conditions of the environment, unaffected by
personal and emotional matters and with predictable performance. In-
stead, human performance is difficult, sometimes very difficult to
predict. Apart from what the human engineer wants that the operator
brings with him in the task situation, like his skills, experience
etc., he always brings more: his motivations, beliefs, wishes, ex-
pections and, in particular, his variability.

Any given operator in a man-machine system differs from all


other operators and will frequently show remarkable variability in
behaviour from day to day and even from moment to moment. In fact,
people never do anything exactly the same way twice. This problem
area is one of the main reasons that, until now, no precise ap-
proaches exist for modelling human behaviour in all its complexities
and with all its interactions.

So, engineering according to man's capabilities generally seems


to be limited by at least one basic human characteristic, i.e. his
variability. If prediction of human performance is possible at all,
then only within certain error bounds.

PERFORMANCE SHAPING FACTORS

In man-machine system analysis the human being often is treated


like other components or elements of the system. But, because of his
basic variability, man differs from all other elements. All the pro-
cesses of perception, thinking, decision-making, acting, and learn-
ing are influenced by what are called performance shaping factors
(Swain & Guttmann, 1980). These factors are not only operating in
isolation but very often interact to a complex degree, making it
very difficult for the designer of equipment and workplaces to assign
weighting factors to each to calculate their relative influence on
task performance. Nevertheless, careful investigation of the effects
of performance shaping factors on task performance is a prerequisite
to human engineering.

Generally, performance shaping factors are subdivided in task/


equipment factors, environmental or situational factors, procedur-
al factors, organismic factors and stressors. Table 1, taken from
Swain & Guttmann (op. cit.) reflects a survey. It is not within the
SOME LIMITATIONS TO ERGONOMIC DESIGN 81

scope of this contribution to discuss all the performance shaping


factors in detail. The aim is to illustrate that good human engin-
eering practices cannot do without considering; all other factors
beyond those that are strictly related to equipment design. Swain
& Guttmann (cp. cit.), in analyzing human error probability in
nuclear power plants, i.e. the probability that when a given task
is performed an error will occur, also include the use of well-
designed written procedures and checklists, and frequent practice
of appropriate responses to potential emergencies in the applica-
tion of good ergonomic practices (see Table 2).

Table 2. Estimated Decreases in Human Error Probability (HEP)


Resulting from Application of Good Ergonomic Practices
to Nuclear Power Plants (from Swain & Guttmann, 1980)

If done: Decrease in HEP


(factors)

Good human engineering practices in 2 to 10


design of controls and displays

Use of well-designed written procedures 3 to 10


and checklists to replace typical nar-
rative style procedures

RedesiC))Il of displays or controls that > 10


violate strong populational stereotypes

Redesign of valve labelling to indicate "'V 5


their functions (including a clear indi-
cation of the system with which a valve
is associated) and also to show clearly
their normal operating status

Frequent practice of the appropriate 2 to 10


responses to potential emergencies of
other abnormal situations (practice
includes periodic recertification in
dynamic simulators and talk-throughs
conducted at least once per month for
the major potential problems)

The range in the factors in the right column of Table 2 refers to


human variability, i.e. inter-individual variability.
82 J.MORAAl

The question arises, then, whether it is possible for the human


engineer to deal effectively with the variability between individu-
als.

ANTHROPOMETRIC VARIABILITY

One of the most frequently encountered sources of variability


equipment designers have to deal with is body shape. Anthropometric
data show normally distributed characteristics, like length and
weight. Tables 3 and 4 present data of an anthropometric survey of
2000 British car drivers (Haslegrave, 1980). In using these kind of
data the general strategy, then, is to decide upon the percentile
range as a basic for e.g. vehicle design. The problem is, how to in-
clude all the relevant information from such a data bank in the de-
sign? Otherwise stated, what is the relative weight of each of the
dimensions of a data-file like the one presented in Table 3 and 4?
One way to solve this problem is to condense the data by applying
factor analysis, to reduce the large number of dimensions to a
manageable proportion, but containing all relevant information. A
comparative study of 15 factor analyses of anthropometric data
showed that only two major factors were common: length and weight
(see Haslegrave, op. cit.). However, Haslegrave concludes that these
two factors only are not sufficient in reconstructing body size and
shape for a certain population and, therefore, insufficient to con-
struct representative dummies for design and safety research.

After normalizing the data from Tables 3 and 4 for length and
weight, Haselgrave was able to extract other factors influencing
body shape, namely, torso depth, limb length and torso width, to-
gether accounting for about 33% of the variance in the male and
female samples. These factors then, seem to represent main features
of body shapes.

The distributions of the values of the constructed indices


for these factors (together with those for mean body diameter) are
presented in Table 5. These body indices were suggested as a basis
for the development of anthropometric dummies.

In a follow-up study, Hutchinson & Haslegrave (1980) showed


that pattern of body shape even can be revealed by the use of part-
ial correlations successively eliminating dimension like age, sta-
ture and weight, so that one can do without the need for the com-
plexities of a technique as factor analysis.

PERFORMANCE VARIABILITY

Besides anthropometric variability there are, of course, nu-


merous other human characteristics showing variability, like those
en
0
~
m
r-
~
::::j
Table 3. Anthropometric data of male drivers >
-t
(from Haslegrave, 1980) 6
z
en
Sample Standard Percentile Value -t
Size Mean Deviation 0
m
Dimension N it S 5 10 20 30 40 50 60 70 80 90 95 99 ::u
c;')
0
Weight, kg 1584 74·3 11-17 51·5 58-0 61'2. 65'3 68·2 7()O8 73-3 76-0 79·1 82'7 88'4 94'1 104'8 Z
Stature, mm 1584 1738-1 68·00 1573 1626 1655 1682 1704 1722 1737 1755 1773 1794 1824 1851 1892 0
Eye height, mm 1584 802·7 33-95 713 746 760 776 786 795 804 811 821 830 846 857 881 ~
Shoulder height, mm 1584 62H 29-20 550 573 585 598 606 614 621 629 636 646 657 668 690 n
Neck width, mm 1583 140-4 11·86 109 123 127 131 134 137 140 142 145 149 ISS 161 172 C
m
Shoulder width, mm 1583 37(}7 23'02 314 3~4 342 353 360 365 371 375 381 388 397 410 435 en
Chest breadth, mm 1583 311'0 21'89 266 279 285 293 300 304 ~IO 315 320 328 338 349 372 C5
Seat breadth, mm Z
1582 375·2 26-08 322 336 344 354 361 367 )73 379 387 395 410 421 447
Chest depth, mm 1582 247:2 24·26 192 209 217 227 234 241 247 254 260 267 278 286 306
Stomach depth, mm 1582 267'6 35·27 206 220 227 237 246 254 262 273 283 296 314 330 365
Thigh depth, mm 1584 16(}S 14·81 125 136 142 149 153 157 160 164 168 172 179 185 197
Knee height, mm 1583 563-4 28'19 493 520 529 S40 548 556 563 570 578 586 600 611 631
Buttock-knee
length, mm 1582 610·7 29'16 543 563 574. 586 596 604 611 618 626 635 647 659 684
Arm length, mm 1073 786·3 34'78 708 730 743 757 768 776 784 794 804 815 832 844 875
Shoulder-elbow
length, mm 1583 379·4 17-71 339 351 357 365 370 375 379 384 388 394 403 409 422
Elbow-fingertip
length, mm 1584 468·7 2(}74 421 436 443 452 458 463 468 474 479 485 496 S03 519
Shoulder slope,
degrees 1578 0 0 0 0
2l-1° 3-45° 13·2" 15·5< 16'8" 18'1" 19·2° 20·2' 21-1' 22'0° 22'9° 24-0 25'6 26'9 29-0

(XI
Co)
(Xl
~

Table 4. Anthropometric data of female drivers


(from Haslegrave, 1980)

Sample Standard Percentile Value


Size Mean Deviation
Dimension N x S 5 10 20 30 40 50 60 70 80 90 95 99

Weight, kg 416 62'1 9·52 4J8 49·8 51·8 54·4 56·6 58·5 61'1 63·0 65'6 68·6 74·3 79·9 90·7
Stature, mm 416 1624·5 56·01 1493 1537 1553 1578 1594 1609 1623 1637 1652 1672 1699 1719 1763
Eye height, mm 416 759·7 29·03 689 707 722 735 746 754 762 769 775 784 795 805 821
Shoulder height, mm 416 580·2 25·00 526 540 548 557 567 574 580 587 593 602 614 623 635
Neck width, mm 415 129·7 13042 97 108 112 120 123 126 130 133 137 140 147 152 160
Shoulder width, mm 416 339·4 22·68 289 304 313 320 327 334 338 344 350 356 368 381 398
Chest breadth, mm 416 281·8 21·55 240 250 258 265 270 274 279 284 291 297 311 319 350
Seat breadth, mm 416 364·0 26·92 310 325 333 342 349 356 362 369 376 385 399 412 437
Chest depth, mm 415 251·7 28·34 203 212 220 228 234 239 247 255 264 273 291 305 333
Stomach depth, mm 416 253-3 34·11 197 207 215 224 233 239 248 258 267 278 297 320 359
Thigh depth, mm 416 156·0 15·66 126 134- 138 143 147 151 155 159 162 167 176 181 204
Knee height, mm 415 540·6 26-73 479 499 507 518 526 534 540 547 553 562 575 588 606
Buttock-knee
length, mm 415 600·9 26·89 540 560 567 577 586 594 601 608 616 622 636 646 666
Arm length, mm 257 721·8 30·34 656 672 681 698 707 715 724 729 736 742 760 771 801
Shoulder-elbow
length, mm 416 351·4 15·82 316 327 330 338 343 347 351 355 360 365 371 378 393
Elbow-fingertip
length, mm 416 426·8 18·32 382 397 403 412 418 422 426 432 436 442 450 456 473
Shoulder slope,
u
degrees 413 22·2 3·77 12·8 15·8 17·2 19-4 20·3 21-2 22·2 23-2 - 24·6 25-7' 27·0' 28T 30·l

~
s:
0
:tl
»
»
r
SOME LIMITATIONS TO ERGONOMIC DESIGN 85

Table 5. Indices of body size and shape (from Haslegrave, 1980)

INDEX MEAN STANDARD MEDIAN RANGE


DEVIATION

Mean body Male 232.7 15.92 231.8 182-314


Diameter, rom Female 220.0 15.84 218.1 185-300

Torso Depth Male 2.2108 0.15424 2.1966 1. 777-2. 718


Index Female 2.2913 0.15722 2.2717 1.924-2.729

Torso Width Male 4.5502 0.17617 4.5482 3.976-5.444


Index Female 4.4851 0.17799 4.4790 3.861-5.201

Limb Length Male 1.1636 0.02354 1.1631 1.087-1.250


Index Female 1.1819 0.02456 1.1822 1.123-1.252

that have to do with performance. In practical situations the total


range of variability mostly is restricted by having people selected
and trained. Furthermore, those who are performing far below the
group average can be retrained or transferred, while those perform-
ing consistently at a high level can be promoted.

Wechsler (1952) measured a large number of human abilities.


He noted that the usual measure of variability, the standard devi-
ation, was highly sensitive to the shape of the distribution.
Therefore, he defined a different measure, the total range ratio
as " ... the ratio of the highest score to the lowest score of a
group of people homogeneous with respect to what is being measured,
but excluding the extreme scores, i.e. the lowest and highest tenth
of 1% of the population" (see Swain & Guttmann, 1980).

Wechsler found that, when taking quantity of output (not qual-


ity), the ratio of the scores of the best performers to the scores
of the worst performers generally to be in the order of 3 : 1 to
5 : 1, while for production tasks in particular, the ratio approach-
ed 2 : 1. Rook (see Swain & Guttmann, Ope cit.), investigated per-
formance variability in number of defects (accuracy) and concluded
that " •.. the best workers are almost never more than three times
better than the worst workers. In fact, a ratio of two to one is
more typical."

HUMAN PERFORMANCE AND HUMAN BEHAVIOUR

In human engineering design of machi'1es end equipment the basic


data on human capabilities not seldom are drawn from studies on
86 J. MORAAL

performance, i.e. what people can do, instead of behaviour, i.e.


what people actually do. The following example is taken from road
user behaviour studies.

Good task performance of traffic participants means a.o. that


they stop in time for a red traffic light. From human performance
studies we know that 8% of the men and 2% of the women fail to
discriminate between red and green. Therefore, traffic lights were
adapted by shifting the red somewhat to the orange and the green
to the blue, making discrimination possible for practically the
whole population. Given that good care is taken to aspects of vis-
ibility, conspicuity, placement etc. of the traffic lights, it can
be assumed that this particular information is presented in an
optimal way.

However, when observing actual traffic behaviour it turns out


that, at least in some countries, traffic participants, car drivers
as well as cyclists and pedestrians often violate the rules. The
required behaviour for red and green is unambiguous, whereas dif-
ficulties may arise with the presentation of amber: One has to stop,
except when, e.g., because of too short a distance, stopping is not
well possible, leaving the road-user with a decision problem. Fac-
tors entering this decision-making process are the road-user's moti-
vation, the predictability of the situation, an estimate of the ef-
fects of ignoring the amber (or even red) light, like conflicts
with other traffic participants, an estimate of the effects of a
sudden stop, like discomfort, waiting time, following traffic, etc.
Not seldo~ the outcome of this decision-making process is running
the red light.

In looking for a solution to this problem the literature con-


verges to a prolongation of amber duration. However, in the opinion
of Gazis et al. (1960), "Drivers are inclined to ignore a long amber
phase and treat i t as merely a continuation of the green phase. As
many drivers, if not more, will go through the red light when the
amber phase is too long, as will do so if it is too short". Recent
evidence seems to point to the conclusion that for car drivers the
decision to stop is made independent of the amber duration, at least
in the range of 3 to 5 s. Deceleration level seems to be much more
essential. Fig. 1 is an example of a distribution of the probab-
ility to stop as a function of the deceleration level.

From several studies it seems that the 85th percentile of the


probability to stop is about 2.0 to 2.3 mis', which is experienced
as an acceptable or comfortable level. A possible solution to po-
sitively influence the so-called "red-light discipline" of car
drivers is to install detectors in the pavement at an adequate
distance from the traffic lights, such that the duration of the
green phase is prolongated for those cars having passed the de-
tectors, while the drivers that still have to pass the detectors
SOME LIMITATIONS TO ERGONOMIC DESIGN 87

1.0 .-. • speed limit= 30 mph


a.

.'\
o \/\ (48.3km/h)

. •
-; 0.8

~0.6

:c
.8 0.4
ea. 0.2

decleration level (m/s2)

Fig. 1. Probability to stop just before the marking line as a


function of the level of deceleration (Olson & Rothery,
1972)

are still able to stop with a comfortable deceleration level. In


a before/after study of this measure Zegeer & Deen (1978) observed
a significant decrease in number of conflicts, like violations,
sudden stops and accelerations through amber.

This example illustrates that when design of equipment to im-


prove human functioning, is based on performance data alone, there
is a serious chance that the ultimate behaviour one has in view is
still not reached. In the example the ultimate aim of traffic lights
is not only discrimination of the different colours but to have
people stop when it is necessary, and safely to do so. The only
validation then lies in a registration of dynamic actual behaviour
over time in the real situation.

A point of general discussion raised here is that it may answer


the question why human engineers or ergonomists often complain that
ergonomics is overlooked by deSigners of industrial and military
systems. In Edwards' view (1981) it is because ergonomists tend to
be laboratory- rather than problem-centred. We could say performance-
rather than behaviour-centred. In the laboratory the problem is re-
moved from its context, extrinsic variables are eliminated and con-
trolled experimentation in which all the variables are systematical-
ly manipulated becomes feasible. But, as Edwards (op. cit.) states:
"Unfortunately, the removal of a problem from its context can rare-
ly be accomplished without the loss of concomitant factors whose
importance is unknown and whose neglect is likely to invalidate the
solution for real-world application". This is of course not to say
that laboratory experiments have no value for ergonomics, although
according to Chapanis (1976) their value is limited and at best are
"rough and approximate models of real-life situations". Or, as
De Greene (1980) puts it: "In short, much human factors/ergonomics
research produces isolated segments, slices, or vignettes of human
88 J.MORAAL

behaviour". What has to be underlined here is not any controversy


that might exist between laboratory- on the one hand and field-tests
and experimentation on the other hand, but the difference between
studies in which the subject knows that he is being studied versus
unobtrusive registration of his behaviour in realworld tasks. As
long as the subject knows that he is being studied there is always
a chance that he is doing better than normally because he is spec-
ially motivated or behaves according to what he thinks the experi-
menter wants him to behave.

Video or film techniques can, of course, be used in the un-


obtrusive observation of overt behaviour, for example, traffic
participants. Meanwhile, also systems are developed and being test-
ed for the continuous, unobtrusive monitoring of covert behaviour,
like ECG, EEG, respiration, blood pressure, limb movements, produc-
tion output and sencondary task performance (Knight et al., 1980).
According to the authors such methods ..... would significantly aid
in: (1) validating laboratory results in a realistic work environ-
ment, and (2) providing an adequate qata base for developing first-
hand insight of sources of work-stress and consequences of long-term
exposure to adverse work environments". Such a system has to be un-
obtrusive with respect to impact on workers and production rates,
capable of long-term data collection, and must permit simultaneous
monitoring of various dependent variables of many workers.

CONFLICTING CRITERIA

When design criteria in human engineering pertain to efficiency,


comfort and safety, there will be no problem when various design
actions converge in reaching all three. But not seldom the three
are in conflict like, for example, when designing for safety leads
to a degradation of performance. This may be illustrated by an ex-
ample from the smallest man-machine system: clothing.

In caring for safety of military people various types of (NBC)


protecting clothing are developed. Table 6 shows part of the results
of a study in which three types of protective clothing were com-
pared for three categories of military personnel, with regard to
performance in tasks like fire power, communication and mobility.
It is clear from Table 6 that degradation of task performance does
occur and this could mean that task performance becomes ineffective.
The results of several studies on the degrading effect of protective
clothing point to the conclusion that degradation possibly can be
overcome by having soldiers well-trained and motivated, together
with good physical fitness, and familiarity with and confidence in
his equipment.

From this problem it may be clear that one of the aims of


ergonomics, i.e. deSigning for safety (and health), can lead to
SOME LIMITATIONS TO ERGONOMIC DESIGN 89

degraded task performance. So, if the human engineer wants to safe-


guard task performance at a certain level he could find himself
very much limited in designing safe equipment. The problem of safe-
ty is, however, much broader than that for the soldier, fireman or
any other working man. It has to do with toys and bicycles as well
as with spacecraft and nuclear power plants, and has become a matter
of legislative concern in most societies. However, as Chapanis (1979)
states: "There is no way to make anything perfectly safe". Accord-
ing to this author it is therefore necessary that ergonomists de-
velop a cost-benefit mentality. Costs, not only in terms of design
costs but also in terms of operating the system at a degraded level.
"Every recommendation we make, every ergonomic change we propose
has a cost associated with it. We need to be constantly aware of
those costs and to assure ourselves that the benefits of what we
propose outweigh the costs. Unless we do so, the world at large will
viewus as unrealistic theoretics and dreamers" (Chapanis, op. cit.).

FUTURE RESEARCH

In future research human factors engineering or ergonomics


should heavenly concentrate on the factors that limit its activities.
Because of the variability as a basic human characteristic, human
engineering can contribute to, but not determine optimal task per-
formance. Therefore, the more sophisticated techniques are develop-
ed and used for dealing with variability, the more adequate equip-
ment and machines will be designed.
In reaching a better data base, i.e. not only performance-
but also behaviour-oriented, more and better techniques should be
developed and used for the (unobtrusive) observation and re.gistra-
tion of behaviour. For example, the registration of road-user be-
haviour the development of equipment for the automatic analysis of
video images is under way.
As was already said, a cost-benefit mentality in ergonomics
should be developed, with cost taken in a broad sense, e.g. also
the required amount of training. Finally, for whatever machines are
designed, people are needed for manning and operation of them.
In view of today's high level of technology and compleX systems
one of the urgent actions is to design for simplicity. With high
automation one would expect that systems could be staffed by per-
sonnel with lower-level skills. However, the opposite seems to be
true which is, according to De Greene (1980) because of the. fact
that "... adequate human performance data are not available for
determining how to simplify tasks to coincide with personnel who
are more available".
In conclusion then: Some (or many?) limitations to ergonomic
design, but at the same time as many challenges to shift the scene
to a more adequate understanding of man-machine interaction.
co
o

Table 6. Operational Degradataion with Chemical Warfare Protective Clothing

FIRE POWER COMMUNICATIONS MOBILITY

MOP * I** II III I II III I II III

INFANTRY Rounds per hit Messages unanswered Road march time


4 rv 10 rv 12% t
M-14 5.4 5.9 8.6
(%) 9+*** 59+ Time to install wires March rate
15 ' 25' 26 min 4%
< + 10% +
M-60 10 5.7 10.7
(%) 43 + 7+ Getting specific man Assault time
1-2 ' rv 4+ min <10% + >100% +
M-79 5.3 3.0 8.4
(%) 43 + 58+ Voice versus hand signals
rv 50% +
ARMOR: N.B. - MOP III also required closed hatch

Rounds per hit Target identification Road march time 9% +


# missed
M-73 NSD 1 7 25 March rate 13% +
13% 28% 42% ~
M-79 6% t 104% + Attack time 20% + s::
~~ of transmissions 0
:tJ
105 0.3r+ 0.6r + 19+ 28+ Attack difficulty 3x »
»
r
fg
ARTILLERY ~
m
r-
Time from receipt at FDC to battery order In firing sections Last unit ~
Area adjust 27% t 137% t responses across SP 192% t
26 OK 11%+ 24%+ ~
Registration 100% t 75% t Clear hasty oz
position 45% t en
Accuracy a
Transfer 33% t 94% t > 95% rv 45% Enter-to all m
:D
ready 31% t Gl
o
Z
o
~
*MOP = Means of Protection n
**1 = normal combat clothing; II = open NBC clothing; III = closed NBC clothing with mask, gloves, cm
etc. en
*** t = increase; = decrease.
+ C5
z

CD
~
92 J. MORAAL

REFERENCES

Chapanis, A., 1976, Engineering Psychology, in: "Handbook of In-


dustrial and Organisational Psychology," Dunnette, ed., Rand
McNally, New York (N. Y.).
Chapanis, A., 1979, Quo Vadis, Ergonomia, Ergonomics, 22:595.
De Greene, K. B., 1980, Major Conceptual Problems in the Systems
Management of Human Factors/Ergonomics Research, Ergonomics,
23:3.
Edwards, M., 1981, The Design of an Accident Investigation Procedure,
Applied Ergonomics, 12:111.
Gazis, D., Herman, R., and Aradudin, A., 1960, The Problem of the
Amber Signal Light in Traffic Flow, Operations Research,
8:112.
Haslegrave, C. M., 1980, Anthropometric Profile of the British Car
Driver, Ergonomics, 23:437.
Hutchinson, T.P., and Haslegrave, C. M., 1980, Determination of Pat-
terns of Human Body Measurements by Use of Partial Correla-
tions, Ergonomics, 23:475.
Knight, J. L., Geddes, L. A., and Salvendy, G., 1980, Continuous, un-
obtrusive, performance and physiological monitoring of in-
dustrial workers, Ergonomics, 23:501.
Olson, P. L., and Rothery, R. W., 1972, Deceleration Levels and
Clearance Times Associated with the Amber Phase of Traffic
Signals, Traffic Engineering, 42:16, 42:63.
Swain, A. D., and Guttmann, H.E., 1980, "Handbook of Human Reliabi-
lity Analysis with Emphasis on Nuclear Power Plant Applica-
tions," Sandia Laboratories, Albuquerque (New Mexico).
Wechsler, D., 1952, "Range of Human Capacities," Williams & Wilkins,
Baltimore.
Zegeer, C. V., and Deen, R. C., 1978, Green-Extension Systems at
High-Speed Intersections, lTE Journal, 11:19.
ANTHROPOMETRIC REFERENCE SYSTEMS

Hans W. Jurgens

Anthropologisches Institut, Neue Universitat

Kiel/FRG

Research into the morphological and biomechanical variability


of a population requires an adequate data base. Although there are
no serious technical or theoretical problems connected with estab-
lishing such a data base, the data available for most western na-
tions are sparse and unsatisfactory; the male and female working
population is particularly poorly represented. The Federal Republic
of Germany is an exception, as data are available for a sample of
over 40.000 people, that may be considered to be representative of
the entire population with respect to regional and social factors.
The sample was taken and evaluated according to standardized methods
and formed the basis of DIN 33 402. This research was carried out by
or in close collaboration with the Anthropological Institute of the
University of Kiel.

The availability of comparable anthropometric data for the


whole nation permits investigation into various ergonometrically
important parameters and into combinations of such parameters. We
shall refer below to the evaluation of this data base.

THE ERGONOMIST'S TARGET GROUP

In Applied Anthropometrics and in Ergonomics in general it has


become accepted that the environment should be adapted to suit the
needs of persons from the 5th to the 95th percentile. Smaller and
larger people can only be catered for in special cases. This limita-
tion has proved to be appropriate from the technical and the econo-
mic point of view.

93
94 H. W. JORGENS

This seems quite clear at first glance, but Table 1 shows the
very considerable variation in values for three particular body mea-
surements, namely height, ankle-knee length and buttocks width, if
the percentile groups are differentiated for example according to
age and sex. The range of variation that then becomes apparent may
be as much as 41 cm in height or 13.8 cm in buttocks width. Trying
to take just this variation into account leads to technical diffi-
culties; it is probably not possible for a young man of the p5 group
for buttocks width to rest his arms on the arm rests of a chair that
suits the requirements of an elderly lady of the p95 group for the
same parameter. Similar examples will be found for other measure-
ments.

Table 1. Variation of body measurements according to percentile,

constant p50 male sex race, regional


male sex race,regional social grouping
race, regional social grouping
social grouping

variable age age age


p5-p95 p5-p95
sex

stature (60y.) 168, 1cm (p 5,60y.)158,lcm (9,P 5,60y.) 148,3cm


(19y. ) 177, 1 cm (p95,19y.) 189,3cm (d,p95,19y.)189,3cm
9,Ocm 31,2cm 41,Ocm

lower leg (60y. ) 42,lcm (p 5 ,60y. ) 39,Ocm (9, p 5,60y.) 34,6cm


length (19y. ) 45,5cm (p95,19y.) 49,7cm (d,p95,19y.) 49,7cm
3,4cm 10,7cm 15,lcm

seat (19y. ) 35,5cm (p 5, 19y. ) 32,2cm (d, P ,5, 19y. ) 32,2cm


breadth (60y. ) 36,7cm (p95,60y. ) 39,8cm (9, p95,60y.) 46,Ocm
1,2cm 7,6cm 13,8cm

HOW MANY PEOPLE TO A PERCENTILE?

The considerable variation already apparent between persons of


the 5th and 95th percentile has led in many cases, particularly for
military reqUirements, to reduction of the variation range by con-
structing tanks and submarines for a much more limited portion of
the population.
ANTHROPOMETRIC REFERENCE SYSTEMS 95

On the other hand we are faced with continuous demands for


equality, meaning that people outside the 5 to 95 percentile limit
should also be taken into consideration in design and construction
of everyday articles and in the organization of offices and facto-
ries.

A central question here, with regard to both practical/techni-


cal and economic factors, is how many people are there to a percen-
tile?

A fairly direct answer to this question is possible by first


establishing how many adults there are in a given population. These
figures are available from annual statistic records and are given in
Table 2 for various populations:

Table 2. Percentages of adults in some populations

Japan 67 %
Fed.Rep.Ger. 65 %
U.K. 63 %
China 63 %
Eire 58 %
USA 57 %
Turkey 56 %
India 56 %

This figure shows that in various countries allover the world


there are about 60 % adults, except in developing countries, where a
high number of children reduces the total percentage of adults. It
may generally be assumend that the adults for which these statistics
are available are roughly half male and half female. With the aid of
population records, that are available for every country of the
world, at an average of 60 % adults or 30 % each men and women, the
percentiles may now be calculated for a given population.

This calculation gives a figure of 180.000 as one percentile of


the adult males in the Federal Republic of Germany, while the cor-
responding figure for the USA is 680.000.

WHAT PRICE SAFETY?

It has become common practise to adapt every day articles and


offices and factories to the needs of the 5th to 95th percentile,
but this limitation is unsatisfactory when questions of safety pre-
cautions are to be considered: if the limit is 5th to 95th percen-
tile, then 10 % of the population are not covered by safety mea-
sures. This has led to adaption of the 1st to 99th percentile limits
in this case, leaving only 2 % of the population unprotected.- Ex-
tending the limits still further is frequently not possible because
96 H. W. JORGENS

the variations at the two ends of the scale are so extreme that ac-
counting for these minorities would frequently lead to complications
or even dangerous situations for the large majority otherwise c~ter­
ed for. This means the jobs must in some cases exclude particularly
small or particularly tall people.

To give an idea of the metrical changes caused by extending the


limits from 5th - 95th to 1st - 99th percentile, Table 3 shows the
data for men for one sample measurement, functional upward reach
with both arms. The increase in the total range is 18,7 em, enough
to cause considerable technical problems and additional costs.

Table 3. Differences of safety ranges depending on the percentiles


considered upward reach, both arms, males

p 5 194.8 cm p 95 224.8 cm
p 1 183.7 cm p 99 232.4 cm
Difference 11.1 cm 7.6 cm
Differences combined 18.7 cm

CALCULATED AND REAL PERCENTILE DATA

Both in theory and for practical purposes percentiles are


usually calculated according to the normal distribution curve, be-
cause the available data are often not sufficient to calculate the
percentiles directly from the anthropometric data. Table 4 gives a
comparison of the real percentiles established from the primary data
for height, buttock circumference and weight with the percentiles
calculated on the basis of normal distribution.

It is evident from the comparisons that there are few dif-


ferences between calculated and real percentile values for the
parameter of height - or for any length parameter. This is because
body length measurements are virtually represented by a curve of
normal distributions. The measurements corresponding to weight, on
the other hand, do not correlate, and Table 4 clearly illustrates
that differences may occur here when the measurements are calculated
on the basis of normal distribution.

HEIGHT AS A POINT OF REFERENCE

It has become accepted for practical purposes to use height as


a general point of reference. This is partly because height is a
very obvious personal characteristic and therefore also socially
relevant, partly because it is a parameter that can easily be mea-
sured.
ANTHROPOMETRIC REFERENCE SYSTEMS 97

Table 4. Differences between real and calculated percentile data


(males, age 19)

stature real calculated


p 5 165.2 cm 165.1 cm
p 95 186.7 cm 186.5 cm
p 1 160.9 cm 160.7 cm
p 99 191.2 cm 190.9 cm

Hip circumference real calculated


p 5 85.4 cm 84.4 cm
p 95 103.6 cm 102.4 cm
p 1 82.5 cm 80.6 cm
p 99 108.1 cm 106.2 cm

Weight real calculated


p 5 56.0 kg 54.2 kg
P 95 85.0 kg 86.5 kg

Whatever the reasons, height is the measurement most commonly


recorded and most people know how tall they are - although the mea-
surements given on questioning very often differ from real values.
Because height is also a socially relevant parameter, shorter people
tend to quote greater values and very tall people, especially women,
tend to state lower values when asked about their height.

For practical ergonomic questions height is of relatively


little importance, e.g. for beds, stretchers and some architectural
questions. Other measurements, for example the height of a person
when seated, are very rarely measured although they are of much
greater practical consequence.

The question arises as to whether this preference for height


measurements is justified. The answer must be seen in the light of
three factors:

- If any attempt is to be maGS to reconstruct other body measure-


ments on the basis of height, it should be recalled that children
are differently proportioned than adults and that there are also
proportional differences between men and women, between consti-
tutional and racial types, so that an attempt to calculate or
estimate a particular measurement such as leg length or knee-ankle
length on the basis of height can lead to considerable errors.

- It is generally possible to calculate body length measurements on


the basis of height within a biologically homogeneous group of
people' (adults of one sex from a uniform population) :
98 H. W. JORGENS

The correlation between total height and sitting height is r =


0.7, that between total height and eye height, sitting, is r =
0.8. On the other hand there is usually poor correlation between
total height and width or circumference measurements: The cor-
relation between total height and hip breadth, sitting, is r
0.37, that between total height and weight is r = 0.45. This al-
lows us to conclude that total height may only be used as an in-
dicator for other length measurements.

- The main disadvantage in using height as an indicator and point


of reference lies in the lack of uniformity of the individual
length parameters making up total height. Some factors are very
stable, such as cranium length or length of thigh or lower leg,
while other factors are very susceptible to change, such as plan-
tar arch, spinal curvature and hip inclination. Age can influence
height owing to the functional change in structure of the neck of
the femur, which becomes increasingly right-angled with advancing
age.

The different components making up total height show variations


due to accupation, age, lessening turgor over the period of the day
(especially in the invertebral discs) and also differ in rate of
growth: a 20-year-old can experience both growth and shrinkage in
the following years of his life if fallen arches lead to height re-
duction in the lower part of the body, while continued growth of the
spine increases his upper body height. Repeated measurement of the
total body height may show no change, while the sitting height will
have increased and the leg length decreased. Chance factors and ex-
ternal influence may further distort the picture. All these matters
imply that if it is really necessary to have a general indicator
and point of reference for measurement correlation, then this should
be the sitting height and not the total height, which has been shown
to be so liable to give a wrong impression. It is in any case prefer-
able to choose an indicator measurement that is relevant for the
question at issqe.

SPECIAL PROBLEMS

The anthropometric principles outlined above apply generally


to measurements for an entire population and should be supplemented
by discussion of two special problems when measuring clearly defined
subsections of the population, namely the physically handicapped and
people employed in a particular kind of job.

- The physically handicapped

As a general rule applicable to private life, to offices and fac-


tories, right down to safety precautions it has become commonly
accepted, as described above, to adapt the limitations set by the
ANTHROPOMETRIC REFERENCE SYSTEMS 99

requirements of people of the 5th to 95th percentile for every


day matters and of the 1st to 99th percentile for questions of
safety. The decision to retain these limits is based on practical
and economic arguments and is basically common sense. People
outside these limits are not adequately catered for by standard
products, normal occupational conditions or by the most usual
environmental characteristics. They know this is so, they have to
come to terms with this, and it is the duty of our society to
help them as far as possible. However, society can not guarantee
totally equal opportunities and possibilities for all of them
unless considerable disadvantages are to be accepted by the large
majority of the population.

This common-sense attitude breaks down in the face of physically


handicapped people. The same arguments as those used above would
justify catering for handicapped people of the 5th to 95th handi-
capped percentile. This would enable the not unlimited resources
to be utilized economically and to the best advantage to help as
many as possible.

Time and again it can be seen that emotional arguments outweigh


rational ones and thatrepeated attempts are made to cater equal-
ly for a I I the handicapped, which in practise often results
in catering for extreme cases while improvements that would be
technically and financially possible and that would benefit the
majority of handicapped people just are not carried through.

- Limiting the anthropometric range

When particular technical, physical factors make it seem sensible


and practical it is always worth considering whether there is any
point in suiting the environmental conditions of a place of work
to the needs of the usual limitation of the 5th to 95th percen-
tile, or whether it is not better to choose a given subsection of
the total population. Just as it is advantageous for baseball
players to be taIlor for bobsleigh riders to be heavy, it can be
appropriate for the people who work in planes, submarines or
tanks, for example, to be small. These considerations do playa
part in designing and planning tanks and submarines and examples
are available to demonstrate that calculations are based on maxi-
mal requirements of the 70th height percentile, and that sometimes
the upper limit is set as low as the 30th percentile.

This type of consideration may seem technically very appropriate,


but it does require careful calculation of the limitations imposed
on the potential users of the material in question.

To take an example from the population of the Federal Republic of


Germany once again, let us consider the figures available for
military personnel born in 1965. If we assume that limiting tank
100 H. W. JORGENS

troops to those not above the 30th height percentile leaves enough
men for tanks, we must recall that the number available will be
halved in the course of the next ten years. Therefore the question
must then be reconsidered, whether the 30th percentile for height
must still be regarded as the upper limit. Added to this, the
secular acceleration trend is still leading to a constant, if
small,increase in height, so that it is not possible to limit
indefinitely the space available for the person carrying out a
given job. It must be generally accepted that the ergonomist
cannot only think in terms of technical ergonometric factors, but
that if he is basing his calculations on the requirements of a
population of humans, he must always relate his findings to the
population in question.

Another point that must also be taken into consideration here is


the fact that we cannot only set the percentile limit for factors
of size; in view of the constantly increasing technical require-
ments placed on people in civil and military occupations, the
percentile limits for intellectual and professional requirements
must also be raised, thus reducing the number of candidates eli-
gible by reason of size for a given job. This means that the
values chosen to ensure an adequate subsection of the population
for the task in question must not only select the physically small
people but also those who are intellectually sufficiently well
qualified; in addition to this, other personal factors will playa
role in deciding a person's suitability for the job to be fulfill-
ed. The correct basis for ergonometric calculations cannot be con-
sidered to have been established until all these factors have been
taken into account to determine the real extent of the population.

REFERENCES

Buchi, E. C., 1950, Anderung der Korperform beim erwachsenen Men-


schen, Anthrop. Forschungen, Heft 1, Wien.
Hertzberg,· H. T. E., 1963, "Anthropometric survey of Turkey, Greece,
and Italy," Pergamon, Oxford.
Jurgens, H. W., 1975, Korperformen und Korperhaltungen; Beurteilung
von MeBwertdifferenzen; Die Kombination von KorpermaBen; Bio-
metrische Voraussetzungen, in: "Handbuch der Ergonomie,"
Bundesamt fur Wehrtechnik und Beschaffung, ed., Hanser,
Miinchen.
Jurgens, H. W., Habicht-Benthin, D., and Lengsfeld, W., 1973, Kor-
permaBe 25- bis 40-jahriger Manner zur Prufung der anthro-
pometrisch-ergonomischen Bedeutung altersbedingter Veranderun-
gen der Korperform, Forschungsbericht aus der Wehrmedizin,
Bonn.
Jurgens, H. W., Helbig, K., and Lengsfeld, W., 1975, Korperform und
Eignung, Forschungsbericht aus der Wehrmedizin, Bonn.
Lewin, T., and Jurgens, H. W., 1969, tiber die Vergleichbarkeit von
anthropometrischen Daten, Z.f. Morphologie und Anthropologie,
61:33.
ENGINEERING ANTHROPOMETRY

Karl H. E. Kroemer

Virginia Polytechnic Institute and State University

Blacksburg, Virginia/USA

This paper* attempts to summarize and abstract the current sta-


tus of engineering anthropometry. It indicates sources of anthropo-
metric data needed by the engineer. It discusses the practical ap-
plication of anthropometric information to the modelling and design
of manned systems for optimal fit to the human, for highest safety,
and best performance.

PREFACE

Assessment of the physical dimensions of the human body and


application of this knowledge to the design of tools, equipment, and
work are certainly among the oldest arts and sciences. It would be
an easy task if all anthropometric dimensions, of all people, would
follow a general rule. Thus, philosophers and artists embedded their
ideas about the most aesthetic proportions into ideal schemes of
perfect proportions. "Golden sections" were developed in ancient
India, China, Egypt, and Greece, and more recently by Leonardo Da
Vinci, or Albrecht Durer •. However, such canons are fictive since
actual human dimensions and proportions vary greatly among indivi-
duals. Hippocrates (about 460 - 377 BC) taught that there are four
temperaments (actually, body fluids) represented by four body types.
The psychiatrist Ernst Kretschmer (1888 - 1964) proposed that three
typical somatotypes (pyknic, athletic, asthenic) could reflect human
character traits. In the 1940's, W. H. Sheldon and his co-workers
devised a system of three body physiques (endo-, meso-, ectomor-
phic). The classification was originally intuitive, and has only
more recently been developed to include actual measurements.

*Based on a presentation at the AIlE 1981 Spring Annual Conference

101
102 K. H. E. KROEMER

Today's engineers assess human dimensions and physical capabi-


lities in physical units, and use these to design hand tools, work
stations, equipment, and work task to fit human dimensions and
capabilities. Thus, "Engineering Anthropometry" is the application
of scientific physical measurement methods to human subjects for the
development of enqineering design requirements (Roebuck, Kroemer,
and Thomson, 1975). Therefore, engineering anthropometry is one of
the backbones of "Ergonomics", or "Human Factors", which study human
characteristics for the appropriate design (\f the living and working
environment.

A separate branch of applied anthropometry developed in the


late 1900's. Body structures, kinetics and kinematics of the human
body, the mechanics of the musculo-skeletal system, etc. were speci-
fically investigated. This field of scientific endeavor, now called
"biomechanics", was defined (by Roebuck, Kroemer, and Thomson, 1975)
as the interdisciplinary science (comprising mainly anthropometry,
mechanics, physiology, and engineering) studying mechanical struc-
ture and behavior of biological materials.

The relationships between physical anthropometry and biomecha-


nics are so close that it is difficult and probably useless to draw
demarcation lines between them. The knowledge about the physical
characteristics of the body is obviously basic to each, and design-
ing the man-made environment from tools to tasks such as to suit
human dimensions and to meet human capabilities is the common de-
sired result.

SECULAR CHANGES IN BODY SIZE

Looking at medieval body armor one cannot help but notice that
today's males would have a hard time fitting into these small
shells. Such secular increase in body size is also apparant from the
everyday experience of children being larger than their parents.
While evidence for long-term gains in body size is only indirect,
statistical comparison of anthropometric data taken by the military
services on large samples with consistent techniques proves in-
creases in stature in the neighborhood of one centimeter per decade
throughout much of this century. In the U.S.A., a gain in body weight
of about 2 kg per decade has become apparent during the second half
of this century.

Regarding the reasons for such secular changes in anthropo-


metry, and the expected future development, one may assume that here-
ditary capabilities for the achievement of one's optimal body dimen-
sions had often not been fully utilized due to less than optimal
living conditions. Recent improvements in hygiene and nutrition faci-
litate reaching inherited growth potentials. Thus, longitudinal body
dimensions of families that were well-off over generations no longer
ENGINEERING ANTHROPOMETRY 103

have significant increases, while those families still improving


their socio-economic status show gains in stature. With favorable
living conditions for most the 'increase should diminish, and body
height would reach a new statistical value asymptotically (Stoudt,
1978) •

For most engineering applications, the relatively small changes


in body dimensions are of little consequence - but should be con~
side red for the design of manned systems that are meant to be used
well into the future.

SURVEY SAMPLING

The selection of the subject sample to be measured is highly


critical because it determines the validity of the data, and the
cost of the procedure. Performing a survey of every citizen of a
nation is time-consuming but yields an exact anthropometric picture
of that population. Counterexamples of highly selected samples would
be visitors to automobile exhibitions (excluding those not interest-
ed in car shows, or living in remote areas), or telephone owners. On
the other hand, if one is interested in, say, the weights of shop-
pers at supermarkets, exactly such persons should be sampled.

The most comprehensive and most expensive technique in samp-


ling is to measure all subjects, such as routinely done in military
entrance exams. Thus, the whole population of concern is measured.
If the population group is not "captive", random selection from a
master file, such as the census, is in order. While likely to yield
a highly accurate picture of the population, this is certainly a
very expensive method. However, sampling techniques stratified by
age, sex, social status, region, etc., are feasible and have been
proposed for nationwide surveys (Churchill and McConville, 1976). If
the selection criteria are well defined, quaSi-quota matching can be
used. Many combinations of these techniques are feasible, such as
measuring a rather large sample in selected key dimensions, and
taking additional measurements on every third or tenth subject.

The more specialized the sample and the dimensions taken, the
more this method resembles "subgrouping", such as applied to pre-
selected and pre-measured individuals usually representing the ex-
treme end of the population distributions, e.g., the tall-heavy,
or short-light. Such subgroups are often used by engineers to check
their designs for fit to the extremes of the user population, as-
suming that the mid-ranges will be accommodated if the design fits
the unusual persons. The ultimate is, of course, measuring indivi-
duals in order to fit equipment exactly to the person, such as in
tailoring space suits for selected astronauts.
104 K. H. E. KROEMER

MEASUREMENTS NEEDED

The first question is: Who needs anthropometric data for what
purpose? A manufacturer of loosely-fitting garments is interested in
different anthropometric data than a physical anthropologist. The
manufacturer of face masks needs very different information than an
automobile designer. To determine the size and location of safe
openings in equipment so that the operator cannot be caught in them
requires information other than needed to determine leg room, and
chair adjustments, for workstations used by seated operators.

The needs are closely related to the question of the necessary


exactness of the data. How much uncertainty can the anthropologist,
manufacturer of face masks, or automotive engineer tolerate? This
user requirement determines how much inaccuracy is acceptable in the
prediction of, say, civilian data from military information, or in
the exactness with which the initial measurer must take the data.

A problem probably more critical than the exactness of data


(which seems to be satisfactory for most application purposes) is
how to translate the traditional static anthropometric data into
dynamic information, which would depict reliably the functional di-
mensions of the human body in motion. Body dimensions are tradi-
tionally taken on subjects assuming a rigid, standardized posture in
which the body segments are at 0 and 90 degrees to each other. These
are not the conditions in which work is performed, and for which
equipment must be designed. At this moment, there is still no theo-
retically sound procedure or acceptable "recipe" available that
allows reliable translation of static standard data into functional
measurements.

The number of measurements to be taken on the subject depends on


thethree aspects mentioned, i.e., who needs the data, how exact
must the data be, and how do they translate from the static to the
dynamic condition. Accordingly, and depending on other specific con-
straints, dimensions actually taken range from very few (twelve in
the 1967 U.S. HANES survey) to very many (189 taken in 1967 USAF
Survey). The ISO (International Organization for Standardization),
through its Technical Committee 159, Ergonomics, Sub-committee 3,
will provide a key list of measurements to be taken in the near
future (ISO Draft No. 7250).

There seems to be a general concensus that, even with the most


traditional computational methods, 20 to 40 body dimensions at most
taken per subject in a survey should suffice to predict all but the
most spec-ial other dimensions. A recent analysis has shown that one
might need just two or three dimensions to predict practically any
others in which an engineer is interested. The two dimensions mostly
needed are height (stature) and weight which combined yield throuqh
regression equations or matching, accurate p~edictions of other
body dimensions (McConville, Robinette and Churchill, 1981;
ENGINEERING ANTHROPOMETRY 105

Robinette and Churchill, 1979; Robinette and McConville, 1981).


Table 1 lists such body dimensions calculated for the current ci-
vilian u.s. population.

To carry it to the extreme: since it has been shown that just


asking subjects for their height and weight yields rather exact
figures (with predictable overstatements for height and understate-
ments for weight) one might get away with that extremely simple
measuring technique in the future, relying on predictive equations
for the remainder of the data.

MEASUREMENT TECHNIQUES

Traditional standard anthropometry relies on a small set of in-


struments: anthropometer, several types of calipers, measuring tape,
grip strength dynometer, and a weight scale. These are effective
measurement tools if used by a skilled anthropometrist, but (with
exception of the scale) represent obsolete technology with respect
to acquisition and recording of the data. Many attempts have been
made to improve these standard instruments, such as adding potentio-
meters to anthropometer, caliper and tape and to use their electric
outputs as direct inputs to the computer recording the data.

Photography does not need physical contact, but can relate only
to the surface contours of the body which are, with few exceptions,
determined by compressible tissue and give little information about
the underlying bony landmarks preferred by the anthropologist and by
the engineer. Standard photography, with its techniques derived from
cartography, provides 3-D pictures of the surface contours of the
human body. To overcome the difficulty of relating surface features
to underlying bony structures, the use of "point marks" has been
proposed (Reynolds, 1982) which would allow locating limbs (re-
presented by the point marks) with respect to each other in spatial
coordinates" This technique has yet to be applied to living subjects
in general surveys.

In fact, the relation of surface measurements to the underlying


skeletal structures of the body is of major concern to the anthropo-
metrist, biomechanicist, and engineer. Usually', the human body is
modelled as a basic system of rigid links conncected by articulations
with known degrees of freedom, covered by masses of given distri-
bution and contours, powered by muscles acting around the joints,
etc. -- very much in the tradition of Borelli's model "De motu ani-
malium" of anno 1680. Such models require knowledge of the re-
lationships between the surface measurements, as traditionally pro-
vided by anthropometry, and the body structures representing links
and artiCUlations. At this time it appears questionable if current
(often deterministic and overly simplistic) biomechanical models can
represent the characteristics of the human body sufficiently, and if
o

Table 1. u.s. Civilian Body Dimensions, Female/Male in cm or kg, for Ages 20 to 60 Years*

Percentiles
5th 50th 95th Std. Dev.

Stature (Height) 149.5/161.8 160.5 /173.6 171.3 /184.4 6.6 / 6.9


Eye Height 138.3/151.1 148.9 /162.4 159.3 /172.7 6.4 / 6.6**
Shoulder (acromion) Height 121.1/132.3 131.1 /142.8 141.9 /152.4 6.3 / 6.1**
Elbow Height 93.6/100.0 101.2 /109.9 108.8 /119.0 4.6 / 5.8**
Knuckle Hei2ht 64.3/ 69.8 70.2 / 75.4 75.9 / 80.4 3.5 / 3.2**

Height, sitting 78.6/ 84.2 85.0 / 90.6 90.7/96.7 3.5/3.7


Eye Height, sitting 67.5/ 72.6 73.3 / 78.6 78.5 / 84.4 3.3 / 3.6**
Shoulder Height, sitting 49.2/ 52.7 55.7/59.4 61. 7 / 65.8 3.8 / 4.0**
Elbow Rest Height, sitting 18.l/ 19.0 23.3 / 24.3 28.1 / 29.4 2.9 / 3.0
Knee Height, sitting 45.2/ 49.3 49.8 / 54.3 54.4 / 59.3 2.7 / 2.9
Popliteal Height, sitting 35.5/ 39.2 39.8 / 44.2 44.3 / 48.8 2.6 / 2.8
Thigh Clearance Height,
sittin2 10.6/ 11.4 13.7/14.4 17.5/17.7 1.8/1.7

Head Breadth 13.6/ 14.4 14.5 / 15.4 15.5 / 16.4 .57/ .59
Head Circumference 52.2/ 53.8 54.9 / 56.8 57.7/59.3 1.63/1.68 ?"
::t
Interpupillary Distance 5.1/ 5.5 5.83/ 6.20 6.5 / 6.8 .44/ .39
!T1
~
;:g
0
m
3:
m
;:g
m
Z
G)
Z
m
Percentiles m
:lJ
5th 50th 95th Std. Dev. Z
G)
~
Forward Reach, functional 64.0/ 76.3 71.0 / 82.5 79.0 / 88.3 4.5 / 3.6** Z
-I
Elbow-Fingertip Length 38.5/ 44.1 42.1 / 47.9 46.0 / 51. 4 2.2 / 2.2** J:
Hand Length 16.4/ 17.6 17.95/ 19.05 :lJ
19.8 / 20.6 1.04/ .93 0
Hand Breadth, Metacarpale 7.0/ 8.2 7.66/ 8.88 8.4 / 9.8 .41/ .47 "'0
0
Hand Circumference, 3:
Metacarpale 16.9/ 19.9 18.36/ 21.55 19.9 / 23.5 .80/ 1.09 m
-I
:lJ
-<
Chest Depth 21.4/ 21.4 24.2 / 24.2 29.7/27.6 2.5 / 1.9**
Elbow-to-Elbow Breadth 31.5/ 35.0 38.4 / 41. 7 49.1 / 50.6 5.4 / 4.6
Hip Breadth, sitting 31.2/ 30.8 36.4 / 35.4 43.7/40.6 3.7/2.8

Buttock-Knee Length, sitting 51.8/ 54.0 56.9 / 59.4 62.5 / 64.2 3.1 / 3.0
Foot Length 22.3/ 24.8 24.1 / 26.9 26.2 / 29.0 1.19/ 1.28
Foot Breadth 8.1/ 9.0 8.84/ 9.79 9.7/10.7 .50/ .53

Weight (in kg) 46.2/ 56.2 61.1 / 74.0 89.9 / 97.1 13.8 /12.6

*Courtesy of Dr. J. T. McConville, Anthropology Research Project, Yellow Springs, OH 45387,


and Dr. K.W. Kennedy, USAF-AMRL-HEG, WPAFB, OH 45433
** Std. Dev. estimated by Kroemer

...o
'-I
108 K.H.E.KROEMER

classical anthropometry can provide the information needed for


state-of-the-art modeling.

Borelli's 300-year-old model of the human body incorporated


muscles moving the body segments. Generally, a muscle connects
two body segments rotatable about a common joint. If the muscle is
under tension, it applies torque to each segment. The magnitude of
torque depends on the amount of internal muscle force, on the di-
stance between the joint and the location of the muscle or tendon
attachment to the bone, that is, the lever arm, and on the pull
angle between the vector of muscle force and the limbs. Usually,
neither pull angle nor lever arm are known, and for practical pur-
poses,the muscle is inaccessible. Therefore, internally developed
muscle force usually cannot easily be measured directly.

However, for engineering purposes the force applied to an out-


side object is of primary interest. Since this force depends on the
inherent strength of the muscles as well as on the prevailing mecha-
nical advantages, the location of the force-measuring device must be
specified with respect to the body, usually to the next joint, to
make the measurement meaningful.

Human muscular strength is, for engineering purposes, expressed


as the linear, or translational, force exerted at the interface with
the measuring device in terms of a vector, having magnitude and di-
rection. Torque measurements (force multiplied by the prevailing
lever arm) or pressure measurements (force evenly distributed over a
known surface area) can be conceived as modified assessments of
linear force.

By definition, strength data refer to maximal efforts only. The


relevance of such maximal data to submaximal, optimal, reasonable or
acceptable conditions is discussed in some detail by Roebuck,
Kroemer, and Thompson (1975).

Measurement of grip strength has often been part of anthropo-


metric surveys. Usually, a mechanical device was used that gave
somewhat (amount depending on the special design features) under
force application. The deflection of the device was calibrated and
used as a measure of strength exhibited by the subject. For muscle
strength assessment other than grip strength, a large number of mea-
surement tools have been used (Roebuck, Kroemer, and Thompson,
1975). Non-recording devices, usually pointer instruments, are suit-
able for quick checks, but do not satisfy the need for yielding a
true record of the strength exerted, as controlled strength tests
require (Kroemer and Marras, 1981). A very convenient means to
attain an analog record of the strength exertion is via strain-gage
technology.
ENGINEERING ANTHROPOMETRY 109

Many persons and institutions are, at'this time, working to


develop better measuring techniques to assess human body characteri-
stics, and to better understand human body mechanics so that one can
describe realistically body dimensions, body mechanics, and func-
tional capabilities needed for work place design. Most of these re-
searchers presented their results at a special symposium, held in
1980 in England. This meeting discussed the topics of anthropometry
and biomechanics, including human voluntary muscle strength capabi-
lities. The proceedings of this symposium provide detailed infor-
mation (Easterby, Kroemer, and Chaffin, 1982).

MODELS AS "DATA MASSAGE" SYSTEMS

Before anthropometric measurements can be used as input to


data storage and retrieval systems, they must be checked for errors.
Errors used to be quite frequent with traditional anthropometric
surveys, in which data were written down manually and then tran-
scribed into computer-readable information. However, even newer data
recording techniques suffer from a great number of possible errors
in information input and in computational routines. In fact, since
often no human is involved who could apply judgement, grossly wrong
data can be displayed by the computer. TWo editing routines, de-
signed to eliminate errors and outlyers, have proved to be highly
successful. The XVAL method sorts out extreme values (ten highest
and ten lowest) for each data category. EDIT tests each selected
actual data point against a predicted value based on other inputs
contained in the sample (Kikta and Churchill, 1978). Numerous other
checks can be performed, among them visual presentation of indivi-
dual data points, or of frequency distributions, as graphic dis-
plays, e.g., on the cathode ray tube (NASA, 1978).

One purpose of the model utilizing computerized calculating


routines is to present data as predicted values for defined popula-
tion samples. The simplest model is that of the "average person" re-
lying solely on mean values, accumulated over several data cate-
gories. This mythical person, having only average body dimensions,
was proven to be nonexistent in the early fifties and pronounced
useless and dead repeatedly since -- but some engineers still be-
lieve in this ghostly figure.

Another concept assumes that a large female can be represented,


for most purposes of applied anthropometry, by the body dimensions
of an "average'" man; relatedly, that the dimensions of a small man
are similar to those of an "average" female. (Small females are not in-
cluded in this scheme.) This concept has been scrutinized recently
and found to be inappropriate and misleading in many anthropometric
aspects (Robinette, Churchill, and McConville, 1979).
110 K. H. E. KROEMER

A more sophisticated and appropriate means to represent select-


ed body dimensions utilizes the fact that given percentages values
can be calculated from the mean by adding or subtracting certain
multiples of the standard deviation. Clearly, the percentile ap-
proach is fully correct and very useful for one given dimension,
provided the variable is distributed normally. However, it is fall-
ible to add percentile values in order to derive, say, a 95th per-
centile phantom having 95th percent body dimensions throughout. For
statistical reasons, percentile values are not additive -- the "50th
percentile person" is the same ghost as the "average person".

Stature and weight are the best single predictors of all other
body dimensions. This model has been checked recently by Robinette
and Churchill (1979) and found to be fully valid. Using regression
equations which need as input variables body height and weight, equa-
tions have been developed that predict other body dimensions with
high accuracy (Robinette and McConville, 1981). Body dimensions cal-
culated through regression equations are additive.

A sizable number of computer models have been developed in


order to interface human body dimensions with the geometry of the
work station. Underlying all of them is the Borelli concept of rigid
body lengths connected by joints. Most models are distinguished by
the number of links and joints, degrees of freedom, and the methods
to drive the model, particularly to positions body parts in space.
Other major distinctions are the linking of the model to the work
space, and the manner in which the underlying anthropometric data
bases are incorporated in the model. At this time, most computer
models of the human body are deterministic, although some attempts
have been made to include probabilistic techniques.

Models such as BOEMAN, COMBIMAN, CAPE, CAR, or SAMMIE (see


Easterby, Kroemer, and Chaffin, 1982; Kroemer 1983) have been
rather successful as means to store, retrieve and massage anthropo-
metric data. However, major problem areas still need improvement.
These include: mode of operation, assumptions about underlying links
and joint centers, representation of voluntary versus enforced move-
ment, mass distribution, contours, time related strength and work
capacities, transition from the seated to the standing or walking
posture, representation of external constraints such as clothing or
obstacles in the path of movement, etc. The graphics to be employed,
the internal computation times needed, and the whole complex of in-
teraction with the user-designer need significant improvements. One
solution may be to use rather simple models for gross statements,
and to employ more complex models only when there is a need for more
detailed information. For simple design and evaluation tasks the use
of basic tools is still very appropriate, efficient, and successful.
For example, templates and manikins recently developed by Juergens
and Kennedy have proven to be of high value for design tasks.
ENGINEERING ANTHROPOMETRY 111

INTERFACING THE OPERATOR MODEL WITH THE EQUIP.MENT MODEL

As just discussed, a number of model approaches have been used


for checking existing work station designs, or for designing new
work places, to achieve optimal fit to the human operator. Among
others, major differences among these models lie in the interfacing
of the operator with the equipment, i.e., cockpit, cab, machine,
bench or table, chair, work objects, tools, controls, visual dis-
plays, etc. Designated interface points connect defined body seg-
ments of the operator with components of the work place or equip-
ment. In the past, the following interface points have been used:

The eye. Called eye design point, eye ellipse, line of v~s~on,
etc., in different industries, this point constitutes the geo-
metrical link between targets and the eyes of the operator, usually
assumed to be seated.
The hip. Depending on industry usage, either an approximation
of the hip joint is used for design purposes (H-point), or the in-
tersection of the seat pan and the backrest planes (Seat Reference
Point, SRP) is used to establish the link with the body of the
seated operator.
The foot. Called heel-rest point, or accelerator-heel point, or
package-origin point, this reference location connects one or both
feet of the sitting operator with the equipment.
The hand. Despite its importance, and because of the diffi-
culties in defining specialized activities, this interaction between
the human body and the equipment is only loosely described such as
reach envelope, as preferred manipulation area, or as work plane
projected onto horizontal bench or table surfaces.
While all of these interface points are important, the inter-
face between the hand and the object is usually critical. However,
it has not been used to the extent desirable because of the diffi-
culty of establishing a hand or finger "point". The body system con-
necting the fingers with the main part of the body is of· such vary-
ing dimensions, of such high variability in its positions, and in-
fluenced by so many external features (design of the work place, of
the object, of the task, etc.) that in the past attempts to link the
operator with the work station through the hands proved to be too
difficult.

SOURCES OF ANTHROPOMETRIC DATA

USAF Anthropometric Data Bank

It contains the results of more than 30 surveys, containing the


data of approximately 100,000 measured individuals. About 1/3 of the
112 K.H.E.KROEMER

surveys concerns u.s. military male and female populations, another


1/3 relates to the body dimensions measured in foreign military po-
pulations, while the remainder concerns u.s. civilian data (AMRL-
BEG, Wright-Patterson AFB, Ohio 45433).

u.s. ARMY

The u.s. Army has not only collected extensive data on its own
personnel, but also obtained the body dimensions of various foreign
populations. (DOD-USA-DRDNA-ES, Natick, Massachusetts 01760)

N~A

NASA published an anthropometric source book in three volumes


which not only contains the newest u.s. data but also discusses, in
considerable detail, sampling, measuring, data recording, and ana-
lysis and retrieval methods. Most of the u.s. Army data are included
in the Air Force data bank, which in turn provided the data input
for the NASA Anthropometric Source Book (1978). (N~A, Houston,
Texas 77058)

ERGODATA

ERGODATA is a new international data source for European anthro-


pometric data, primarily based on French anthropometric surveys but
now also including other national data, such as from Germany and the
U.K. The ERGODATA system is located at the Anthropology Laboratory,
Universite Paris (Prof. A. Coblentz), with the U.K. office at the
University of Birmingham (Prof. K. B. Haley). The system is being
expanded to the u.s. with a branch office located at SUNY in Buffalo,
N.Y. 14260 (Prof. C. G. Drury).

Several books on ergonomics, or human engineering, provide in-


formation on techniques to fit equipment to the human operator. Al-
ready classics, but still very useful, are Shack~I'sApplied Ergo-
nomic Handbook (3rd ed. 1976) and Ethnic Variables in Human Factors
Engineering, edited by Chapanis (1975). Van Cott and Kinkade's ~
Engineering Guide to Equipment Design (1972) consolidated earlier
pioneering works (e.g., by Murrell, McFarland, Woodson) with the
vast information gathered by the military. Though written for use in
North America, its contents also applies to most European people.
(Incidentally: the regularly updated military design handbooks, such
as MIL-STD-1472, or MIL-HDBK-759 in the U.S.A., provide excellent
though somewhat limited information and guidelines for the de-
signer.) The monumental (1047 pages) Human Factors Design Handbook
by Woodson (1981) is certainly a design thesaurus. Recent European
books include Human Factors in Transport Research (edited by Oborne
and Levis, 1980) and Grandjean's Fitting the Task to the Man (1980).
The journal Applied Ergonomics provides continuous commentaries on
many examples of ergonomic design problems and their solutions.
ENGINEERING ANTHROPOMETRY 113

The vast majority of anthropometric information on large popu-


lation groups stems from military surveys. The data were taken on
"captive" population groups, in an organized manner, using con-
sistent techniques. Thus, there are highly reliable data on many
persons at our disposition. However, how representative are these
data for civilian samples? A comparison indicated that U.S. Air
Force males were, on the average, 1.4 cm taller and 3.5 kg heavier
than American civilian males, while u.s. Air Force women were 2.1 cm
taller but 5.7 kg lighter than civilian females (Stoudt, 1978). But
such differences do not make military data invalid for civilian
applications. If key dimensions (such as stature and weight) of both
a civilian and military sample are known, one should be able to
assess body measurements not taken on the civilians but on the
soldiers through matching or regression techniques. Table 1 was de-
rived in that manner.

Unfortunately, data on special population groups, such as


adolescents, elderly, handicapped, etc. are still largely missing.
At this time, though, efforts are under way in several countries to
provide such data.

REFERENCES

Churchill, E., and McConville, J. T., 1976, "Sampling and Data


Gathering Strategies of Future USAF Anthropometry," Wright-
Patterson AFB, Aerospace Medical Research Laboratory (Ohio).
Easterby, R., Kroemer, K. H. E., and Chaffin, D. B. (eds.), 1982,
"Anthropometry and Biomechanics," Plenum, New York.
Kitka, P., and Churchill, T., 1978, Editing Procedure for Anthropo-
metric Survey Data, Wright-Patterson AFB, Aerospace Medical
Research Laboratory, AMRL-TR-78-38 (Ohio).
Kroemer, K. H. E., 1983, Engineering Anthropometry: Designing Work
Space and Equipment to Fit the User, in: "Psychology and
Productivity at Work: The Physical Environment," D. J. Oborne
and M. Gruneberg, eds., Wiley, Chichester, Surrey.
Kroemer, K. H. E., and Marras, W. S., 1981, Evaluation of Maximal
and Submaximal Static Muscle Exertions, Human Factors, 23:643.
McConville, J. T., Robinette, K. M., and Churchill, T., 1981, An
Anthropometric Data Base for Commercial Design Applications,
Anthropology Research Project, NSF Grant DAR-8009861, Yellow
Springs (Ohio).
NASA, 1978, "Anthropometric Source Book," NASA Reference Publication
1024, NITS, Springfield (Va.) (Order Number 79-11-734).
Reynolds, H. M., 1982, The Human Machine in Three Dimensions: Impli-
cations for Measurement and Analysis, in: "Anthropometry arid
Biomechanics," R. Easterby, K. H. E. Kraemer, and D. B.
Chaffin, eds., Plenum, New York.
114 K. H. E. KROEMER

Robinette, K. M., and McConville, J. T., 1981, An Alternative to


Percentile Models, Society of Automotive Engineers, SAE
Technical Paper 810217, Warrendale (Pa.).
Robinette, K., and Churchill, T., 1979, Design Criteria for Charac-
terizing Individuals in the Extreme Upper and Lower Body Size
Ranges, Wright-Patterson AFB, Aerospace Medical Research
Laboratory, AMRL-TR-79-33 (Ohio).
Robinette, K., Churchill, T., and McConville, J. T., 1979, A Compa-
rison of Male and Female Body Sizes and Proportions, Wright-
Patterson AFB, Aerospace Medical Research Laboratory, AMRL-
TR-79-69 (Ohio).
Roebuck, J. A., Kroemer, K. H. E., and Thomson, W. G., 1975,
"Engineering Anthropometry Methods," Wiley, New York.
Stoudt, H. W., 1978, Are People Still Getting Bigger - Who, Where,
and How Much?, Society of Automotive Engineers, SAE Technical
Paper 780280, Warrendale (Pa.).
BASIC DATA FOR THE DESIGN OF CONSOLES

Heinz-Peter RUhmann

Institut fur Ergonomie der Technischen Universitat

MUnchen/FRG

INTRODUCTION

Consoles, information boards and panels are elements especially


sUitabie for the installation of displays and controls. These ele-
ments are usually placed in central departments of a system (control
rooms and stations) and provide for the monitoring and controlling
of technical processes.

The perception of information may result immediately from the


direct view to the process (e.g. control station at a rolling line)
or indirectly from monitoring displays (scale indicators, indicator
lights, plotters, CRT displays, e.g. air traffic control center).
Commands to the process can be caused automatically (by a system for
automatic process control) or by manual operation of the controls
(switches, push buttons, levers, hand wheels, etc.).

At consoles and panels a single operator or a crew is able to


work as well in a standing as in a seated position.

Before discussing design criteria and requirements for the di-


mensioning of consoles, information boards and panels*, the funda-
mentals of anthropometric layout of workstations shall be represented.

*The anthropometric data and dimensions of consoles, information


boards and panels published in this paper correspond to a working
draft concerning the ergonomic layout of control rooms. This draft
has been worked out by the subcommittee AA4/8 UAl of the FNErg
(FachnormenausschuB Ergonomie, standards committee of ergonomics,
DIN).

115
116 H.-P. ROHMANN

FUNDAMENTALS OF THE ANTHROPOMETRIC LAY-OUT OF WORK PLACES

Basic data for the design of work places result from the fea-
turesof the human visual system (lines of sight, visual fields),
from the sitting positions mainly taken up at the work place (e.g.
rearward seating position for predOminantly monitoring a system)
and from the dimensions of the hand reach area.

Lines of sight

For the positioning of displays and controls at work stations


important reference lines are the lines of sight and the fields of
view. These are the direct field of view, the field of view wi thin
which the eyes can fixate objects while the head is kept in a fixed
position and the field of view within which the eyes fixate objects
if the head is moved.

Considering the lines of sight, it must be distinguished bet-


ween the horizontal line of sight, the line of sight related to the
head and the normal line of sight (Fig. 1).

Horizontal line of sight

If the head and the eyes are kept in a straight (tense) position
the line of sight is identical with the horizontal (=0°).

Horizontal line of sight related to the head

If the head is kept in a comfortable (relaxed) posture, imply-


ing that the forward inclination of the head axis to the vertical
lines is about 10°-15°, but the eyes are in a straight position,
then the horizontal line of sight related to the head is about
10°-15° below the horizontal line of sight.

Normal line of sight

If the head and the eyes are kept in a comfortable posture,


the angle between the eye axis and the head axis being about 105°-
110°, then the normal line of sight is 25°-35° below the horizontal
line.

Fields of view

The several fields of view are defined in the following manner.

Direct field of vie",

The direct field of vision contains all objects in the hori-


zontal and vertical plane which can be perceived simultaneously
BASIC DATA FOR THE DESIGN OF CONSOLES 117

Horizontal line of sight

Head and eyes intense


)~ _____ "O Hori zonta1 li ne
of sight
position

fVi/_\:::::::---oo
Horizontal line of sight 1--10°-15°
rel ated to the head

Head in a relaxed, eyes y)' ----10°-15 Horizontal line


in a tense position of sight rel ated
to the head

kI /~k- -- --
,~'05-"OO

i? ,",-::::-
Nonnal line of sight 0°
10°-15°
Head and eyes ina
rel axed position II '
'25°-35° Nonnalline
of sight

Fig. 1. Lines of sight

when the head and the eyes are reposing. One distinguishes between
the monocular and the binocular field of view.

Field of view - fixed head/fixating eyes

In this field of view there are all objects in the horizontal


and vertical plane which can be fixated successively when the head
is kept in a fixed position but the eyes move in order to fixate
objects. One has to distinguish between the monocular and binocular
field of view - fixed head/fixating eyes.

The horizontal and vertical angular range of a "comfortable"


zone for fixating is between ±18° and ±20o.

Field of view - moving head/fixating eyes

This field of view contains all objects in the horizontal and


vertical plane which can be fixated successively while the body is
kept in a fixed position but the head is turned and the eyes are
moved in order to fixate objects. One distinguishes between a mon-
ocular and a binocular field of view - moving head/fixating eyes.

If all direct fields of view around the fixating points as a


centre are drawn one above the other within the visual fields as
118 H.-P. ROHMANN

defined above one obtains the secondary and tertiary field of view
(monocular and binocular).

Only in a small area around the fixating point good visual


acuity exists. If there is a relative visual acuity of 1.0 at the
fixating point, the acuteness of vision diminishes to 0.5 at an
angular distance of 2.5 degrees from the fixating point.

The visual fields shown in fig. 2 - 7 are related to the normal


line of sight. If the task requires an upright head position the normal
line of sight has to be raised to an angle of 10° - 20° below the
horizontal line.

Sitting positions

Considering the design criteria "visibility of signals" and


"reach of controls", one has to distinguish between an upright
(normal) sitting position, a forward and a rearward sitting posi-
tion.

Upright (normal) sitting position

The upright sitting position is typified by a hip angle (angle


u 9 of the body template according to DIN 33 408, part 1) of about
85°-90° and a straight torso line. If the head is kept in a comfort-
able posture - as defined above - the eye point is located on a
vertical reference line tangent to the front edge of the work sur-
face. This eye location results from a horizontal clearance of 30-
50 mm between the front edge of the work surface and the forward
torso contour of the template (Fig. 8). In relation to the hip angle
mentioned the upright sitting position for instance is taken up
while writing.

Forward sitting position

Characteristics of the forward sitting position are a hip angle


of less than 85° and, should the occasion arise, a shoulder pivot
point benched ahead. This sitting posture will be achieved if one
has to grasp objects that are beyond the functional hand reach area.

Rearward sitting position

If a rearward sitting position is taken up the hip angle is


about 100°-105°. Then the eye point is desplaced behind the vertical
reference line as defined above. In good approximation the eye point
moves on a circular arc around the hip hinge point.

Supposing an unchanged inclination of the head (angle between


head axis and torso line) the angle between the normal line of sight
and the horizontal line is reduced to 10°-20°.
BASIC DATA FOR THE DESIGN OF CONSOLES 119

D
.~

M. , ••• ",eel ft.l • • f


..... f.r "I'" .,t •• i ~

. . . . . . . . tUICl U_l • • 1
•••• ,., c.l •• ~ .,t_ "

o ,
(~J
0.1, •• 1 •• ,.el 'l_l •••
• , •• f ... HI t _'t •• 1t

Fig. 2. Vertical dimensions of the direct field of view

The rearward sitting position is corresponding to a body posture


taken up on a seat by someone is observing his environment in a re-
laxed way.

Hand reach

The hand reach or grasping reach of a person is limited to a


three-dimensional surface partly enclosing the human body. Cutting
this surface with any plane will lead to an open or closed curve
which encloses a surface of hand reach facing the operator.

The maximum hand reach capability and the hand control reach
may be defined as follows.

Maximum hand reach capability

According to the measuring methods as defined in DIN 33 402,


part 1 (body dimensions of people) the maximum hand reach capability
of a person can be expressed by a three-dimensional surface contain-
ing all points which can be touched with
- maximum stretched body posture at standing or sitting position,
- shoulder-blades touching a wall, and
- maximum stretched limbs.
120 H.-P. ROHMANN

11.141 .f
.t,. Ii

V'IOm-._....nIl'>;w , 0 d
,~""~:-..:htj..". III 1lI./ • II ow

o •
•• l •••

•• f.I
4.,.cI ".'4 of
c.10., .". it

O,l, •• 1 'l'O'l 1.01. or


O~ .. hr .'.w f , 1'. l . , ••• It

Fig. 3. Horizontal dimensions of the direct field of view

In some cases data of the maximum hand reach capability are


usable to position controls at work stations (e.g. sensor buttons).

Hand control reach

The hand control reach (functional arm reach, grasping reach)


is limited to a three-dimensional surface within which the controls
are to be arranged.

The spatial dimensions of this envelope depend on


- the dimensions of the control,
- the kind of grip necessary for operating the control (e.g. knob
held in a three-finger grasp) ,
- the direction of control operation,
- the amount of control's deflection, and
- the operating forces of the control.

Dimensions of the hand control reach

Basic data for the determination of the hand control reach are
the body dimensions of men belonging to the 5th percentile level.
Data are available for standing and sitting position.
BASIC OAT A FOR THE DESIGN OF CONSOLES 121

Maximum secondary field of


view for light Itimuli

Maximum lecondary field of


view for colour Itimuli

.5° 0°

Optimal field of view


- fixed he.d/fix.tin~ eye. -
-55° for lilht Itimuli

Fig. 4. Vertical dimensions of the secondary field of view and of


the optimal field of view - fixed head/fixating eyes - for
light stimuli

Maximum upward hand control reach at standing position

As long as a standard for a drafting template usable for the


design of work places where the operator has to work in a standing
position is not available, the maximum upward hand control reach
has to be derived from body dimensions of adults (DIN 33 402, part
2). A dimension of reference is the "functional upward reach with
both arms" of the 5th percentile-man (=1910 rom). Supposing a comfort-
able standing posture the maximum upward hand control reach is li-
mited to approximately 1850 rom above the floor.

Side view contour of the hand control reach at sitting position

For the determination of the side view contour of the hand


control reach at sitting position the 5th percentile body template
according to DIN 33 408, part 1 may be used.
122 H.-P. RUHMANN

00

~
0
1111. 110" ~ •• ,.u • •• condery
Vi.W for l'lh, el,.ut,
f,.14 of

~~
~.
.,u.
rod
. bluol,oll" ..

0", •• 1 f,.ld or .,.v


o
- (,c.d h.,d/fl.aLlaa . , •• -
for l"hL ",.yl1

Fig. 5. Horizontal dimensions of the secondary field of view and


of the optimal field of view - fixed head/fixating eyes -
for light stimuli

It has to be taken into account that the maximum sitting height


depends on the design principle of the console (see chapter "Con-
sole"). The vertical distance between the seat and the desk surface
is determined by considering a vertical clearance of 20-30 mm bet-
ween the upper contour of the thigh and the lower side of the desk
surface (Fig. 8).

Assuming an upright sitting position (hip angle 85°-90°) a


horizontal clearance of 30-50 mm between the forward torso contour
and the front edge of the desk has to be designed. The simplified
lumbar pivot is set to an angle of 180° equivalent to a straight
torso line.

By means of four different hand segments the manner of grasping


and operating controls can be taken into consideration.

Thus, the side view contour of the maximum hand control reach
in a plane parallel to the sagittal plane is designed by turning
the arm around the shoulder pivot (Fig. 9). For this procp.dure the
elbow angle is set to 180°.
BASIC DATA FOR THE DESIGN OF CONSOLES 123

100·

Maxiau. te~ti.ry field of


-gOO
view for liaht Iti.uti

-if--r-+---\-++-+-OO

Maxiau. tertiary field of


view for colour Itimuli

~rtl~----+o·

Opti •• l field of view


- aO'Yina bead/fixatiDI eyea -
.65 0 for li.he Iti.uli

Fig. '6. Vertical dimensions of the tertiary field of view and of


the optimal field of view - mOving head/fixating eyes -
for light stimuli

By tilting the torso forward and aside it is possible to extend


the functional hand reach for a short time. The tilting of the torso
is reproducable by reducing the lumbar angle (adjustment range is
168°-195°). The deflection ahead of the shoulder pivot, however,
cannot be reproduced with the drafting template. According to the
percentile level. the horizontal deflection of the shoulder pivot
may be up to 100 mm. Proceeding on a rough estimation it may be as-
sumed that the 5th percentile-male is able to extend the functional
hand reach up to 150-.200 mm forward and aside for a short time.
124 H.-P. ROHMANN

•••• u. "', •• ,' f •• ,. of


9.eW (oJ "Iht .t'.~l.

r .. " 01
: ",.v for colo~r .11 ,.ul,

O,l' •• ' f,.l. of .l.V


- _0.'., •• "f' •• t'.1
~ . , •• -
fOI l"~l 'l~.~"

Fig. 7. Horizontal dimensions of the tertiary field of view and of


the optimal field of view - moving head/fixating eyes -
for light stimuli

Plan view contour of the hand control reach at sitting position

For the time being, a movable drafting template for the design
of the plan view of the human body and the hand control reach is not
available. However, the hand control reach can be defined construct-
ively by approximation in a simple way.

For this purpose the hand-anm-system is considered as a two-


lever-system turning around a hypothetical shoulder pivot point
(SDP) and an elbow pivot point (EDP). The dimensions of this system
BASIC DATA FOR THE DESIGN OF CONSOLES 125

,
Fig. 8. Basic data resulting from an upright (normal) and from a
rearward sitting position

represented in Fig. 10 correspond to the body dimensions of the 5th


percentile-male according to DIN 33 402, part 2.

The range of movement concerning the adduction of the arm at


the shoulder joint (33°-106°) also corresponds to the 5th percen-
tile*. In view of this range it can be assumed that the stretched
arm turns around a hypothetical vertical axis centered in the
shoulder pivot point as long as the elbow pivot point intersects
the sagittal plane. Thus the hand control reach at shoulder height
is limited by two circular arcs with a radius of rA, rA being the
distance between the shoulder pivot point and the grip axis, and
their centres at the right and left shoulder pivot point (SDPr,
SDP1). Tangential to these arcs are two other circular arcs with
the radius rUA, rUA being the distance between the elbow pivot point
and the grip axis, with a common centre at EDPM in the sagittal
plane.

The hand control reach at work surface height can be designed


in the same way. From the side view the shortened radii rA and rUA
are derived. The dimensions of the side view are related to an up-

*For the range of movements at the joints, percentile levels are


derived in the same way as for body dimensions; see Damon et al.,
1966.
126 H.-P. RUHMANN

Fig. 9. Determination of the side view contour of the hand control


reach using the body template according to DIN 33 408,
part 1

ri.ght sitting position. The vertical distance between the seat and
the surface below the work surface is derivable from the thigh clear-
ance height of the 5th percentile-mrueand a vertical clearance bet-
ween the thigh and the surface below the work surface amounting to
20-30 mm. This procedure leads to a vertical distance between the
seat and the surface below the work surface of 150 mm.

By tilting the torso forward and aside an extension of the


functional hand reach is possible far a short time. This extended
hand control reach may be constructed by approximation if one as-
sumes that the pivot points are travelling on a circular arc with
a radius of 150 mm around the hypothetical shoulder pivot point
(SDPr and SDPl). Thus, the envelope enclosing this family of curves
forms the extended hand control reach (Fig. 11).

CONSOLE, INFORMATION BOARD AND PANEL

Console

Consoles are elements for the installation of displays and


controls. The work surface may be horizontal or slightly inclined
to the operator.
BASIC DATA FOR THE DESIGN OF CONSOLES 127

/
/
/

/
I
rt~t.1...
I

I
I ~~
"'J- . . . . . sOP
\I
1·I \ .-;i
I

I "-
\ For the 5th percentile·.. ":

\
\
\
"-
"-,
......

Fig. 10. Construction of the plan view contour of the hand control
reach

Considering the basic design principles of consoles, one dis-


tinguishes between consoles without upper segments, consoles with
a low upper segment (e.g. for the installation of signal lamps and
legend lights), consoles with an upper segment medium-sized (e.g.
for the installation of VDTs) and consoles with stacked segments
(Fig. 12).

The front part of the work surface is usable for writing work,
for the setting up of equipment (e.g. keyboards, Fig. 13 bottom) ,
or for the installation of indicators and controls (Fig. 13 top).

Those controls operated frequently should be grouped within


the privileged area of hand control reach, i.e. in the front part
of the work surface. In order to minimize the parallax for correct
instrument reading, displays should be installed into the upper
segment (s) .
128 H.-P. ROHMANN

Fig. 11. Envelope of the extended hand control reach


----- at shoulder height, ----- at desk height

Viewing the consoles from the top, they can have a straight
or segmented design.

Information board

Information boards are mounted with those displays (indica-


tors, VDTs, plotters, etc.) which serve as primary equipment re-
quired for monitoring and controlling the system. Information boards
are usually placed behind a console without any upper segment or
with a low upper segment.

The surface on which displays are to be arranged is above the


vertical distance HU (Fig. 14). This vertical distance depends on
the geometry of sight relating to the design eye point (see chapter
"Determination of the design eye point") and the obstruction of the
visual field caused by any objects in front of the work station.

As well as for information boards, the geometry of sight is ap-


plicable to control rooms with windows in order to monitor the pro-
BASIC DATA FOR THE DESIGN OF CONSOLES 129

Fig. 12. Design principles of consoles


without upper segment
. • . • . . . with a low upper segment
- - - - - _. with an upper segment middle-sized
-.-.-.-.-.-: with vertical stacked segments

cess by direct view. T,he lower window frame of the objects of in-
terest must be above the line limiting the view ahead.

Viewed from aboved, information boards can have a straight or


segmented design.

Panel

Panels are elements for the arrangement of displays and con-


trols. The mounting area for equipment can be extended by a work
surface. The work surface can be designed horizontal or it is in-
clined towards the operator. Usually the panel forms a boundary sur-
face of the control room (Fig. 15).

In view of the visibility of displays and controls installed


into the vertical plane of the panel and into the work surface, the
geometry of sight shown in Fig. 14 has to be applied in the same
way.

Considering the plan view, panels can be designed straight or


segmented, while in some cases they are L-shaped or U-shaped.
130 H.-P. ROHMANN

Fig. 13. Consoles with various desk surfaces

Location and interrelation of work stations

Depending on the type of the control room, not only the spatial
arrangem~nt of the elements defined above is specified, but also
their interrelation.

Proceeding from common applications and combinations respective-


ly, the following types of control rooms are to be distinguished
(Fig. 16):

- control room only equipped with a panel (e.g. chemical processing


plants, energy distribution plants, Fig. 16 top left),
- control room only equipped with a console (e.g. signal cabin with
direct view to the railway trackage, Fig. 16 top middle),
- control room equipped with a console and a panel (e.g. electricity
power and heat station, chemical engineering plants, Fig. 16 top
right), and
BASIC DATA FOR THE DESIGN OF CONSOLES 131

Fig. 14. Geometry of sight for an information board

Fig. 15. Panel with a work surface inclined towards the operator

- control room equipped with a console, an information board and a


panel (e.g. nuclear power plant, Fig. 16 below).

DIMENSIONS

All vertical dimensions are related to a horizontal reference


plane (e.g. the floor or a pedestal) and the horizontal dimensions
are related to a vertical reference line tangent to that point of
the front work surface which is next to the standing or sitting
w
~

"

~
Fig. 16. Arrangement principles of the work stations according to different types of control ~
rooms ~
o
~
~
~
Z
Z
BASIC DATA FOR THE DESIGN OF CONSOLES 133

operator. Basic data for the recommended dimensions are anthropo-


metric dimensions of men aged between 16 and 20, belonging to the
5th, 50th and 95th percentile (see DIN 33 402, part 2). Some dimen-
sions are derived from a design-work with the body templates accord-
ing to DIN 33 408, part 1.

Clearance dimensions

Clearance dimensions are required for operator mobility, for


the leg space of consoles and for maintenance.

Operator mobility clearance

Basic data for the operator mobility clearance at work places


according to the body posture are represented in DIN 33 402, part 3.
If there is - for instance - a wall behind the operator, the hori-
zontal clearance between the vertical reference line mentioned and
the wall has to be at least 100 mm (see § 24 (1) ArbStattV = regu-
lation for workshops) .

Leg clearance

The dimensions of the leg space shown in Fig. 17 are recommended


for those types of consoles which have a work surface for writing or
for the setting-up of equipment. The horizontal and vertical dimen-
sions are related to the 95th percentile-male.

For the leg space of consoles with work surfaces where displays
and controls are to be ~nstalled, the dimensions of Fig. 18 are re-
commended. In view of these dimensions the 95th percentile-male will
find acceptable seating conditions.

Leg clearance at panels with a work surface

At panels, operators have to work in a standing as well as in a


seated position. In view of a tolerable seating accommodation at
panels with a work surface, the leg clearance dimensions must conform
to those represented in Fig. 19.

An unobstructed space all down the line with a height and a


depth of 100 mm has to be provided for the feet and should be de-
signed even for those panels without work surfaces and for panels
with work surfaces allocated in certain areas respectively.

Clearance for maintenance

With regard to occasional maintenance (e.g. change of racks),


the horizontal spacing between the elements in a control room (e.g.
between a console and an information board) should not be less than
700 mm. This minimum clearance results from a crouching working
134 H.-P. ROHMANN

Fig. 17. Leg clearance at a console with a work surface for writing
or for the setting-up of equipment (no equipment installed
into the work surface)

Fig. 18. Leg clearance at a console with a work surface provided


for the installation of equipment

posture. For a kneeling working posture a horizontal spacing of


1000 mm is necessary (see DIN 33 402, part 3).

Determination of the design eye point

For the lay-out of the visual field and for the determination
of the blind sectors, a design eye point has to be established.

The determination of the design eye point depends primarily on


the tasks of the operators; therefore the extent of automation is
of fundamental importance. One has to distinguish between:

- a fully automatic process, crew mainly monitoring, rare control


operations,
- a partly automatic process, crew mainly monitoring, occasional
control operations, and
- a manual process control, crew has to monitor and control the
process permanently.
BASIC DATA FOR THE DESIGN OF CONSOLES 135

Fig. 19. Leg clearance at a panel with a work surface (depth of


foot clearance should be 100 mm)

In order to determine the design eye point considering control


rooms mainly used for the monitoring of fully or partly· automatic
processes one has to start from the rearward sitting position (see
chapter "rearward sitting position") of the 5th percentile-male.
Facing the arrangement of displays it is furthermore useful to take
a design principle of the console as shown in Fig. 13 and the cor-
responding dimensions (Fig. 18) as a basis. Due to the higher work
surface (in relation to the console type with a work surface for
writing or the setting-up of equipment) and the smaller leg clear-
ance height the result is a lower height of the design eye point
above the floor.

The use of the body template according to DIN 33 408, part 1,


leads to a design eye point located 1200-1210 mm above the horizontal
reference plane and 150-160 mm behind the vertical reference line.

Considering control rooms mainly used for manual process con-


trol, one has to start from the upright (normal) sitting position
(see chapter "Upright (normal) sitting position"). Supposing that
small operators adjust their seat high and big ones low, it is ap-
proved to base the determination of the design eye point on the
body dimensions of the 50th percentile-male, i.e. on a height of the
sitting surface of 450 mm equivalent to the lower leg length of the
50th percentile-male.
136 H.-P. ROHMANN

In that case, the height of the design eye point is independent


of the design principle of the console. The use of the 50th-percentile
body template leads to a design eye point located on the vertical
reference line and 1210 mm above the horizontal reference plane.

Consoles

Consoles without any or with only one upper segment

Consoles without any upper segments and consoles with a low or


middle-sized upper segment which are mainly used for monitoring fully
or partly automatic processes, should have a work surface depth of
less than 790 mm (Fig. 20 top). At a standing position the maximum
forward hand control reach of the 5th percentile-male is limited to
this extent.

If those types of consoles are mainly used for manual process


control, the depth of the work surface should not exceed 500 mm be-
cause the functional forward hand reach in a seated position is equi-
valent to this value (see Fig. 10).

If the whole equipment (displays and controls) is installed into


the upper segment, a smaller work surface can be used for writing
or the additional setting-up of equipment. Preferably the controls
should be grouped in the lower part of the upper segment to provide
easy handling within a comfortable zone of hand reach.

Console types designed for the installation of equipment into


the work surface and certain sectors on this surface designated for
writing must have a work surface at one level.

If there are displays and controls mounted in the work surface


an angle of inclination (ap) of the work surface of less than 10 0 .
does not have any advantage compared with a horizontal work surface
regarding the efficiency of instrument reading and control operat-
ing. On the other hand the disadvantage of an inclined work surface
is a larger overall height, lifting up the line limiting the view
ahead ("view ahead limiting line", Fig. 21). Thus the blind sector
in front of the console is extended.

The angle of inclination of the upper segment (aA) has to be


referred to the designed sitting position. As represented in chapter
"Sitting positions" and Fig. 8, the angle between the normal line
of sight and the horizontal line varies according to the sitting
position. Optimal conditions for instrument reading exist, if the
surface of the upper segment is perpendicular to the normal line of
sight, i.e. the fixation line.

In case of fully or partly automatic process control and main-


ly monitoring tasks, one has to proceed from the rearward seating
~
en
(;

~
»
a
:D
-t
J:
m
C
m
en
C5
z
o
."
n
oz
en
or-
m
en

Fig. 20. Basic data for consoles with and without an upper segment

manual~rocess control fully or partly automatic process control I


a. A 25° - 35° 10° - 20°
I

I
T 500 max. 790 max. w
-- ---------- _ ..- -----------
....
138 H.-P. ROHMANN

Fig. 21. Obstruction of the visual field


horizontal work surface
-.-.-.-.-.-.-: work surface inclined towards the operator

position and from an angle of inclination between the normal line


of sight and the horizontal line of 10°-20°. This angle has a magni-
tude of 25°-35° if an upright sitting position is taken up for manual
process -control.

The overall height of consoles with a middle-sized upper segment


should be limited to 1210 mm if a direct view to the process is re-
quired or if a panel has to be arranged behind the console. Under
these circumstances a larger height of construction would lead to
uncomfortable (tensed) positions of the head and the eyes.

Consoles with vertical stacked segments

At consoles with vertical stacked segments, the horizontal work


surface can be used for writing or for the installation of displays
and controls (Fig. 22 left). The next surface (Al) is especially
suitable for the mounting of controls. Therefore, the depth of the
work surface should not exceed 300-500 mm.

The vertical arrangement of the stacked segments should roughly


form a circular arc with a radius of 550-600 mm around the design
eye point related to the upright sitting position. The upward in-
clination of the surface Al is approximately 30°.

As the surface A2 is designated for the installation of those


displays which require frequent monitoring, this surface has to be
BASIC DATA FOR THE DESIGN OF CONSOLES 139

Fig. 22. Basic data for consoles with vertical stacked segments

perpendicular to the normal line of sight. Hence, the vertical seg-


ment A3 should only contain those displays and controls infrequently
used. The same rule has to be applied to segment A4 which may be in-
clined towards the operator by an angle of 20°-40°.

At consoles with only 3 stacked segments (Fig. 22 right) the


mounting area available for equipment of the corresponding segment
A1 is expanded. The vertical segment A2 is especially suited to the
installation of displays and the representation of flow charts. Dis-
plays requiring frequent reading should be arranged in the lower
part of the segment A2, since they are to be fixated without move-
ments of the head.

This type of consoles should not exceed an overall height of


1600 rom. If this dimension is taken into account, the displays ar-
ranged in the uppermost segment (segment A4 and segment A3 respect-
ively) are within the optimal field of view - moving head/fixating
eyils (see chapter "Field of view - IOClving head/fixating eyes" and
Fig. 6).

Information board

When designing information boards which are for instance placed


behind a console, it has to be considered that the height of the sur-
face for the mounting of displays above the horizontal reference
plane (dimension Hu ' see Fig. 23) is determined by the line limiting
the view ahead. Basic data for the design of information boards are
140 H.-P. RUHMANN

Fig. 23. Basic data for an information board

the visual distance and the back\olard inclination angle (y) of the
mounting surface for displays. These data result from the geometry
of sight related to the design eye point and further data represented
in Fig. 23.

In Fig. 23 the following symbols stand for:

H = vertical distance between the design eye point and the


AP
horizontal reference plane,

- horizontal distance between the design eye point and the


vertical reference line,

vertical distance between an object obstructing the view


ahead (e.g. rear edge of the console or upper edge of a
console-segment) and the horizontal reference plane,

horizontal distance between the obstructing object and


the vertical reference line,

height of the information board's surface for the mounting


of displays (depending on the construction height of
VDTs) •

With regard to the auxiliary data

Tp = depth of the console, and


BASIC DATA FOR THE DESIGN OF CONSOLES 141

T = depth of the clearance for maintenance


W

the dimensions H , D and y can be computed.


u

Example:

HAP = 1210 mm, LAP = 150 mm, HSV = 910 mm, LSV = 850 mm

HIF = 500 mm, Tp = 1050 mm, TW = 1000 mm.

Then:

tg,a = (HAP - HSV) /LSV = 0.353, a = 19.4°,

Hu HAP - tg a (Tp + LAP + TW) = 433.5 mm,

D V«HAP - Hu )/sina)2 - (H IF /2) 2 '= 2319.7 mm,

tg(B/2) = HIF/2D = 0.108, B/2 = 6.1°,

y = a - B/2 = 1 3 • 3 ° .

Panel with a work surface

Viewing the design criteria concerning the visibility of those


instruments which are installed into the surface inclined to the op-
erator, the geometry of sight represented in Fig. 23 has to be ap-
plied in the same way. According to the spatial arrangement (see
chapter "Location and interrelation of work stations"), the blind
sectors on the panel and on the work surface are caused by upper
edges of the console or the information board. If the panel contains
nothing but secondary equipment (equipment of redundancy), it is per-
missible that the work surface is visible from a standing position
only. If the work surface is inclined the angling should be between
20° and 30°.

Adjustment range of the foot rest

Especially for small operators foot rests are a basic require-


ment for using the vertical adjustment of the seat. It is the verti-
cal adjustment of the seat that provides for an optimal instrument
reading and control operating.

Assuming a vertical clearance height of 670-690 mm at the front


part of the console and the lower leg length inclusive footwear of
the 5th percentile-man taken as a basis, the result is a vertical
adjustment range of the foot rest of at least 125 mm, while the foot
rest surface is inclined 10°-25° to the operator (Fig. 24). Viewing
the depth of this surface, it has to be pointed out that the foot
rest has to support the entire foot length of the 95th percentile-
142 H.-P. ROHMANN

Fig. 24. Basic data for the lay-out of the foot rest

male. That is ensured if the depth is at least 330 mm. If foot rests
are fixed to the floor the front edge of the step should be 150 mm
behind the vertical reference' line. For this kind of lay-out a
longitudinal adjustment is not necessary. The horizontal extension
of the step should be at least 500 mm. A foot-operated control pro-
vides for vertical adjustment and, if available, for the adjustment
of the inclination, too.

CONCLUDING REMARKS

Finally, the dimensions recommended for consoles shall be com-


pared with other data.

In Fig. 25 three consoles are drawn one above the other. This
type of console is required for a sitting position with vision over
the top border. It is furthermore assumend that this type of console
is mainly used for manual process control from an upright (normal)
seated position.

One console conforms to those data published in MIL-STD 1472B,


the second one is defined in a French draft standard and the third
one, as mentioned in the Introduction, is the result of a small
working group belonging to the subcommittee of the FNErg.

As it was to be expected, the contours are quite similar but


the dimensions are very different. If the (German!) 5th percentile-
male, for instance, takes up a stretched posture at maximum seat
height, his eye point is below the upper surface of the console's
panel according to the MIL-Standard. The leg clearance height of
the French console is too small, referring to the lower leg length
of the (German!) 95th percentile-male.
BASIC DATA FOR THE DESIGN OF CONSOLES 143

Fig. 25. Consoles for sitting position with vision over top
standard console middle-sized (MIL-STD-1472B)
- - - - - . console according to a working draft (DIN)
type: equipment installed into the desk sur-
face, predominantly manual process
control
-.-.-.-.-.-: console according to a French draft proposal
("Positions and dimensions of men working on
machines and plant")

Consequently, further efforts should be made to obtain uniform


data (e.g. by ISO-work).

As the author pointed out, it is not usefull to define dimen-


sions for a "standard console". Primarily it depends on the oper-
ator's task, as far as the extent of automation is concerned, which
kind of design will be sui table.

REFERENCES

Damon, A., Stoudt, H. W., and McFarland, R. A., 1966, "The Human
Body in Equipment Design," Harvard University Press, Cambridge
(Mass.).
Schmidtke, H., 1981, Arbeitsplatzgestaltung, in: "Lehrbuch der Ergo-
nomie ," H. Schmidtke, ed., Hanser, Mlinchen, Wien.
Schmidtke, H., and Ruhmann, H. P., 1981, Betriebsmittelgestaltung,
in: "Lehrbuch der Ergonomie," H. Schmidtke, ed., Hanser,
Munchen, Wien.
144 H.-P. RUHMANN

Schmidtke, H., and RUhmann, H. P., 1978,"Ergonomische Gestaltung von


Steuerstanden," Forschungsbericht No. 191 der Bundesanstalt
fur Arbeitsschutz und Unfallforschung, Wirtschaftsverlag
N.W., Bremerhaven.
Schober, H., 1960, "Das Sehen," Vol. I, VEB Fachbuchverlag, Leipzig.

Laws, Regulations, and Standards

MIL-STD 1475B, 1974, Human Engineering Design Criteria for Military


Systems, Equipment and Facilities, Department of Defense,
Washington, D. C.
DIN 33 414, Part 1 (Draft), 1982, Ergonomic Design of Control Rooms;
Concepts, Basic Principles, Dimensions for Seated Work Posi-
tions, Beuth Verlag, Berlin.
Pr X 35-104, (Draft), 1980, Positions and Dimensions for Men Working
on Machines and Plant, AFNOR, Tour Europe CEDEX 7, Paris.
ArbStattV, 1975, Verordnung fiber Arbeitsstatten (Regulation for Work-
shops), Federal Republic of Germany.
ERGONOMIC DATA FOR CONSOLE DESIGN

William H. Cushman

Eastman Kodak Company


Rochester, N.Y./USA

INTRODUCTION

Console workplaces are suitable for many types of military


and industrial jobs that involve monitoring a number of displays,
making decisions based on information conveyed by the displays,
and operating controls to carry out the decisions. Typical appli-
cations include air defense, air traffic control, nuclear power
plant operations, industrial process control, and plant security.

The console designer is responsible for the basic configuration


of the workplace and the placement of displays and controls. Visual
displays must be located where they can be easily seen and inter-
preted, and controls should be placed where they can be easily
reached. Hence, the anthropometric characteristics of the operators
and their perceptual-motor capabilities and limitations must be
known. Other factors that must be considered in order to maintain
operator alertness and to minimize fatigue over long periods of
time include operator comfort and safety (including body posture) ,
mental workload, and task complexity.

A considerable amount of ergonomic data has been obtained to


assist designers of console workplaces. This paper reviews those
data concerning functional arm reach for horizontal and vertical
work surfaces, three-dimensional reach envelopes, work surface
height, primary and secondary visual work zones, reaction time as
a function of control type and location, console configuration,
and console dimensions. When properly applied, this information
will enable designers to design consoles that ensure a high level
of operator performance and system reliability.

145
146 W. H. CUSHMAN

The data reviewed in this paper and recommendations for console


design were obtained primarily from technical publications originating
in the United States and Canada. For a discussion of console design
from a European perspective, see H. RUhmann's paper "Basic Data for
the Design of Consoles".

FUNCTIONAL ANTHROPOMETRY

One of the prime requirements for the layout of any console


is that all controls be located where they can be easily reached
and manipulated. To achieve this, it is necessary to know just how
far various percentages of the operator population can reach to
perform anticipated tasks. Static anthropometric measurements
(e. g., Webb Associates, 1978, Vol. II) have historically been less
than satisfactory for predicting dynamic reaching capabilities.
Functional anthropometric measurements, such as the ones described
below, have been more successful.

Work Area for a Horizontal Surface

Most consoles have an extended shelf or other horizontal sur-


face which may be used as a writing surface or for supporting books,
telephones or other small items. Sometimes controls and operating
diagrams are also placed on this surface.

Barnes (1949) has described the normal working area for a


horizontal plane as the area that can be conveniently reached with-
out extending the upper arm. The outer boundary for the right hand,
according to Barnes, is a semicircle determined by sweeping the
right hand and forearm across the work surface while the upper arm
remains at the side of the body in a natural position. The boundary
for the left hand is also a semicircle and is similarly determined.
The normal work area was defined as the area enclosed by the two
semicircular arcs. For males, the radius of each arc is 15 1/2 inches
(39 cm). Barnes also described the maximum working area for a hori-
zontal surface as the area that can be reached by extending the arm
from the shoulder.

In later editions of his book, Barnes (e.g., 1958) cited a


study by Farley (1955) which provides additional information con-
cerning the dimensions of the normal and maximum work area and in-
cludes data for females as well as males. The radii of the arcs for
females and males for the normal work area are 14 inches (36 cm)
and 15 1/2 inches (39 cm), respectively. For the maximum work area,
the radii are 23 1/2 inches (60 cm) for females and 26 1/2 inches
(67 cm) for males. The work areas proposed by Farley, although
similar in shape to those proposed by Barnes, are significantly
smaller. Barnes assumed that the elbows rest near the front edge
of the work surface, while Farley assumes a horizontal distance
ERGONOMIC DATA FOR CONSOLe DESIGN 147

of about 7 1/2 inches (19 cm) between the resting position of the
elbows and the front edge of the work surface.

Squires (1956) has proposed a somewhat different horizontal


work surface contour --a prolate epicycloid-- because the upper arm
and elbow do not naturally remain in the same position as the hand
and forearm move across the work surface, and because it is diffi-
cult to move the hand and forearm more than about 120 0 with the
elbow as a fixed pivot. The parametric equations for determini~g
the coordinates of the outer boundary of the work area are given
on page 2 of Squires'report. These data are useful in designing con-
soles where displays and controls must be arranged compactly and
operator movement is relatively restricted.

A comparison of methods for determining the work area for a


horizontal surface has recently been made by Das and Grady (1981).
They concluded that Squires' method was better than the method pro-
posed by Barnes (1949) or Farley (1955).

Work Area for a Vertical Surface

The boundaries of the maximum work area for a vertical surface


beyond a shelf have been determined by Chaffee and Emanuel (1964).
Shelf depth was varied from 0 (no shelf) to 25 inches (64 cm) and
shelf height from 28 inches (71 cm) to 30 inches (76 cm). The results
of the investigation are given in a series of figures showing the
maximum work areas for the right hand of seated 5th, 50th, and 95th
percentile males for various shelf depth and shelf heights.

The Chaffee and Emanuel (1964) data may be used to help decide
where to place controls on vertical surfaces, such as instrument
racks, if the operators are to remain seated. The data also suggested
that the depth of a console shelf should not exceed 20 inches (51 cm)
and that 28 inches (71 cm) is the preferred shelf height. The latter
recommendation is one inch below the height recommended by Ayoub
(1973). (See section on Work Surface Height for additional details.)

Reach Capabilities in Three Dimensions

One of the most recent and extensive studies of reach capab-


ilities has been conducted by Kennedy (1978). The purpose of the
study was to derive complete three-dimensional grasping-reach
envelopes for seated 5th, 50th, and 95th percentile men and women
and to present the results in a form that could be readily used by
designers. (See also Kennedy, 1964).

Kennedy's apparatus consisted of a rotatable chair beneath a


rigid arch having movable rods attached at 15 0 intervals along one
side. If extended inward, the rods would all converge at the center
of the arch, a point 24 inches (61 cm) directly above the seat re-
148 w. H. CUSHMAN
ference pOint (SRP)*. Thus, the center of the arch was at shoulder
height and remained fixed as the seat was rotated between trials.
On each trial, the subject grasped the knob at the end of each rod
with the thumb and forefinger of the right hand and pushed outward
as far as possible (arm fully extended) without pulling the shoulder
away from the chair backrest. Reach data --the distances between the
ends of the rods and the center of the arch-- were obtained for vert-
ical planes radiating at 15 0 intervals from an imaginary vertical
line passing through the center of the arch and SRP.
The results of the Kennedy (1978) study were presented in both
tabular and graphic formats for 5th, 50th, and 95th percentile men
and women. Both sets of data indicate the boundary of the reach enve-
lope for selected horizontal (X-Y) planes and vertical (Y-Z and X-Z)
planes. Data for horizontal (X-Y) planes from 6 inches (15 cm) below
the SRP to 54 inches (137 cm) above are given. Tables and graphs for
Y-Z vertical planes (frontal planes) include planes from 24 inches
(61 cm) aft of the SRP to 36 inches (91 cm) forward; data for the X-Z
vertical planes (fore-aft planes) include those planes from 24 inches
(61 cm) to the left of the SRP to 36 inches (91 cm) to the right.
In the Kennedy (1978) study described above, and most other
studies of functional reach capabilities, the subjects have been
required to sit with their shoulders against the seat backrest.
Few, if any, console operators are required to maintain this
posture. Therefore, in a practical sense, the sizes of the reach
envelopes have usually been underestimated. This issue has been
addressed in a study by Roth, Ayoub, and Halcomb (1977). These
investigators measured the reach envelopes for seated males and
females who were permitted to lean forward as much as 100 to achieve
maximum reach. Data for restrained subjects were also obtained with
the same apparatus so that the reach envelopes for unrestrained and
restrained subjects could be compared.
The shape and volume of the three-dimensional reach envelope
change significantly with the orientation of the hand. Studies by
Dempster (1955) and Dempster, Gabel, and Felts (1959) included an
analysis of photographic traces showing the extent of hand move-
ment over a series of frontal planes as the hand assumed each of
eight different orientations, while grasping a special handgrip de-
vice. Dempster coined the term "kinetosphere" to describe the enve-
lope for a sing~hand orientation. The largest kinetospheres were
obtained when the hand assumed a 0 0 orientation (i.e., when the hand-
grip was vertically oriented). The smallest kinetospheres were ob-
tained when the hand assumed either a supine orientation or a 90 0
orientation. Hence, when deciding where to place a control on a con-
sole, the designer should consider the reach envelope for the hand
orientation that must be assumed when operating that control.

*SRP is defined as the midpoint of the line formed by the inter-


section of the seat surface and backrest.
ERGONOMIC DATA FOR CONSOLE DESIGN 149

Seat characteristics also may alter the reach envelope of a


seated operator. The angle of the backrest, for example, affects
the relationship between the body and the space immediately forward.
In one investigation, Ely, Thomson, and Orlansky (1956) identified
the optimum areas for the placement of manual controls for various
seat backrest angles. The distance between the SRP and the farthest
forward position within the optimum area decreased by about 7 inches
(18 cm) as the seat backrest angle was increased from 0° (vertical)
to 20°.

Studies such as Kennedy (1978), Roth et ale (1977), Dempster


et ale (1959), and Ely et ale (1956), clearly show that factors
such as anthropometry, body restraint, body positioning, type of
task, and so forth may have a profound effect on the outcome of
any functional reach investigation. Hence, the user of functional
anthropometric data must be very cautious when making extrapolations
for new applications where conditions are anticipated to be signif-
icantly different. One solution is to apply correction factors to
compensate for the differences (Stoudt, 1978).

The studies discussed above are only a few of many concerning


human reach capabilities in three dimensions. Reviews of the lite-
rature have been made by Kennedy (1978) and Stoudt (1978).

WORK SURFACE HEIGHT

The optimum height for a horizontal work surface depends on


the visual and manipulative requirements for the tasks that are to
be performed. For tasks typically performed at consoles by seated
operators (e.g., writing), the preferred height for a horizontal
work surface (based on anthropometric and biomechanical data) is
27 1/2 to 29 1/2 inches (70 to 75 cm) for females and 29 to 31 inches
(74 to 79 cm) for males (Ayoub, 1973). The recommended height for
a surface supporting a typewriter or video display terminal (VDT)
keyboard is slightly lower -- 25 1/2 inches (65 cm) for females and
27 inches (68 cm) for males (Grandjean, 1981). However, the vertical
distance between the seat surface and work surface and seat adjust-
ability are probably more important than the absolute height of the
work surface.

VISUAL CONSIDERATIONS

Consoles must be designed so that the operator is able to see


all critical displays from the normal working position. As a first
step, static anthropometric data (e.g., eye height sitting) may be
used to predict the position of the eyes for 5th and 95th percentile
operators. Viewing angles and viewing distances should then be con-
sidered, along with the reach capabilities of the operators, to
150 w. H. CUSHMAN
determine acceptable positions for the work surfaces, displays, and
controls.

Recommendations for preferred and maximum viewing angles in


both lateral and vertical directions may be found in Morgan, Cook,
Chapanis, and Lund (1963) and MIL-STD-1472C. The recommendations
for eyes only and head and eyes rotations are given in Table 1.
Note that the reference line of sight is horizontal (parallel to
the ground) rather than 5° below horizontal (the standard line of
sight) or 15° below horizontal (the normal line of sight). The
preferred range is the range through which the operator can view
displays and controls "with speed and accuracy by eye rotation
alone". The maximum range is the range through which the operator
can view displays "without straining the neck or eye muscles".
Designers will find the recommendations given by Morgan et ale
(1963) to be very useful because ranges of acceptable viewing angles
are specified rather than single optimum values.

An attempt to determine the preferred amount of downward head


and eye rotation has been made by Lehmann and Stier (1961) in a
study described in some detail in Grandjean's (1981) book. They
found that the preferred amount of downward head and eye rotation
for seated workers was 32 - 44°; for standing workers the preferred
range was 23 - 37°.

The preferred and maximum ranges given in Table 1 apply to con-


ventional displays such as dials with pointers, horizontal and ver-
tical scales, and digital readouts. Flashing light indicators, how-
ever, can often be detected without moving either the eyes or head.
Kobrick (1965) measured reaction times for lights flashed at various
locations in the peripheral visual field. He found that flashing
light indicators could be detected about equally well at peripheral
locations as far out as 65° or more in all directions from the line
of sight, except directly upward. In the central region of the upper
hemisphere, performance declined beyond about 40°. The author con-
cluded that some flashing indicators could te safely placed at peri-
pheral locations, permitting the central visual space to be more
effectively occupied by displays that require constant monitoring.

MarOR PERFORMANCE

The time required to activate a control depends on both the


type of control and its location. Sharp and Hornseth (1965) obtained
performance times for the operation of three types of controls
(knobs, toggle switches, and push buttons) at each of 36 locations
to the left of the mid-sagittal plane and above the SRP. The data
for the closest console position were presented as two-dimensional
performance-time contour maps. The best performance was obtained
when the controls were approximately 25° to the left of the mid-
ERGONOMIC DATA FOR CONSOLE DESIGN 151

Table 1. Preferred and Maximum Viewing Angles*


(Adapted from Morgan et al., 1963)

Preferred Range Maximum Range

Eyes only rotated 15°L to 15°R 35°L to 35°R


OOU to 30°0 25°U to 35°0

Head and eyes rotated 95°L to 95°R


75°U to 85°0

sagittal plane and 25 inches (64 cm) above the SRP. The fast per-
formance area for the toggle switch was considerably smaller than
the corresponding areas for both the knob and push button. This
finding suggests that the selection of a location for a toggle
switch is more critical than the selection of a location for the
other two types of controls.

In a somewhat similar study, Arora (1976) measured the time


to activate push-button switches with the right hand as a function
of reach distance, direction of reach, and angular inclination of
the instrument panel surface. The best performance (shortest re-
sponse times) occurred when the reach was approximately 30° to the
right of the mid-sagittal plane. Performance time increased linear-
ly with reach distance, but panel inclination had no Significant
effect. These findings are consistent with the results of the Sharp
and Hornseth (1965) study described above and several others cited
in the Arora (1976) paper.

CONFIGURATION OF SEGMENTEO CONSOLES

The number of controls and displays for some consoles is so


great that a single panel design is not feasible. In these situa-
tions, a segmented design will usually improve operator performance
significantly. MIL-STD-1472C recommends a segmented design whenever
the panel width for a single seated operator exceeds 44 inches
(112 cm).

A systematic evaluation of segmented console configuration has


been carried out by Siegel and Brown (1958). They built a full-scale
mock-Up of a segmented console and varied the angle between the
front panel and side panels in 10° increments from 35° to 65° (see
Figure 1). Subjects (either individually or in pairs) followed a
sequence of instructions to operate controls (toggle switches,
rotary switches, push buttons, etc.) at various locations on the

*Assumes horizontal line of sight with observer looking straight


ahead. L= left, R= right, U= up, D= down.
152 w. H. CUSHMAN

1+--48 in. (122 em)---t

~~~----------~
~\{,~
~,(\.

~"3

Fig. 1. Console arrangement for the Siegel and Brown (1958) study.

front panel and two side panels. Number of seat movements, magnitudes
of seat displacement, number and extent of body movements, and number
of partial and full arm extensions were recorded. Subjective evalua-
tions and preference rankings were also obtained.

Data for single operators consistently favored the console


arrangement with the side panels angled at 65°. For example, both
the number of seat movements and the sum of seat displacements de-
clined as the angle of the side panels was inGreased.

The findings for paired operators were somewhat inconsistent.


Controls on the front panel became easier to operate as the angle
of the side panels was decreased. However, the controls 9n the side
panels became easier to operate as the angle of the side panels
was increased, but then there was a tendency for operators to block
each other's view. As a compromise, the authors concluded that the
side panel angle for two-operator segmented consoles should be
between 50° to 55°.
ERGONOMIC DATA FOR CONSOLE DESIGN 153

DIMENSIONS FOR CONSOLES

Ergonomic data concerning reach capabilities and visual per-


formance have not always been presented in a readily usable form.
One solution to the problem is to use these data to derive sets of
preferred console dimensions for different applications. Kennedy
and Bates (1965), for example, provide recommendations for five dif-
ferent types of consoles based on operator posture (sitting or
standing) and whether or not the operator is required to see over
the top. Console profiles were developed from anthropometric data
(e.g., eye height sitting and reach measurements) and evaluated
with mock-Ups. The recommended dimensions for each type of console
are given in Table 2 and Figure 2.

Consoles designed to the specifications in Table 2 will ac-


commodate 95 % of the u.S. Air Force male population and 60 % of
the U.S. Air Force female population. Obviously, these recommenda-
tions are not appropriate for populations that differ significantly
from the reference population of U.S. military personnel. For example,
if designing for 95 % of U.S. industrial workers, one would need to
consider decreasing dimensions A, B, and D because of the high per-
centage of women in the U.s. industrial work force. Some clearance
dimensions such as F (thigh clearance) would have to be increased.

In addition to the data in Table 2, Kennedy and Bates (1965)


also recommend that the width of the console panel not exceed 36 in-
ches (91 em) for consoles where vision over the top is not required
and not exceed 44 inches (112 em) for consoles where vision over
the top is required. These recommendations, as well as those given
in Table 2, have been incorporated into MIL-STD-1472C with some
minor modifications.*
Perhaps the most comprehensive review of recommendations for
console dimensions may be found in a paper by Ayoub and Halcomb
(1976). In a series of tables (B-9 through B-12), the authors sum-
marize the console design recommendations given in over 40 books,
scientific articles, technical reports, and standards. A discussion
of major console design issues is also included. One of the problems
one encounters when using these data is that different investigators
have used different reference points. For example, one author recom-
mends a minimum knee clearance of 12 inches (30 em) measured from
"the beginning of the desk to the front of the knees at knee level".
Another recommends a minimum knee clearance of 46 inches (117 em)

*In MIL-STD-1472C, the recommendations for maximum console height


(A in Figure 2) and vertical dimension of panel (B in Figure 2)
are 1 1/2 inch (4 em) less than the values shown in Table 2 for
consoles where vision over the top is required. These modifica-
tions were probably made to accommodate the increasing number of
women joining the U. S. Army, Nav.l·' and Air Force.
154 W. H. CUSHMAN

Table 2. Console Dimensions


(Adapted from Kennedy and Bates, 1965)

Dimensions in Inches*
Type of Console A B G H

Sit-stand 62 26 36 35
(157) (66) (91) (89)

Sit (with vision 47t-58 a 22 25t-36 24f-35


over top) (121-147) (56) (65f-91) (62 -89)

Sit (without 51!.-62 b 26 25~-36 24t-35


2
vision over top) (131-157) (66) (65t- 9 1) (62 -89)

Stand (with 62 26 36
vision over top) ( 157) (66) (91 )

Stand (without 72 36 36
vision over top) ( 183) (91 ) (91 )
*Equivalent dimensions in em are given in parentheses.
anA" must never be more than 29~ in. (75 em) greater than J.
b"A" must never be more than 33~ in. (85 em) greater than J.

1
A

Fig. 2. Console dimensions diagram.


A - Maximum console height (see Table 2)
B - Vertical dimension of panel (see Table 2)
C - Console panel angle --------------------- 15°
D - Minimum writing surface depth ----------- 16 in. (41 em)
E Minimum knee clearance ------------------ 18 in. (46 em)
F - Minimum thigh clearance ----------------- 8t in. (22 em)
G - Writing surface height (see Table 2)
H - Leg height clearance (see Table 2)
I Foot clearance -------------------------- 4 in. (10 em)
J - Seat height (adjustable)
ERGONOMIC DATA FOR CONSOLE DESIGN 155

measured from "the seat back to the front of the knees" (Ayoub and
Halcomb, 1976, p. B-55).

Recommendations for console dimensions may also be found in a


number of human factors design guides. Two such guides are Van Cott
and Kinkade (1972) and Woodson (1981).

CONCLUDING REMARKS

In addition to the data discussed above, there are other data


(i.e., data concerning human capabilities and limitations, work
surfaces, seating, selection of displays and controls for specific
applications, cognitive functioning and decision/making, effects of
workload on operator performance, and software) that console designers
will find to be useful. Some of this information appears in other
papers in this volume.

Design methods are also important, but they are beyond the
scope of the topics to be discussed at this NATO Advanced Research
Institute. Nevertheless, designers should be familiar with at least
some of the powerful design tools that are available. These tools
include procedures for determining panel layouts (e.g., link analysis
and linear programming), computer-aided design, mathematical model-:-
ing, and mock-up construction and evaluation.

When using ergonomic data, the designer must pay special atten-
tion to the conditions under which the data were obtained and apply
correction factors if needed. It was pointed out earlier that factors
such as body dimensions, hand orientation, seat characteristics, type
of restraint, and type of control have a significant effect on reach
envelopes. Application of reach data without making allowances for
these variables may lead to an unsatisfactory design. The amount of
effort that should be spent in modifying existing data so that they
may be used for new applications will depend upon the criticalness
of tasks that the operators perform and the consequences of operator
errors.

In the future, ergonomic investigations concerning console de-


sign should be directed more toward determining ranges of accept-
ability and the consequences of deviating from established recom-
mendations. Such data will help designers make better trade-offs.
Ergonomists should also continue to develop methods for modifying
existing data so that it will be more useful for new applications."
Mathematical modeling and computer-aided design will, no doubt, play
an important role in this effort.
156 W. H. CUSHMAN

REFERENCES

Anonymous, 1981, Human Engineering Design Criteria for Military Sy-


stems, Equipment and Facilities, MIL-STD-1472 c.
Arora, S., 1976, Effect of magnitude, direction of hand-movement
and plane angle on reach time, Proceedings of the Ninth An-
nual Meeting, Human Factors Association of Canada, 40.
Ayoub, M. M., 1973, Work place design and posture, Human Factors,
15:265.
Ayoub, M. M., and Halcomb, C. G., 1976, Improved seat, console, and
workplace design: Annotated bibliography, integration of the
literature, accommodation model, and seated operator reach
profiles, Pacific Missile Test Center, TP-76-1.
Barnes, R. M., 1958, "Motion and Time Study," Wiley, New York.
Chaffee, J. W., and Emanuel, A., 1964, Maximum arm reach on instru-
ment racks from the seated position, Boeing Company Report
D2-90549.
Das, B., and Grady, R. M., 1981, Comparative analysis of the normal
working area in the horizontal plane, Proceedings of the
25th Annual Meeting of the Human Factors Society, 762.
Dempster, W. T., 1955, Space requirements of the seated operator,
WADC-TR-55-159.
Dempster, W. T., Gabel, W. C., and Felts, W. J. L., 1959, The an-
thropometry of the manual work space for the seated subject,
Am. J. Physical Anthropology, 17:289.
Ely, J. H., Thomson, R. M., and Orlansky, J., 1956, Layout of work-
spaces, in: "Joint Services Human Engineering Guide to Equip-
ment Design," WADC-TR-56-171.
Farley, R. R., 1955, Some principles of methods and motion study
as used in development work, General Motors Engineering
Journal, 2 :20.
Grandjean, E., 1981, "Fitting the Task to the Man," Taylor & Francis,
London.
Kennedy, K. W., 1978, Reach capability of men and women: A three-
dimensional analysis, AMRL-TR-77-50.
Kennedy, K. W., 1964, Reach capability of the USAF population:
Phase I. The outer boundaries of grasping-reach envelopes for
the shirt-sleeved, seated operator, AMRL-TDR-64-59.
Kennedy, K. W., and Bates, Jr., C., 1965, Development of design
standards for ground support consoles, AMRL-TR-65-163.
Kobrick, J. L., 1965, Effects of physical location of visual stimuli
on intentional response time, J. Eng. Psychol., 4:1.
Lehmann, G., and Stier, F., 1961, Mensch und Gerat, in: Handbuch der
gesamten Arbeitsmedizin, Vol. I," E. Baader, ed., Urban und
Schwarzenberg, Berlin.
Morgan, C. T., Cook, J. S., Chapanis, A., and Lund, M. w., 1963,
"Human Engineering Guide to Equipment Design," McGraw-Hill,
New York (N. Y.).
ERGONOMIC DATA FOR CONSOLE DESIGN 157

Roth, J. T., Ayoub, M. M., and Halcomb, C. G., 1977, Seating, con-
sole and workplace design: Seated operator reach profiles,
Proceedings of the 21st Annual Meeting of the Human Factors
Society, 83.
Sharp, E. D., and Hornseth, J. P., 1965, The effects of control lo-
cation upon performance time for knob, toggle switch, and
push button, AMRLrTR-65-41.
Siegel, A. I., and Brown, F. R., 1958, An experimental study of
control console design, Ergonomics, 1:251.
Squires, P. C., 1956, The shape of the normal work area, U.S. Navy
Medical Research Laboratory, Report No. 275, New London
(Conn. ) .
Stoudt, H. W., 1978, Arm-leg reach and work,space layout, in: "Anthro-
pometric Source Book," Vol. I, Webb Associates, ed~ NASA
Reference Publication, 1024.
VanCott, H. P., and Kinkade, R. G., eds., 1972, "Human Engineering
Guide to Equipment Design," U.S. Government Printing Office,
Washington, D. C.
Webb Associates, eds., 1978, "Anthropometric Source Book," Vol.I;
"Anthropometry for Designers," Vol. II; "Handbook of Anthro-
pometric Data," Vol. III; "Annotated Bibliography of Anthro-
pometry," NASA Reference Publication, 1024.
Woodson, W. E., 1981, "Human Factors Design Handbook," McGraw-Hill,
New York.
ERGONOMIC DATA FOR THE DESIGN OF BODY SUPPORT

Heinz Schmidtke

Institut fur Ergonomie

Technische Universitat Munchen, MUnchen/FRG

INTRODUCTION

In the preceding papers on console design it has been shown


that optimal conditions for the operator depend upon certain dimen-
sions of the consoles and on seat characteristics. Only by taking
into consideration both aspects, i.e. shape and dimensions of con-
soles, work tables, etc., and shape dimensions of body support for
the seated operator, will it be possible to optimize the whole work
system. It is furthermore necessary, under certain circumstances,
to guard against vibration and acceleration forces which are like-
ly to be transmitted from the environment or the technical system
through the body support used by an operator.

In highly sophisticated technical systems (e.g. modern mili-


tary aircrafts), ,body support is regarded as an important part of
the system and therefore adaptable not only to different tasks per-
formed by an operator, but also to a large number of the operator
population. In many cases however, the adapt ion of the seat to its
user is rather poor. Because this misadaptation may adversely af-
fect performance as well as health, it seems prudent to apply er-
gonomic knowlege on a large scale in the design process of seats,
particularly in view of the fact that the sum total of all types
of body support constitute the largest class of tools used in in-
dustry, traffic, administration, and in private homes.

REFERENCE POINTS ON SEATS

Amongstthe biggest problems encountered in the process of seat


design lies in the definition of a reference point from which the
designer can calculate:

159
160 H. SCHMIDTKE

- the position of the eye point or the eye ellipse (in relation to
the field of vision),

- the reach envelope according to thP. enveloping surface (in rela-


tion to the position of controls) ,

- the position of the heel point (in relation to the position of


foot-operated controis), and

the positioning of fasten~ng elements for safety belts in dynamic


systems.

As a matter of fact, the problem of reference points, has to


date not been adequately sol~ed. The reason for this lies not so
much in the lack of a reliable reference point as with a plurality
of reference points generally found in literature. The following
reference points have occassionally been raised:

- the Hip-Point (H-Point) according to SAE J 826b, ISO 5353, DIN


33 408, part 1;

- the Seating Reference Point (SgRP), according to SAE J 941 e;

- the Reference-Point (R-Point), according to ISO 6549;

- the Seat Index Point (SIP), according to ISO 5353, ISO 6682, and
SAE J 1163;

- the Seat Reference Point (SRP), according to ISO 3462, DIN 1857,
part 2, and ISO 4253, DIN 1857, part 1, DIN 24 091 E, DIN 1858;

- the Seat Reference Point (SRP) or Neutral Seat Reference Point


(NSRP), according to MIL-STD-1472b and HEL-STD-S-6-66.

Apart from the poor definitions of a number of the aforemention-


ed reference points (e.g. HEL-STD-S-6-66: "Seat reference point is
point where seat back and seat cushion intersect"), a major diffi-
culty in their application may lie in the fact that the aforesaid
reference points are mere theoretical points not to be found in an
actual seat. In order to be able to apply these reference points,
a prototype seat (mockup) has to be manufactured and the measuring
machines adapted to the prototype. This approach makes it possible
to ascertain the extent to which the resulting hardware corresponds
with the original intentions of the designer. At present measuring
machines have been developed for the Hip-point (Fig. 1), the Seat
Index Point (Fig. 2) and the Seat Reference Point in accordance
with ISO 3462 (Fig. 3). In applying a reference point during the
design process, a designer requires drafting tools (e.g. body tem-
plates) which are in close accordance with the measuring machines
mentioned above. The H-point is a reference system found in the
ERGONOMIC DATA FOR THE DESIGN OF BODY SUPPORT 161

Fig. 1. Three-dimensional H-Point-Machine (SAE)

Fig. 2. Device for the determination of the Seat Index Point (SIP)

Fig. 3. pevice for the determination of the Seat Reference Point


(SRP)
162 H. SCHMIDTKE

SAE- and DIN 33 408-drafting template and on the three-dimensional


H-point-measuring machine. The position of the H-point is identical
in the SAE- and DIN-drafting template for the 50 th percentile male
(measuring machines see Fig. 1 - 3).

For practical purposes it is fairly complicated to operate the


H-point-measuring machine. The SIP-measuring machine, on the other
hand, lends itself towards easier operation. In view of the afore-
mentioned, namely that both the H-point and the SIP are theoretical
points which cannot be located in a real seat, it would be conceiv-
able to use the SIP-machine provided that the SIP is located close-
ly enough to the H-point marked on the drafting templates. An ex-
periment performed by Ruhmann (1982) of our Institute, using 15 dif-
ferent seats, possibly demonstrates that the SIP can be localized in
a vertical plane up to about 6 mm above - and 10 mm in front of the
H-point (Fig. 4). Although the seats varied greatly, both in design
and in upholstery, the variation between the H-point and SIP was
found to be rather small. We are consequently of the opinion that in
using the SAE- or DIN-drafting templates for design work, the SIP-
machine should be applied for the ha~dware. The relation between H-
point, SIP and SRP (according to ISO 3462) are shown in Fig. 5.

ERGONOMIC DATA FOR SEATS ATTACHED TO THE FLOOR

Floor-fixed seats are primarily used for all conceivable types


of consoles as well as in dynamic systems. In adapting the seats
both to the requirements of the task and to varying body sizes of
the users, several principles of design have to be taken into con-
sideration:

The adjusting range in the horizontal and vertical olane must be


suitable for operators of the 5 th and 95th percentile. Further-
more, if, in addition to male operators, cognizance is taken of
female operators, the lower critical limit should be in accordance
with the anthropometric data of the 5 th percentile female.

- The minimum vertical adjustment of the seat should ensure that it


is possible to adjust the eye point of the 5 th percentile operator
(male and/or female), sitting in a relaxed position, to a height
of 1210 mm above floor level.

- Floor-fixed seats should be adjustable parallel to the median


plane of the operator to such a degree that male and/or female
operators from the 5th to the 95th percentile are able to reach
the controls comfortably.

- In order to provide for relaxed sitting positions, both in control


operations as well as in monitoring tasks, an adjustment of seat
inclination is desirable.
ERGONOMIC DATA FOR THE DESIGN OF BODY SUPPORT 163

• Small lateral upholstery or broad seat and backrest between the lateral
upholstery, hard Isoll upholstery
• Marked lateral upholstery, hard upholstery

4_----10mm -----t
10 mm tolerance area according to ISO 3462

·BMW- Test -seat I


BMW 7er (L" ·BMW 7er (5)
MAN - truck Peugot 604
(1)
·Cltroen CX
I
.. I 10mm

..
KM-LokO: .. +.cBMW 5er .. :MW 3er (SF,W)


Volvo 244 Ford Granado
BMW 200, I
BMW- Test - seat lift-truck
(2) Sable
"tAudi100


--tH-~kt
Direction of movement
I
Fig. 4. Location of the Seat Index Points of 15 different automobile
seats relative to the H-Point

II
99 104

~-+-" SRP and zone


~---121 ----I of deviation
---131---

Fig. 5. Experimental data of location of SIP and SRP relative to


the H-Point
164 H. SCHMIDTKE

- It should be possible to adapt the front part of the floor-fixed


seats in order to reach underneath the work plate of the consoles.
In such a case the seat is within easy reach only if it is pos-
sible to turn it around the shaft.

- It has to be ensured that all controls regulating seat adjustment


and -rotation, are situated within easy reach and mounted in such
a position as to prevent the entanglement of wearing apparel and
accessories. It has to be taken into consideration that under
certain circumstances, adjustment controls have to be used with
protective gloves.

- Seat and back rests shouid, in relation to the functional aim of


the technical system, be upholstered in such a way, to ensure that
the vibrations are sufficiently damped. In a few cases it might
even be necessary to add an additional damping system to the sub-
structure of the seat.

- The natural frequency of the seat-occupant-system should be so de-


signed that an amplification of input acceleration force, parti-
cularly in the range of the principle human natural frequency (ap-
proximately 3 to 6 Hz) would occur neither vertically nor horizon-
tally. In addition, the angular seat resonance (back-slap, peak
at 11 Hz with a relatively large standard deviation due to occu-
pants posture) should be minimized.

- Seat and back rest should have a cover providing certain lateral
guiding, with a sufficient water vapour permeability and an ability
to absorb water vapour.

- Arm rests should be upward-folding and latched in this position.

In considering these principles the designer needs standard


data based on ergonomic requirements. These data are the following
(see Fig. 6 and Fig. 7):

a} Size of seat
The depth of the seat should be between 430 and 450 mm, the width
between 400 and 460 mtl.

b} Size of back-rest
The minimal height of the back-rest should be 500 mm above the
seat. The clearance between seat and the lower flange of the
back-rest should not exeed 150 mm. The width of the back-rest
should be adapted to the width of the seat, with a minimum width
of 400 mm and, under certain circumstances, devices for neck-
rests are necessary.

c} Back-rest inclination
It is conceivable that out of economic reasons a design without
ERGONOMIC DATA FOR THE DESIGN OF BODY SUPPORT 165

Fig. 6. Data for the design of seats attached to the floor (side
view)

Fig. 7. Data for the design of seats attached to the floor (front
view)
166 H, SCHMIDTKE

an adjustable back-rest inclination may be in demand. In this


case the inclination of the back-rest should be fixed at an angle
of 95 0 to 100 0 in relation to the surface of the unloaded seat.

d) Seat adjustment
In order to meet the requirement of adjusting the eye point of
the 5 th percentile operator to a point not less than 1210 rom
above floor, the height of the seat should be made adjustable,
i.e. for male and/or female operators from 450 rom to 530 rom
(= 80 rom).

An effe·ctive contact point in hand and foot operated controls for


the 5 th and 95th percentile man can be achieved if the'seat is ad-
justable in the longitudinal plane, going out from a neutral po-
sition:

- for male operator: ±60 rom (120 rom); and

- for male and/or female operator: ±75 rom (150 rom).

e) Shape of seat and back-rest


In order to facilitate more effective absorption of lateral forces
in dynamic systems, the upholstery of the seat should be folded
upwards in the last quarter of the back position of the seat, the
fold to measure approximately 35 rom along the back edge but taper-
ing towards zero at the front edge. At the back-rest 250 rom above
seat, the value for lateral guiding should be approximately 40 rom
whilst going towards zero at the upper and lower edge. The back-
rest should be vaulted in breadth and in length in order to opti-
mize body contact with the back-rest.

f) Arm-rest
Floor-fixed seats, in general, should be fitted with upward-fold-
ing armrrests. The height of arm-rests above seat should be 190 rom
to 230 rom. Armrrests will be accepted by users if their surface
dimension is about 50 rom multiplied by 300 rom (maximum 350 rom).
If the inclination of the back-rest is adjustable, it should be
possible to keep the arm-rest constantly in a horizontal posi-
tion.

g) Isolation from random vibration in dynamic systems


Because user comfort depends to a large extent on his isolation
from annoying vibrations, the damping characteristics of loaded
seat should ensure the avoidance of any kind of amplification of
input acceleration forces, at least in the range from 3 to about
11 Hz (3 - 6 Hz human response mode, approximately 11 Hz seat
"back-slap"-mode). Fig. 8 and Fig. 9 show the transfer function
of two different seats in z:
ERGONOMIC DATA FOR THE DESIGN OF BODY SUPPORT 167

'r~y

- SlhlOt on pr or 0 racesogn

tJ1It ckJ~ seat

Fig. 8. Transfer function of a standard seat in a cross-country


vehicle (vibration amplification in the critial frequency
range i.e., the torso natural frequency)

Z.O 3.0 5.0 10 20 30 50


frequency

- sltuohon pnor to redesign


-<>- sotuahon oller redesign

VhW.! Improvement compored


ruu.:I with standard seat

~ Impo"meot compnred
~ with stondord soot

Fig. 9. Redesign of the standard seat (the transfer function of the


new seat shows a, remarkable improvement of the' damping power
in the critical frequency range)
168 H. SCHMIDTKE

amplification of acceleration forces in the critical range from


1 - 6 Hz and from 9 - 15 Hz;
- a newly-designed seat without any amplification in the critical
range but with a considerable damping characteristic between 1
and 20 Hz.
Because seat damping depends upon expected input forces, an eva-
luation of the damping systems is possible only empirically by
measuring the transfer function of the seat loaded with an user
or a dummy.

As important as damping of seats is the damping of vibration of


the seat back. Angular resonance, caused by a vibration mode at
the hinge pOint where the seat back folds forward and backward,
(Varterasian and Thompson, 1917) reaches its main peak at 11 Hz
with a considerable standard deviation. In this respect it is
remarkable to note that light-weight user will receive heavier
back-slap blows, as he occupies a greater mass.

ERGONOMIC DATA FOR WORK-CHAIRS

Within the category of work-chairs fallsa broad variety of


seats, ranging from a simple stool to a very comfortable easy-chair.
Despite chairs sometimes being regarded as status symbols, ergono-
mists can only be expected to attend to those design aspects which
are correlated to the functionality of man. Because tasks may re-
quire a sitting posture using, on the one hand, only the front part
of the chair (e.g. work on a assembly line), or, on the other hand,
the whole chair (e.g. office work in a vertical position" of the
trunk or inspection work where the operator can lean back), these
differences have to be taken into account together with different
anthropometric data of men.
The following ergonomic principles have to be applied to chair
design:
- The adjusting range of height should be oriented on the Sitting
height of the 95th percentile male and the 5 th percentile female.

- Work-chairs generally require a chair-back. The minimal dimen-


sion of the chair-back has to be adapted to the whole lumbar re-
gion. If sitting postures are expected whereby the front- and the
back part of the chair are alternatively used, the chair-back
should either be able to swiveled or be adjustable on the longi-
tudinal axis of the chair.

- The seat of a work-chair must have a shape that allows slight


variations of sitting postures. Lateral profiling of the seat has
to be avoided.
ERGONOMIC DATA FOR THE DESIGN OF BODY SUPPORT 169

- The inclination of the seat has to be matched with the preponderant


sitting posture. If during work a forward sitting posture is to be
expected, the seat should be adjustable in the horizontal plane or
inclined slightly forward. Independent from the angle of the slope,
the. front edge of seat has to be rounded off in order to minimize
surface pressure at the thigh.

- In order to guarantee good stability the design of work-chair should


be of such a nature as to safeguard against a topping over caused by
a shifting of the center of the body mass. Roller-brakes are re-
quired if work-chairs are supplied with swivelling roller.

- Controls for seat-adjustment should be within easy reach.

- Arm-rests may be appropriate for work-chairs utilized for office-


oriented tasks.

In order to transform these general principles into hardware,


the following should be considered:

a) Size of seat
The depth of the usable part of the seat should not exceed 400 mm
lest small female operators encounter difficulties in making pro-
per contact with the chair-back. The width is irrelevant in so
far that the value will not remain under 400 Mm.

b) Size of chair-back
If the chair-back is rigidly mounted the ~n1mum height should be
450 mm and the width 350 Mm. The vertical shape of the chair-back
. should be concavely dished, particularly in the lumbar region.
>~ '!he height of the main lumbar support should be between 80 and
100 mm above the point of contact with the seat.

c) Chair i~clination
It should be possible to adjust the seat of work-chairs used for
different sitting postures between - 2° and + 5° in correspondence
to the horizontal plane (see Fig. 10).

d) Height adjustment th
A work-chair is usable both for the 5 percentile female as well
as for the 95 th percentile male, provided the adjustment range
goes from 530 mm (upper position of seat) to 380 mm (lower posi-
tion). Regarding work on assembly lines, it may be necessary to
consider the real height of the work bench. This may be achieved
in the following way:

- Work-chair in upper position


530 mm + difference height work bench (mm) - 750 mm
- Work-chair in lower position
380 mm + difference height work bench (mm) - 750 mm
170 H. SCHMIDTKE

-'.. 0°
c:rI
A '\

\ s~
l'
B _5°

Fig. 10. Chair inclination


A: normal seating position
B: backward inclined seating position
C: forward inclined seating position

Example: Height work bench 860 mm.


530 mm + (860 mm - 750 mm) = 640 mm upper position.
380 mm + (860 mm - 750 mm) = 490 mm lower position.
Foot-rests should be available.

ERGONOMIC DATA FOR STANDING-SUPPORT (SUPPORT STOOL) (PERCH)

Under standing-supports we include those aids which enable the


operator, working in a standing posture, to relieve the load to the
muscles of the legs and the back. As a matter of fact, standing-sup-
ports should not be considered as aids in compensation of poorly de-
signed work places originally planned to be operated in sitting
postures. Standing-supports can, at least partially, reduce the
disadvantages of a standing posture (e.g. skeletal injuries or
vascular diseases) in such cases where work has to be performed under
high forces (e.g. filing) or in large work pieces. But standing-sup-
ports are only suitable for use if they have a high degree of tilt-
resistance and provided they prevent the buttocks from slipping off
the seat as a result of the transmission of high muscular forces.

The following ergonomic priniples have to be considered in the


design of standing-supports:

Standing-supports in the longitudinal and lateral axis should be


of such a nature as to prevent a tilting over either to the side
or to the back even when it becomes necessary to shift the center
of the trunk or to transfer the higher muscular forces to a work
piece.
ERGONOMIC DATA FOR THE DESIGN OF BODY SUPPORT 171

B·B
F====+-L....L...£ 2. ) '

{~, 8
0\ . 0\
)50

Fig. 11. Data for the design of a standing-support (perch)

- Standing-supports should enable an adjustment of the seat height


in at least three different up- or down-positions.

- The guiding ways or the shaft for the vertical adjustment of the
seat should be designed in such a manner as to prevent injury to
the back of the operator as well as the entanglement of the op-
erator's clothing.

- The application of a standing-support requires a non-slip floor


preventing both the support and the feet from slipping.

- A locking device for both using and non-using positions is re-


quired if standing supports are connected through a revolving
mechanism to the working place instead of them being built as
mobile units. In the non-using position the support should not
interfere with the operator.

On the basis of the principles enunciated above the following


standard data are available (see Fig. 11):
172 H.SCHMIDTKE

a) Height adjustment
A standing-support may be used by both a tall man and a small
woman provided the range of seat adjustment goes from 700 rom
above floor to 850 mm.

b) Stability
Sufficient tip over-stability will be achieved if the front and
back contact surfaces of the substructures has a width of not
less than 450 rom and a distance in the longitudinal axis of about
500 rom. The contact surface with the floor should be coated with
antislip material.

c) Seat design
A standing-support should have a saddle-shaped bucket seat with
a back-width not less than 350 rnm and a minimal depth of 250 rnm.

d) Seat inclination
By using a standing-support both feet will be in contact with
the floor. In order to minimize body pressure the seat should be
tilted forward in progressive steps ranging from 0° to - 5° and
to - 10°.

ERGONOMIC DATA FOR FOOT-RESTS

Foot-rests as devices for body support have to be assigned to


the work place under the following conditions:

- if, in relation to the seat height, the feet fail to make suf-
ficient contact with the floor;

- if, in the case of out-stretched legs, an excessive flection of


the ankle joint is sought to be prevented;

- if, in dynamic systems, the lower extremities can support the


body stabilization by the influence of acceleration forces.

In most cases foot-rests achieve to reduce surface pressure to


the legs, particularly in the area of the front edge of seat. Only
by applying foot-rests can smaller people make full use of the
height adjustment of seats. This however constitutes a requirement
for establishing an optimal view as well as an optimal grasp enve-
lope at work.

In the design of foot-rests the following principles have to


be applied:

- Foot-rests should be adjustable in the vertical as well as in the


horizontal plane. The adjustment range should meet the needs of
the 5 th percentile male and/or female;
ERGONOMIC DATA FOR THE DESIGN OF BODY SUPPORT 173

foot-rests should have a tread sufficient in size to support the


whole footi

- foot-rests should be manufactured by material with low thermal


conductivity in order to prevent heat absorption from the feeti

- foot-rests for use with more or' less out-stretched legs should
have an adjustable angle of incidencei

- in order to prevent a loss of body stability by gliding off with


the feet, the leg support should be coated with anti-slip-materi-
ali

- the tread should be sufficiently wide to allow for a change in


the leg position.

These principles can be realized by applying the following


data (see Fig. 12):

a) Height adjustment
If the clearance underneath the table board is about 700 mm, a
5 th percentile female will reach the average height of the eye
point (1210 mm) only by adjusting the seat height to 520 mm.
Because the length of the lower legs (the length of the shoes
including) of the 5th percentile female is about 395 mm, a foot-
rest of at least 125 mm height is needed. A 5 th percentile male
will reach the same sitting conditions by a 30 mm height of the
foot-rest. It may be concluded that the height adjustment should
range from close to 0 up to 125 mm.

b) Adjustment in the horizontal plane


There will be no need for an adjustment in the horizontal plane
if the minimum depth of the foot-rest is 330 mm and the position
of the ,front edge of the foot-rest is located 150 mm behind the
vertical plane going through the front edge of the table. It is
advisable to have an angle of inclination for the foot-rest of
about 10° to 25°.

c) Width of foot-rests
A width of the foot-rest, equivalent to the width of leg room,
should be aimed for. 400 mm to 500 mm should be the minimum.

d) Adjustment mechanism
In the design of the adjustment mechanism for height and incli-
nation, consideration should be given to the possibility of the
operator having undergone a foot operation.

e) Design of chair-fixed foot-rests


In the case of work places requiring (relatively) high chairs as
alternatives to floor-fixed or bench-fixed foot-rests, it is
174 H. SCHMIDTKE

50

Fig. 12. Data for the design of foot rests

possible to connect the foot-rest to the chair. This can be done


either by means of a ring shaped (Fig. 13) or, preferably, by a
sector-shaped foot-rest (Fig. 14). For both types a height ad-
justment is necessary ranging from 380 mm to 520 mm below the
upper part of front edge of seat. The front edge of foot-rests
should exceed the front edge of the seat at least 50 mm. If this
type of chair is not floor-fixed mounted, the required degree of
stability is such as to make it impossible for the chair to over-
turn even by standing on the foot-rest with trunk bent forward
as far as possible without hold. This position may be achieved
if the dimension of tilt resistance meets the condition 0,4 D
- a > O. If for instance the distance from shaft-axle to front
edge of seat is 220 mm and the distance shaft-axle to front edge
foot-rest in the median plane is 270 mm (= a), the critical limit-
ing value 0,4 D - a = 0 for the diameter (D) of the circle en-
closing the subconstruction of the chair will be 675 mm. Suffi-
cient tilt resistance may be achieved if the diameter is >675 mm.
ERGONOMIC DATA FOR THE DESIGN OF BODY SUPPORT 175

Fig. 13. Chair-fixed and ring-shaped foot rest

ERGONOMIC DATA FOR NECK-RESTS

Neck-rests in dynamic systems are passive safety devices aimed


toward reducing the risk of injury to the body resulting from im-
pact. Without a neck-rest there will be a relative motion between
trunk and head due to innertia stresses which, in their turn, would
result in sheer forces on the cervical vertebra. In order to reduce
this type of highly dangerous injury to the body,occuring mainly
in road accidents, neck-rests should not only be available, but also
carefully adapted to the individual needs of the user. This can be
done by applying the following principles:

- Neck-rests will act as supporting elements of the head in case of


high acceleration forces in the longitudinal plane only if they
are rigidly linked to the steel structure of the seat.

- Neck-rests should be adjustable to the height of the mass center


of the head of the user population.

- The material of neck-rests should be highly energy-absorbent.

- Neck-rests should have a vertical contour avoiding contact with


the head during normal sitting posture, however simultaneously
minimizing free head motion brought about by acceleration forces.
176 H. SCHMIDTKE

-0 0

Fig. 14. Chair-fixed and sector-shaped foot'rest

The neck-rest design should not impair the vision in all direc-
tions.

For realization of these principles several ergonomic data are


available:

a) Height of neck-rest
For the neck-rest to be completely integrated into the back-rest
of the seat, the horizontal plane through upper edge of the back-
rest should be at least 775 mID above H-point (approximately 875 mID
above unloaded seat). This will meet the requirements of the 95 th
percentile male. For the purpose of safety aspects however, it is
necessary to take the anthropometric data of the 99 th percentile
male into acount. For the aforesaid reason the value of 775 mID
above H-point has to be extended to a minimum of 800 mID.
ERGONOMIC DATA FOR THE DESIGN OF BODY SUPPORT 177

Fig. 15. Profile of a neck-rest

Where a neck-rest, which is adjustable in height, is separated


from the back-rest of the seat, it should be noted, in order to
prevent a deflection of the cervical spine in an accident, that
the distance between the upper edge of the back-rest and the lower
edge of the neck-rest should not exeed 100 mm. The adjustment
range, related to the horizontal plane through the upper edge of
the neck-rest, should vary from 660 mm (5 th percentile female) to
775 mm (95 th percentile male) above H-point. The values for the
1st percentile female are 640 mm and for the 99 th percentile male
880 mm. If there is no continuous height adjustment the distance
between the catch points should not exeed 20 mm. The guide block
for height adjustment should be mounted in the back-rest in such
a manner that a variation of distance from head to neck-rest may
be prevented.

b) Width of neck-rest
The neck-rest should have a minimum width of 200 mm in order to
prevent a head-slide off if the application of acceleration forces
is transversal. In addition, slide-off effects may be reduced by
shape (see Fig. 15).

c) Vertical contour of neck-rest


The vertical contour of neck-rest should be adapted to the normal
sitting posture (inclination head against trunk 10° - 15°). The
requirement of an acceptable distance (d) head to neck-rest will
be met by 25 mm < d < 75 mm.

d) Upholstery
In order to reduce head acceleration the structure of neck-rest
should be built up, the inner part with rigid expanded plastics
or steel-deformation elements, and the surface padded with soft
upholstery.
178 H. SCHMIDTKE

REFERENCES

Ruhmann, H. P., 1982, Vergleichende Darstellung und Messung von


Sitzbezugspunkten an Fahrzeugsitzen, z. Arb.wiss., 36:41.
Schmidtke, H., ed., 1981, "Lehrbuch der Ergonomie," Hanser, Miinchen.
Schmidtke, H., 1981, Arbeitsplatzgestaltung, in: "Lehrbuch der Er-
gonomie," H. Schmidtke, ed., Hanser, MUnchen.
Schmidtke, H., and Ruhmann, H. P., 1978, "Ergonomische Gestaltung
von Steuerstanden," Forschungsbericht Nr. 191 der Bundesan-
stalt fur Arbeitsschutz und Unfallforschung, Wirtschaftsver-
lag NW, Bremerhaven.
Varterasian, J. H., 1982, On measuring automobile seat ride comfort,
Proceedings Internat. Automotive Engineering Congress,
Detroit.
Varterasian, J. H., and Thompson, R. R., 1977, Teh dynamic charac-
teristics of automobile seats with human occupants, Pro-
ceedings Internat. Automotive Engineering Congress, Detroit.
THE DESIGN AND LOCATION OF CONTROLS:

A BRIEF REVIEW AND AN INTRODUCTION TO NEW PROBLEMS

Walter W. Wierwille

Virginia Polytechnic Institute and State University

Blacksburg, Virginia / USA

INTRODUCTION

This paper is intended to provide a brief review of conventional


ergonomic control design and placement, and to indicate future di-
rections that control research and design may take. A control is de-
fined here as a device used by the hands or feet (limbs) of an
operator to apply signals or control inputs to a system. Controls
represent the primary means of commanding most systems. There are
some exceptions of course, such as voice-commanded systems.

Discussion will be limited to controls that require a degree


of prec1s10n as opposed to force or power. Manual devices used for
applying appreciable forces should be treated as a separate topic,
with emphasis on work physiological and biomechanical aspects. In
this paper the emphasis is on controls in which the forces required
of the operator are small and there is a need to get the correct in-
put into the machine or system.

When the literature on controls is examined, it will be found


that numerous design guidelines are available. These guidelines have
been developed by pioneering human factors researchers and designers.
To be sure, there is so~ research literature, but not as much as
might be expected. When problems were encountered that did not seem
to have a ready solution, a brief research study might then have
been conducted. As a result the design guidelines are to a great
extent based on experience.

This paper will also present two new areas associated with
control design. The purpose of including these areas is to indicate
how little is known about important particular classes of controls

179
180 W. W. WIERWILLE

and to show what future trends may take place in control design.
The problems are typical of those that may be encountered, but they
are by no means the only ones that will be encountered.

REVIEW OF CONVENTIONAL CONTROL DESIGN AND LOCATION

There are several textbooks which provide excellent overviews


for the design controls, determination of their relative placement,
and their required associations with displays. To mention a few
recent ones, the books by Van Cott and Kinkade (1972), McCormick
(1976), Kraiss and Moraal (1976), Hutchingson (1981), Schmidtke
(1981), and Woodson (1981) provide good summaries of what is current-
ly known. In this section the basic principles will be described so
that proper insight can be developed.

General Principles

One of the most important principles is that the control must


fit the requirement. For example, if continuous adjustment is re-
quired, a designer should use a continuously adjustable control. On
the other hand, if there are ~ alternatives in the requirement, the
control itself should have ~ discrete positions.

Actually there are four different types of controls in common


use, each with a distinct purpose: 1) activate-deactivate (switches),
2) discrete setting (multiple position devices), 3) quantitative-
continuous, and 4) data entry. The data entry devices usually con-
sist of a bank or matrix of depressable switches, but are classified
separately because of their distinct purpose.

In regard to quantitative-continuous controls, control display


ratio becomes important. Usually a tradeoff exists in selecting the
correct ratio. If the control has high gain, slew rate will be rapid,
but fine adjustment time may be long. On the other hand, low gain
can produce slow slewing rates, but with short fine adjustment times.
The correct value of gain depends on the specific application and
the relative importance of slew rate into the neighborhood of the
desired value and time to reach the desired value with a certain
tolerance, once the neighborhood is reached.

Also in regard to continuous controls, it is sometimes necessary


to have both a precision setting and a wide range of settings. This
requirement is usually best met by selecting multi-turn rotary con-
trols such as handwheels. If the slew rate must also be high, a
"crank" handle should be included with the handwheel.

Another very important principle in control design is that of


control-display compatibility. In general, an attempt should be made
to keep the direction of movement of the control the same as the
THE DESIGN AND LOCATION OF CONTROLS 181

direction of movement of the corresponding display. This implies


that if possible, they should move in the same plane. This principle,
if it can be applied in a given application, will minimize or elimi-
nate "control reversals" (movement of the control in the wrong ini-
tial direction). There are of course many situations where it is un-
desirable to keep the control and display movement in the same plane.
These situations occur because displays are usually in the vertical
plane, whereas from an anthropometric point of view, the correspond-
ing control creates stress, since it would project outward from a
vertical panel and be moved either horizontally or vertically.

When control-display compatibility cannot be achieved by using


the same plane and directions for the control and corresponding dis-
play, then the correct population stereotype should be employed. For
example, for a display pointer moving to the right, a corresponding
rotary control is best located below the display and should move
clockwise. population stereotypes of a wide variety of controls have
been developed and tested. These can be found in the textbooks
mentioned previously. It is very important to avoid conflict with
an existing stereotype.
It is also important to distribute controls among the limbs so
that no single limb (the right" hand, for example) is overloaded. If
one limb is overloaded, that limb must time-share between controls,
or between axes, thereby increasing average response time to dis-
play stimuli.

Once the limb to be used to operate a control has been selected,


the control must be tailored and located to fit that limb. The size
of the control, its shape, and the directions and limits of move-
ment must be carefully designed to match the limb. Otherwise, pro-
blems may arise due to operator fatigue and joint stress as well as
due to imprecise setting and activation.

There are several other criteria for locating controls, and


to some extent, they conflict with one another. First, it is de-
sirable to have controls located near the corresponding displays.
This tends to eliminate the problem of determining which control is
associated with which display. However, there may be good reasons
for not co-locating controls and displays. They include: 1) display
position for optimum viewing is usually not near the control posi-
tion for optimum anthropometric fit and activation by the given
limb; 2) visual blockage of portions of a display panel may occur;
and 3) controls may be ganged and therefore associated with more
than one display.

When controls and displays cannot be co-located, there are


four criteria that can be used in locating them. They are the same
four as are used for instrument arrangements: frequency of use,
importance, sequence of use, and functional grouping. Usually, if
182 W. W. WIERWILLE

there is a group of displays, control locations and design are dic-


tated by the same principle. For example, if all the engine in-
struments are grouped together (that is, by function), then the cor-
responding engine controls should also be grouped together, even
though they may not be co-located with the displays. In any case,
the controls must be fitted and located according to the limb used.

In regard to control arrangement, one technique that has been


used successfully in industrial situations is to mimic the display
(instrument) arrangement in the control arrangement. For example,
if the displays are in two rows of three each, the related controls
should be in two rows of three each. Such a procedure aids the op-
erator in associating each control with its corresponding display.
When the arrangement cannot be mimicked for one reason or another,
a standard logical progression of controls should be used.

Controls must be easily identified. There are several methods


for accomplishing this, each with its own advantages and disad-
vantages. Controls may be shape-coded for example. Shape coding has
the advantage that the control can be identified without illumination,
.but i t has the disadvantage of being ineffective when a combination
low illumination and heavy clothing (gloves, boots) must be used.
Labeling has the advantage of allowing identification without me-
morization, but it requires illumination and the ability to read
the language in which the labels are written. Color coding can some-
times be used for identification, providing the number of controls
is less then perhaps ten and they are illuminated. However, color
blindness and memorization can cause problems. Finally, controls
may be coded by operational direction. For example, in a vehicle,
movement of a lever to the right might cause the vehicle to move
to the right. Usually this technique works well as long as the
number of axes to be controlled is small. On the other hand, there
may be a learning period involved in using controls that are op-
erationally coded.

Regardless of the application, the control itself should have


the proper reaction to movement. A multi-position switch, for example,
should contain a detent, so that is snaps or clicks between positions.
This prevents setting the control between defined positions and i t
also aids in avoiding inadvertent changes in setting due to vibra-
tion or accidental contact by the operator. Continuous controls
should also possess proper reactive responses. In particular they
should exhibit frictional resistance to movement or spring center-
ing, so that they remain in a set, predictable position when un-
attended.

In some cases, special precautions must be taken to avoid in-


advertent actuations. This can be accomplished by one or more of
the following techniques: locating the control out of the normal
THE DESIGN AND LOCATION OF CONTROLS 183

workspace of the operator, requiring a dual axis motion for acti-


vating, heavy detenting, or providing a physical barrier around the
control.

Manual Control

Manual control involves a situation in which the relationship


of the displays and controls is not simply one to one. To illustrate,
consider a situation in which there is a single continuous control
and a corresponding display. A movement of the control causes pro-
portional movement of the display, which is nearly instantaneous.
In other words, the control and display are related by a gain (con-
trol/display ratio). There are many situations, however, in which
the relationship between a control and a display is much mane complex
than a gain. For example, in process control, a change in a control
setting affects the balance of the process, which eventually results
in changes exhibited on the display. In other words, "dynamics" are
introduced between the control input and the display output.

Books (e.g. Kelley, 1968; Frost, 1972) and hundreds of papers


have been written on manual control system design. Manual control
is a sophisticated subject that cannot be reviewed adequately in a
few paragraphs. Many controls are being used in closed-loop manual
control applications. In fact, if an operator observes a display and
uses the information to adjust a control, the system is then closed-
loop and is actually a manual control system. (To give an idea of
the ubiquity of manual control systems, one need only observe that
all vehicles that are not automatically guided are manual control
systems.) The matn point to be made here is that controls, partic-
ularly continuous controls, cannot be examined in isolation. They
must be examined in the context of the closed-loop operator-machine
system.

In manual control systems, specific types of controls have


been developed. These controls are distinguishable not so much by
their anthropometric shape as by their reactive characteristics.
In the next few paragraphs, these controls will be reviewed.

Two pure extremes in controls used for manual control systems


are the isometric and isotonic controls. An isometric control pro-
duces an output signal that is proportional to the force applied.
A pure isometric control ("stiff stick") does not deflect however.
If there is more than one axis to the control, then there is an
output signal for each axis, proportional to the force applied
along (or about) that axis. An isometric control has the inherent
characteristic of returning the output signal to zero whenever there
is no force applied. Usually, therefore, it is used with dynamics
containing integration. If one integration is used, the system out-
put eventually reaches a constant position after the control is
released.
184 W. W. WIERWILLE

The control representing the other extreme is the isotonic


control. Theoretically, this control provides no resistance to move-
ment, and its output is proportional to displacement. If this con-
trol has more than one axis, the output signal in each axis is pro-
portional to the displacement along or about that axis. When re-
leased, this control maintains its position and produces a constant
output. It therefore has position memory.

Most controls used in manual control applications fall between


the two idealized extremes. One of the most widely used types of
control is the spring centered joystick. Usually this type of con-
trol produces an output proportional to deflection. But, since de-
flection is nearly proportional to the force applied, the output is
then also nearly proportional to the force.

Another important class of controls falling between the extremes


is that in which the control maintains a set position when released
by the operator. This class includes track-ball controls, hand-
wheels, many types of knob controls, and many types of lever controls.
Actually most of the quantitative-continuous devices described earlier
fall in this class. These controls generally maintain a fixed posi-
tion by means of frictional resistance. They are actually physical
realizations of isotonic controls. The output is proportional to de-
flection, the same as isotonic controls.

Control reaction in manual control systems may be extremely


important. Control reaction may be as simple as the frictional re-
sistance of the control itself or as complex as the force feedback
from the control surfaces of an aircraft. Auto enthusiasts speak of
"steering-wheel feel" which to an appreciable extent involves the
reflected forces that appear in the steering wheel. These forces
are a result of the applied forces, the vehicle and steering response
characteristics, and the wind and road conditions.

Control reaction is dependent on the ph~/sical design of the


control system. In a "fly-by-wire" system, there is no mechanical
connection between the control and the dynamic system being con-
trolled. Instead, electrical, pneumatic, or hydraulic signals are
taken from the control and used as inputs to the system being con-
trolled. Control reaction under these conditions is largely a re-
sult of the control mechanism itself. To regain "control-feel",
fly-by-wire systems are often designed so that reactive forces from
the process being controlled are fed back and actively applied to
the control. This type of system is called a bilateral system,
because there are two interactive systems operating in opposite
directions. One takes signals from the control and applies them to
the system. The other takes signals from the system and applies them
to the control.
THE DESIGN AND LOCATION OF CONTROLS 185

In some cases conventional (positive) manual control systems,


which have a mechanical connection between the control and the sys-
tem, already have adequate mechanical power and reactive feel. If
they do not have adequate mechanical power, they can be power assist-
ed. Unless precautions are taken, however, power assist will usually
reduce or eliminate reactive forces from the system. Therefore, to
maintain this feel, the design of the power assist must be sub-
stantially more sophisticated.

Both fly-by-wire and power assisted systems must be designed


for fail-safe operation, if they are used in real vehicles or other
systems that present a hazard if control is lost. In fly-by-wire
systems, the usual method of providing fail-safe operation is through
back-up or redundant systems. For power assisted systems, the same
technique can be used, or the system can be designed so that it re-
verts to a positive mechanical system if it fails.

In concluding this brief review of controls, it should be


emphasized that to an extent, control design should be considered
an iterative process. A designer should test any control selected
for a given function. In fact, it is probably best to try more than
one type or style of control. Most design information has been ob-
tained through design experience, and this experience may be mis-
leading in specific circumstances.

There are several techniques that can be used to test and


compare specific controls. First of all, performance can be compared
using objective measures in controlled tests. Care must be taken in
conducting such tests to avoid biases and to,obtain statistically
reliable results. Similarly, operator preferences can be obtained
through detailed attribute rating scales. However, once again, care
must be taken to avoid bias and to obtain statistical reliability.
And finally, anthropometric fit can be checked. One of the best
methods for accomplishing this is through EMG (electromyogram) anal-
ysis. Gartner (1981), for example, has shown that EMG's can be used
effectively to detect muscular-skeletal strain, that is, strain
caused by forcing the skeletal structure into unnatural positions.
This type of strain is difficult to eliminate in controls having
multiple axes and in controls having substantial excursions.

NEW PROBLEM AREAS

The purpose of this section is to direct attention toward some


of the unresolved problems appearing in control design. More spe-
cifically, two major areas will be described, both of which are a
result of advancing technology. No doubt there are other important
unresolved problems as well, but the purpose here is to stimulate
thinking in new directions - not to provide an exhaustive treatise.
186 W. W. WIERWILLE

Bang-bang Controls

There is a class of controls that does not fall neatly into


any of the previous categories. These controls are the so-called
bang-bang controls. They are ordinarily used in manual control sys-
tems that involve positioning or pursuit velocity matching.

A simple example of a bang-bang control is one in which two


momentary-contact push buttons are placed next to each other. De-
pressing the right pushbutton causes a corresponding display pointer
to move to the right at constant speed. Depressing the left push-
button causes the pointer to move to the left at constant speed.
When the pushbuttons are released, or when both are depressed, the
pointer remains stationary.

Bang-bang manual controls can be found in numerous, important


applications:

a. Rapid traverse positioning of machine tools

b. Crane dynamics

c. Construction equipment with hydraulic controls

d. CRT cursor positioning

e. Pitch trim, flap extension, and speed brake controls on jet


transport aircraft

f. Spacecraft attitude control systems.

In general, these systems are characterized by the need to slew


rapidly from one position to another, and then to be set precisely
thereafter. The desired slew rate and the desired position accuracy
vary from system to system.

Often, their dynamics are complex and cannot be represented by


a simple step input to an integrator. Mass and friction may cause
an appreciable speed build-up characteristic when the system manual
control ist first depressed (or deflected), and a coast d9wn when
the control is released. Therefore, the human operator must perform
an estimation process in guiding the output to the desired final
value. In some cases several inputs to the control must be made,
often in opposing directions. The operator in some cases uses the
reverse direction as braking.

The only existing literature on these systems appears to be an


early paper by Pew (1966), in which the dynamics were not varied.
Because there was so little literature, we decided to conduct a pre-
liminary investigation of them at Virginia Polytechnic Institute
THE DESIGN AND LOCATION OF CONTROLS 187

and State University. William F. Beatty carried out the experiment.


It was designed to determine optimum regions of design parameters.
The experiment will be reviewed briefly here.

The dynamics used in each axis of this two-axis study were as-
sumed to have the form

where Fist the constant force applied when one of the control but-
tons is depressed, x is the position of the output, M is the equi-
valent mass, and D is the equivalent viscous friction of the system.
The dynamics can be rewritten in transfer function form as

Xes) F/D
U(s) s
s (1 + D/M)

where U(s) ! for a step input, F/D is the system gain and D/M is
the system co?ner frequency in radians per second.

~e control device used was a set of four pushbuttons on the


points of a square. The square's corners were oriented so that they
appeared as a "diamond" to the operator. To make the display spot
move upward, the operator depressed the top pushbuttoni to make it
move down, he/she depressed the bottom pushbutton. Similarly, the
right pushbutton caused movement to the right, and the left pushbut-
ton caused movement to the left.

The dynamiCS were simulated on an EAI-S80 hybrid computer and


the output spot was displayed in x-y coordinates on a 21 inch
(S3.3 em) oscilloscope. Scoring and timing were also performed on
the hybrid computer. The dependent variable used in this experiment
was the integrated product of mean square position error multiplied
by time. Thus, the longer it took to reach the final value and the
greater the excursion in reaching the final value, the larger was
the score. Small scores were then most desirable.

Thirty six subjects (18 males and 18 females) participated in


the experiment. Each received one of nine sets of dynamics (settings
of F/D and D/M). Each subject's task was to position the spot so
that it resided at the center (target area) of the oscilloscope
screen. The spot was initially placed at one of eight locations
equidistant from the center of the screen. The visual subtense of
the target area about the center was 2.4°, and the visual subtense
of the distance from the starting positions to the center was 28.6°.
The subject received a warning that a trial was about to begin when
the spot moved to the starting position. The beginning of a trial
was signaled by means of an audio tone. At that time the subject
moved the spot to the target area as quickly as possible. Once the
188 W. W. WIERWILLE

spot was within the target area for 1 second, the trial ended. Each
subject completed sixty-four trials, the first thirty-two of which
were practice.

The results of the experiment were analyzed using a central-


composite design and response surface analysis (Clark and Williges,
1973). They are plotted in Figure 1. As the plot shows there are two
optimal design regions, designated by the numerals I and II. In re-
gion I the system gain, F/D, is high and the system corner frequency
DIM is low. It was observed that in some cases, subjects overshot
the target area, but were able to return to it quickly because of
high gain. In region II, the system gain is low and the corner fre-
quency is high. In this region, the system dynamics are acting much
the same as an integrator. Even so, there is a clear-cut optimum
gain, which could not have been predicted without experimentation.
The remainder of the plot shows that increased values of the cri-
terion measure occur when pairs outside regions I and II are used.

In general the results of this experiment suggest that impor-


tant, practical problems in control design and manual control para-
meter selection have not been adequately examined. Moreover, clear-
cut optima appear to exist. Therefore, if a human factors engineer
is willing to take the time to perform experiments, helshe is likely
to be able to optimize the parameters of the design for the specific
circumstances and measures selected. In control design and in manual
control, the majority of our results have evolved through what might
be termed "design practice" and simplified laboratory experiments.
It is suggested here that designs developed from this background
should be tested and optimized.

This initial experiment also points up some of the additional


problems that remain with bang-bang controls. Does display subtense
affect the optimal parameter settings? Would substitution of a three-
position joystick change the settings? How would the resul~be chang-
ed if a digital (numerical) readout was used in place of an analog
display? These questions can only be answered by further studies.

Multi-function Controls

The rapid advances in technology during the last two decades


have forced new burdens on operators of these systems. Systems now
perform more functions than they did earlier, and the functions are
themselves more complex. To an extent, operators in systems are
taking on more of a monitoring and supervising role, rather than an
active manual role. However, they must always be able to take over
an automated system and control it manually if necessary.

Because systems are becoming more complex, the controls that


operators use are also becoming more complex. When spacecraft were
first developed, controls that could be used for three rotational
THE DESIGN AND LOCAnON OF CONTROLS 189

8
OJ

CREATBR TMA" 12S0


6

,/0
~ • 10~1200

)
D. AS U IUCID

:z ) S 6 7 8 9
D/II
Fig. 1. Scores shown by regions of the response surface

axes had to be developed. In addition, methods for inputing trans-


lational commands had to be developed as well.

In more recent aerospace applications, multiple functions are


being added to a single control. This process began with the intro-
duction of a trigger switch on a two-axis joystick. Since then it
has become common practice to include "a few extra" functions on
joystick controls. Woodson (1981) describes a "six-function joystick".
The siX switch functions are in addition to the primary purpose of
the stick itself, which is two-axis continuous control. Woodson
states, "Six-functions are probably the maximum number that should
be considered for a single controller; more functions make it diffi-
cult for pilots to keep from actuating the wrong,one, either because
they forget which switch is which or because of inadvertent motion
inputs ••• " (page 594).

Most recently, the F-18 aircraft has been designed with two
controls, one for each hand. The control for the left hand has the
primary purpose of controlling thrust, and the control for the right
hand has the primary purpose of pitch and roll control.
190 W. W. WIERWILLE

The F-18 represents a distinct change in the philosophy of air-


craft design. North (1977) stated this change in philosophy as fol-
lows:

The F-18 has been designed as a complete head-up air-


craft. The head-up display presentation as the primary
flight instrument gives all the parameters needed for
flight and weapons delivery. This, combined with the
hands-on-throttle-and-stick concept, allows the pilot
to select and fire any of the weapons or to control the
radar and change its modes in the combat arena while
keeping his eyes on the target (page 64).

In other words the concept, insofar as controls are concerned,


is to eliminate the need for visual reference to the controls. The
pilot operates the controls by feel and by relative location within
each hand.

To give an idea of how complex this philosophy becomes in prac-


tice, the left hand control has been drawn for the F-18 with the
various auxiliary control devices designated (Figure 2). First,
the control itself is actually two slide controls one controlling
the throttle of each engine. They can be operated together or se-
parately, and can be locked in any position. In addition there are
four auxiliary control functions operated hy the left thumb, two
more are operated by the left index finger, and five more are op-
erated by the remaining three fingers of the left hand. As the figure
shows, the types of controls used for the auxiliary functions vary.
Momentary pushbuttons, slide switches, a rocker switch, a toggle
switch, a two-axis isometric control, and a bi-directional isotonic
control are included.

The right hand control for the F-18 is similar to that de-
scribed by Woodson (1981). The control itself (Figure 3) is a two-
axis joystick for pitch and roll control of the aircraft. It con-
tains seven auxiliary control functions, three of which are per-
formed by the right thumb, another by the thumb and index finger,
another by the index finger alone, and still two others by the re-
maining fingers. Again, the types of controls used for the auxil-
iary functions vary.

A great deal of time, effort, and talent has gone into the de-
sign, development, and testing of these controls. Clearly, they are
effective and appear to work efficiently. While there may be small
revisions in the design as more experience is gained, the design
philosophy is likely to continue and to become more widespread.

Multifunction controls such as those of the F-18 aircraft have


ramifications for the general problem of control design. The new
THE DESIGN AND LOCATION OF CONTROLS 191

Three-position slide switch,


momentary, center hold

Three-position Right throttle control


rocker switch

Left throttle control

Momentary
pushbutton

Momentary
pushbutton
Three-position slide
switch, momentary in Momentary
one direction, pushbutton
hold in center and
in other direction,
(Bang-bang)

"'",--B'-:::~~~"1
' \ =,"01

FRONT
(away from pilot)

Finger locks for


Two-axis throttle controls
isometric control,
combined with
momentary switch
Fig. 2. Left hand (throttle) control of the F-18 aircraft

philosophy compromises several well-known design principles as out-


lined earlier in this paper, including the following:

1. Prevention of inadvertent control actuation,

2. Provision for display-control compatibility,

3. Provision for positive identification (coding) of controls,

4. Provision for even distribution of controls over the limbs,


and
192 W. W. WIERWILLE

Four momentary-position
center-off pushbutton

Four-position
momentary switch
(Two-axis bang bang)
Momentary
Pushbutton

Three-position toggle
switch and pushbutton

Two-axis
control stick

Two-detent

lever
switch
Momentary
pushbutton

Fig. 3. Right hand (stick) control of the F-18 aircraft

5. Provision for grouping controls according to a principle,


such as, by sequence of use.

While not all of the above are absolutely discarded, they are
to an extent subjugated to a new design philosophy.

What are the new principles that come into being along with
this new design philosophy? They appear to include the following:

1. Design of the controls so that the human operator need not


observe them when critical visual demands are present (i.e.,
the need for eye-hand coordination is eliminated),

2. Design of each primary control so that the hand (or foot)


remains in contact with the primary control throughout
cricital operations of the system, and
THE DESIGN AND LOCATION OF CONTROLS 193

3. Addition of auxilliary controls to the primary control for


the hand in such a way that they can be actuated by the
thumb and fingers without loss of contact of the hand itself
with the primary control.

The purpose of this discussion is not to criticize the design


or use of multifunction controls. Rather, the purpose is to point
out that a subtantially different set of guidelines accompany multi-
function controls, and that these guidelines are not fully reconcil-
able with existing guidelines for conventional controls.

It is suggested that research on multifunction controls should


be conducted. Since they are presently being used and will, no doubt,
continue to be used, the more we know about them, the better. The
kinds of research to be undertaken should include:

1. Determination of the maximum number of usable functions and


best method for accessing or call-ups,

2. Development of an alternate identification scheme (other


than operator memory),

3. Determination of the best method for eliminating inadvertent


actuations,

4. Coordination of CRT display formats with control functions,


and

5. Determination of whether separate keypads might provide an


alternative to multiple~function controls.

Finally, it is suggested that whenever a multifunction control


is used in a system, it must be thoroughly tested in that system.
Because so little is presently known about them and because the guide-
lines associated with them are tentative, comprehensive testing is
mandatory.

CONCLUDING REMARKS

This paper has two main conclusions that can be succinctly


stated. First, design of controls is based heavily on design experi-
ence and some supporting laboratory experiments. As a result, a new
design should always be tested experimentally before it is placed
in operation. Second, there are several important existing control
design problems that have received little or no research emphasis.
Among them are the bang-bang control problem and the multifunction
control problem. Both will become more important as system complexity
increases.
194 W. W. WIERWILLE

ACKNOWLEDGEMENTS

The author acknowledges the helpful suggestions of Mr. John G.


Casali of Virginia Polytechnic Institute and State University,
Dr. Frank Gomer of McDonnel-Douglas Astronautics Company, and Dr. Sam
Schiflett of the Naval Air Test Center.

REFERENCES

Clark,C.,and Williges, R. C., 1973, Response surface methodology


central composite design modifications for human performance
research, Human Factors, 15:295.
Frost, G., 1972, Man-Machine dynamics, i~ "Human Engineering Guide
to Equipment Design,' H. P. VanCott and R. G. Kinkade, eds.,
Government Printing Office, Washington, D. C.
Gartner, K. P., 1981, Electromyography and applications, in: "Manned
Sys tems Des ign: Methods, Equipment, Applications," J. Moraal
and K. F. Kraiss, eds., Plenum, New York (N. Y.).
Hutchingson, R. D., 1981, "New Horizons for Human Factors in Design,"
McGraw-Hill, New York (N. Y.).
Kelley, C. R., 1968, "Manual and Automatic Control," Wiley, New York
(N. Y.).
Kraiss, K. F., and Moraal, J., eds., 1976, "Introduction to Human
Engineering," Ttlv Rheinland, Kc5ln.
McCormick, E. J., 1976, "Human Factors in Engineering and Design,"
McGraw-Hill, New York (N. Y.).
North,D. M., 1977, Single-plot effectiveness of F-18 tested in si-
mulator, Aviation Week and Space Technology, 106:64.
Pew, R. W., 1966, Performance of human operators in a three-state
relay control system with velocity-augmented displays, IEEE
Transactions on Human Factors in Electronics, 77:83.
Schmidtke, H., ed., 1981, "Lehrbuch der Ergonomie," Hanser, Miinchen.
VanCott, H. P., and Kinkade, R. G., eds., 1972, "Human Engineering
Guide to Equipment Design," U.S. Government Printing Office,
Washington, D. C.
Woodson, W. E., 1981, "Human Factors Design Handbook," McGraw-Hill,
New York (N. Y.).
THE HAND-MACHINE INTERFACE

Colin G. Drury

Department of Industrial Engineering


State University of New York at Buffalo, New York/USA

ABSTRACT

Three types of interface between the human hand and the world
of machinery are distinguished. They are handles for moving objects,
handrails for steadying the human and controls for transmitting in-
formation to a machine. For each type of human/machine interface the
shape, size and texture of the handle need to be considered. The
literature is reviewed on handles, handrails and hand controls, giv-
ing consistent recommendations on interfaces which fit the hand.
Handle position and angle are considered for two types of tasks,
manual materials handling and push/pull tasks. Both laboratory and
field studies show that handle positions on boxes should encourage a
'diagonally opposite' grip which has both horizontal and vertical
stability. Handle angle should be such as to minimize radial and
ulnar deviations of the wrist. For pushing and pulling tasks the
handle should be 900 - 1100 mm above ground level and foot obstruc-
tions should be avoided to allow free walking while pushing and
pulling.

INTRODUCTION

Where the hand meets the machines and objects of the worla of
work is an obvious, if complex, inferface. We have many words for
these hand/machine interactions: touching, holding, pushing, grasp-
ing, pulling, steadying. The machine part of the interface can be of
three types:

1. A handle for moving an object.


2. A handrail for steadying the operator.
3. A control for transmitting information to a machine.

195
196 C.G.DRURY

These interfaces must be designed with the human body in mind,


not just the human hand, although this is an important considera-
tion. For example, the placing of push-button controls depends on
both the size of the finger, to ensure that two controls are not
activated at once, and on the visual/kinaesthetic coordination of
the operator so that standards of aiming accuracy can be achieved.
Many recommendations exist on small controls e.g. Grandjean (1980),
Van Cott and Kinkade (1971). Typically we have the data in Table 1.

For larger controls, handrails and handles, we usually require


the whole hand to take part in the gripping action. In this case we
must take into account the type of force being exerted by the hand
on the handle and the type of grip used. Napier (1956) classified
handgripping postures into three types:

1. A hook grip in which the fingers are flexed around the


object and the thumb is not used for gripping.

2. A power grip in which the object is clamped between the


partly flexed fingers and palm with the thumb opposing the
grip and lying along the plane of the palm, and

3. A precision grip in which the object is pinched between the


flexor aspects of the fingers and opposing thumb.

It can be seen that the conflicting requirements of moving,


steadying and controlling can demand different gripping postures.
Generally in moving an object we are concerned with both power and
precision; in steadying the body, power is most important and in
moving a control, precision is required. However, for moving an ob-
ject most handles tend to be ill-designed.

A small survey of manufacturers of handles for luggage and por-


table equipment showed that the main factors in handle design are
visual appearance and cost. Further evidence comes from Woodson
(1971) who reports that off-the-shelf handles appear to be "designed
as decorative appointments" rather than "designed to fit the hand".
He reports insufficient hand clearance, sharp cutting edges and too
small a handle diameter. Most handles, hand-holds or gripping aids
on containers force the worker to use a hook grip (the least effect-
ive) or a power grip. This latter gives a good gripping force and
allows a large surface area of hand to be used but it is inefficient
if accurate control of the container is needed. Frequently, however,
the weight of a container will not allow a precision grip to be
used. A hook grip may be the only one used in practice but many of
the handle design studies in the laboratory have allowed power grips.

Figure 1 shows a classification of the forces which are applied


to handles. The force not shown on the figure is grip force, which
is usually exerted on a handle in order to maintain the control ne-
:tm
:t
>
Z
C
I
s:
>
n
:t
Recommendations for Small Controls
Z
m
Table 1.
~
m
Diameter Length or Force or Spacing :u
Height Torque ~
Min. Opt.
n
m
Push Buttons 12 - 15 mm 10 N 20 mm 50 mm

Toggle SWitches 3 - 25 mm 12 - 50 mm 2.5 - 15 N 25 mm 50 mm

Rotary Knobs (clocks top) 35 - 75 mm 20 - 50 mm <320 Ncm 25 mm 50 mm

Rotary Knobs (continuous) 10 - 30 mm 15 - 25 mm 3.5 - 50 Ncm 25 mm 50 mm

Bar Knobs <25 x 25 mm 12 - 70 mm 13 - 18 Ncm 25 mm 50 mm

....
~
198 C.G.DRURY

Ulnar/Radial
Deviation Torque
Thrust \ \
Force

Fig. 1. Torques and Forces Applied to a Handle

cessary for the production of other forces and torques. The names of
the torques have been chosen on the basis of the action necessary at
the wrist to exert them; the actual muscles used to exert all torques
and forces will of course vary with the nature of the task.

HANDLE/CONTROL DESIGN

There is a reasonable level of agreement in the literature on


shape, size and texture of handles, although this is hardly reflec-
ted in the handle market place!

HANDLE SIZE

The width of a handle or handhold should accommodate at least


95 % of the population. Standard design guides (e.g., the U.S. Army
guide, HEDGE, 1974) and the anthropometry of the hand (Garret, 1971)
suggest a minimum width of 115 mm with 30 - 50 mm clearance all
around to accommodate the fingers and knuckles. Nielson (1978) has
similar recommendations. If use with gloves is anticipated, at least
25 mm should be added to these dimensions.

It is instructive to examine handles in use and see how they


compare with these recommendations. Figure 2 shows a set of hand
saws, collected over the last fifty years, all of which have approx-
imately the same size. Clearly, handles have evolved to meet human
needs. In contrast Figure 3 shows two pans in a set. The handle size
THE HAND-MACHINE INTERFACE 199

Fig. 2. Hand Saws of Different Sizes With Similar Handle Sizes

is scaled to the pan size, presumably for visual continuity, rather


than remaining constant in size to suit the hand which must use the
pan.

HANDLE DIAMETER

Pheasant and O'Neill (1975) measured flexion/extension torques


and found the larger handle the better, at least up to 70 mm dia-
meter, although maximum shear force at the handle surface was great-
est for a handle diameter of 30 - 50 mm. Thrust forces were found to
peak at about 40 mm diameter.

The pronation/supination torque was measured as a function of


handle diameter by Saran (1973) who found that a 25 mm diameter
handle was preferred over either a 19 or 32 mm handle. There were
no differences between handle diameters in terms of electromyogram
(EMG) measures of the muscle groups involved in the task.

Tasks requiring the production of a push/pull force (always


pull in practice) have been used to evaluate handle diameter in a
number of studies. Ayoub and LoPresti (1971) found a relatively flat
optimum between about 25 mm and 64 mm diameter when EMG was mea-
sured. However grip forces were optimum for a diameter of 38 mm.
200 C.G.DRURY

Fig. 3. Pans With Handles of Different Sizes

Khalil (1973) measured EMG activity for three diameters of cylindri-


cal handle, (32, 50 and 70 mm) plus an elliptical handle 50 mm long
x 32 mm wide and a 50 mm diameter sphere. Of all these handles, the
32 mm diameter cylindrical handle was best.

Other recommendations can be made based on different criteria.


If the hand is to fit the handle with any overlap of fingers and
thumb, then Garret's (1971) anthropometric data would suggest 41 mm
as a maximum diameter for a 5th percentile male without gloves. Si-
milarly, guidebook recommendations in human engineering recommend
diameters as follows (quoted from Rigby, 1973; also in the U.S. Army
guide, HEDGE, 1974).

Weight of item Minimum diameter


lb kg mm

15 6.8 6
15-20 6.8- 9.1 13
20-40 9.1-18.2 19
40 18.2 25

These values are quoted without evidence as to their efficacy.

Woodson and Connover (1964) recommend diameters for hand con-


trols such as T-bars and joysticks in the range 12.7 mm to 28.6 mm
with 44 mm maximum diameter for handrails. Damon et al. (1966) re-
commend maximum handle diameters not to exceed 38 mm.

Two studies at SUNY at Buffalo, reported in Drury (1980) found


optimum handle diameters of 20 mm and 31 - 38 mm for manual lifting
tasks. Handrail diameter preferences by handicapped and elderly pub-
THE HAND-MACHINE INTERFACE 201

lic transport users were measured by Brook, Ruffell-Smith and Ward


(1974). They found an optimum diameter of 32 mm for all groups of
subjects. Steinfeld, Czaja and Beer (1981) measured the force with
which people could exert on a handrail, the thrust force in Fi-
gure. 1. They used wood or steel bars of diameters 19 to 51 mm and
five groups of subjects: wheelchair users with and without hand/
reach problems, ambulant/semi-ambulant users with and without hand/
reach problems and finally able-bodied users. All groups showed
optimum diameters between 25 and 38 mm for wood and between 32 and
38 mm for steel.

The overall pattern of these studies is somewhat confusing, but


a handle diameter range of 25 to 40 mm would cover most of the opti-
ma found in the studies. Where a hand must grip a split handle, such
as a pair of pliers, the optimum distance between the grips for ma-
ximum grip strength has been measured. Greenberg and Chaffin (1979)
show that for males and females of all sizes a grip distance of 3"
to 3 1/2" (76 to 89 mm) is optimal. Hertzberg (1955) in a similar
study for pilots found 2 1/2" (64 mm) better than either 1 1/2
(38 mm) or 4" (102 mm) both with and without gloves. Gloved gripping
forces were about 20 % below bare hand forces.

HANDLE SHAPE

Small deviations from a cylindrical handle of constant diameter


need not have a large effect on handle performance. Khalil (1973)
used an elliptical handle of 41 x 31 mm cross section and a spheri-
cal handle of 51 mm diameter and found little difference between in-
tegrated EMG on these two compared with a cylindrical handle of 51 mm
diameter. Pheasant and O'Neill (1975) compared the 13 commerical-
ly available screwdriver handles with smooth and knurled cylinders
of the same diameter. Once the effect of handle diameter was elimi-
nated, there.was no difference between the screwdriver handles and
equivalent knurled cylinders. Tichauer (1973) points out that pro-
vision of finger grooves to give for~fitting handles may not be a
good idea. The finger spacing of each person is different and any
design compromise on 'average' spacing will be a poor fit for many
of the population. Rubarth (1928), in studying shapes of screwdriver
handles, found a cylindrical handle with a rounded end better, by
10 to 20 % in maximum force, than more elaborately-shaped handgrips.

More drastic changes in handle shape have been studied by


Bdbbert (1960) and Nielson (1978). Bobbert made recommendations for
hand grips to be moulded into concrete building blocks, using an-
thropometric, physiological and force production criteria. A trape-
zoidal cross section varying in width from 42 to 65 mm was finally
chosen on the basis of his tests and practical difficulties of cast-
ing concrete. Nielson compared a wide variety of handles for their
suitability in industrial tray design. He recommends handhold cutout
202 C.G.DRURY
or drawer-pull handles of generous size at both ends of the tray for
carrying and low lifts and gripping blocks 19 mm thick and 51 - 76 mm
deep if the tray must be lifted high or over a wide range. This pre-
vents the hand from being locked into the handle, giving extreme ulnar
deviation of the wrist when the load is lifted to high levels.

Figure 4 shows an interesting handle shape on a home power saw.


The handle is tapered, causing it to slip out of the hand when car-
ried. The obvious behavior by the user is to insert the first finger
into the trigger guard to try to prevent this occurrence. This ob-
vious action unfortunately switches on the saw when it is lifted
suddenly! A safety problem may be nothing more than a handle prob-
lem.

HANDLE SURFACE TEXTURE

As expected, most authors recommend a non-slip texture for the


handle surface (Rigby, 1973; Pheasant and O'Neill, 1975; Nielson,
1978) and Pheasant and O'Neill give force production data to support
this recommendation. Perhaps more important is the elimination of
sharp edges, corners, seams or excessive ribbing (Nielson, 1978). A
pilot study at the State University of New York at Buffalo failed to
show significant differences in voluntary holding time between
smooth, padded and sand paper textured handles. Surface texture may
not be as important a variable in handle design as size and shape.
It should be noted that a non-slip texture may also abraid the skin
of the hands and inhibit adjustment of hand position.

As a final example of handles ill-designed for the human hand,


Figure 5 shows a tray designed to be moved by automated equipment.
The moulded handles are designed to be a perfect fit to the "hands"
of the automated equipment but the human operator who must load and
unload the machine was not given the same consideration in design.

HANDLE POSITIONS

The position of the hand-machine interface will be determined


by the task to be performed. We would expect different optimum posi-
tions in at least four types of task:

1. Manual materials handling


2. Pushing/pulling vehicles
3. Handrails for architectural and vehicle use
4. Controls for seated and standing operators.

Only the first two of these will be considered as handrails and


controls must be integrated into overall design of workspaces, mak-
ing any recommendations highly task-specific.
THE HAND-MACHINE INTERFACE 203

Fig. 4. Tapered Metal Handle on Portable Saw

Fig. 5. Tray Designed for Automated Equipment, Not for Human Hands
204 C.G.DRURY

For manual materials handling, an industrial survey was used to


collect data on hand positions used by materials handlers on box-
like objects (Drury, Law and Pawenski - in press). Considerable data
was recorded for each of over 2000 box movements on 27 subjects in
nine industries ranging from parcel delivery, through food warehous-
ing to chemical products manufacturing. Data was recorded for the
subject, for the box and for the task on the form shown as Figure 6.
Of most concern here is the data on handle positions during the task.

The type of task was expected to be straightforward, based on


carrying one or two objects in one or two hands. However, on some
occasions the task type changed during the task, for example from
two-handed to one-handed movement of an object. This fact was also
recorded and coded.

The task itself was seen as having five stages:

1. Pre-grasp - where the object is brought into position for


picking up, but the weight of the object is not yet wholly
supported by the subject.

2. Pick-up - where the subject first takes .the full weight of


the object.

3. Move/carry - where the subject supported the full weight of


the object, moving it from the initial location to the final
location.

4. Put-down - where the weight of the object is at least parti-


ally, if not totally, relinquished.

5. Adjust - where the object is moved (usually slid) into the


final position.

As Pre-grasp height was expected to be the same as Pick-up


height and Put-down height to be the same as Adjust height, only
three stages of the task (START, DURING, STOP) were used to record
the height of the bottom of the object from floor level. These same
three points, between which the object weight is wholly supported by
the subject, were used to determine how far the subject had to reach
over to control the object and whether the subject was twisting or
in a sagitally symmetric condition.

OVer the whole set of 2038 box handling movements, the great
majority (1837 or 90.1 %) were two handed throughout. In a further
105 movements (5.2 %) a change was made from two handed to one hand-
ed or vice versa. The remaining five percent of cases were almost
equally composed of one object handled in one hand or two objects
handled one in each hand. For this two-handed movement data there
were 81 possible pairs of hand positions but only about two-thirds
THE HAND-MACHINE INTERFACE 205

STATE UNIVERSITY OF NEW YORK AT BUFFALO. DEPARTMENT OF nmUSTRIAL ENGINEERING 1980

I MANUAL MATERIALS HANDLING SURVEY I


SUBJECT

Organization Code ~ Subject Code ~ Special _ _ _ _ __


Subject Stature W i n . Subject weight ChN lbs.
OBJECT
..l1.......lL 1516 1718
Height U-J in. Length CIJ in. Width in. CD
Weight ~ lbs. Special _ _ _ _ _ _ _ _ _ _ _ _ _ _ ~

HANDLES
23 24 25
Handle positions DOD Special _ _ _ _ _ _ _ _ _ _ _-1

Handle types -O-g-Q


R TASK

1 Object/2 hands _ 1 object/l hand _


1 2 3
4 5 6 2 Objects/2 hands_ O'ther 0
31
7 8 9
START DURING STOP carry distance ~ ft.

Bottom Height [~]ft. (j ft. LJ ft. number of corners [j


[j ft. [j ft. L] ft.
a
obstacles?

0
Reach Over?

Twist L or R?
38
Oft. ft. L! ft. [jl:j
3
6
2
5
1
4 > LH RH Against
Tia
Top/

=a =a
Stage Posn? Elbow Posn? ElbOlol'
9 8 7
0 0 BodeY
_a
Pre-Grasp'
- _CJ
0 _[j [j -Cj
_a _a
~ick Up

1
~ 2 3
t'lDve/Carry

lPut Down
[j
t) _a LJ
0
_15
_0 _a
_L!
_0
4 5
8
6
~djust [j _0 [j _0
n
_0 _0
7 9

DODD
77 78 79 80
1 2 3
456
789

~
Fig. 6. Manual Materials Handling Survey
206 C.G.DRURY

of these were ever recorded. The great majority of hand position


pairs recorded were only of a relative few of these 81 possibili-
ties. For example, at move/carry one position accounted for over 40 %
of the results, two positions for over 50 %, ten positions for over
90 % and fifteen positions for over 95 % of results. Manual material.
handlers apparently use only a small variety of hand positions. To
analyze this more carefully, the seventeen most frequent hand posi-
tion pairs were tabulated at each stage of the movement. For each
pair both left- and right-handed versions were counted together so
that position 3/7 (the most common at all stages) includes both:

'Left hand in position 3/right hand in position 7' and


'Right hand in position 3/1eft hand in position 7'.

Figure 7 shows the cumulative percentages of each of these


seventeen frequent hand positions at each stage of the movement. The
two positions 3/7 and 1/9 which use hands on diagonally opposite
corners of the box account for about half of all the data at all
stages except for adjust. An adjust stage was only present on 15.1 %
of movements and here different hand positions applied. Now almost
half (47 %) of all movements had both hands on the back edges of the
box, positions 1/7, 1/1 and 3/4. This was usually the only edge
available when putting the box into tight proximity to other boxes.
At all stages, the 'symmetrical on bottom edge' position, 8/8, was
within the five most frequent hand positions, accounting for appro-
ximately 10 % of movements.

Although Figure 7 describes the overall hand-position data, it


is possible to relate the hand position behavior to other variables
of the subject, the box and the task. For example, is 3/7 favored by
taller subjects or is position 4/4 only used for light boxes? The
survey studied natural behavior and thus significant inferential
statistics here indicate relationships but not necessarily causa-
tion. With this in mind, the simplest method possible was used to
relate each hand position to other variables. Each variable, such as
subject height or box weight, was tabulated and split at as near to
the median as was possible with such highly quantized data. The fre-
quencies of movements with each hand position were counted as above
or below the median on each variable. Chi-square tests were then
performed to test whether each variable was related to each hand
position. Tests were performed for all move/carry hand positions,
with a frequency of over 20, to all for safe use of the Chi-square
statistic. BLE in Table 3 is the Biomechanical Lifting Equivalent,
the product of box weight and distance of box center of gravity from
the lumbar spine.

The results are shown at two levels of significance for subject


and box variables in Table 2 and for task variables in Table 3. For
each relationship, the direction is shown by the entry. For example,
Hand Position 4/4 is significantly associated with Box Weight such
-I
:::t
:::t
,.m
Table 2. Pattern of Significant Effects on Subject and Box Variables on Hand Position for Move/
Z
Carry. Upper Case Represents p <-0:001, Lower Case Represents p < 0.05 C
I

Subject Box
,.n3:
:::t
Z
m
Hand Position Frequency Height Weight Height Width Length Weight
Z
3/7 799 LIGHT HIGH WIDE SHORT LIGHT -I
m
::D
1/9 168 tall HEAVY LONG light ."
,.
8/8 159 TALL HEAVY LOW NARROW LONG HEAVY n
m
7/9 145 SHORT LIGHT LOW narrow SHORT LIGHT
2/8 87 TALL HEAVY high LONG
4/4 87 SHORT LIGHT LOW wide SHORT LIGHT
4/6 67 SHORT LIGHT LOW SHORT LIGHT
1/3 58 HEAVY LONG
2/2 50 heavy HIGH LONG HEAVY
6/7 31 short LIGHT narrow SHORT light
4/5 31 tall LOW LIGHT
7/8 24 TALL LOW
5/5 23 LIGHT LOW light

II.)
o-..J
"-I
0
CO
Table 3. Pattern of Significant Effects of Task Variables on Hand Positions for Move/Carry.
Upper Case Represents p < 0.001, Lower Case Represents p < 0.05

Height of Movement Reach-OVer BLE


Hand Fre- Distance
Position quency Carried Start Mid End Start Mid End Start Mid End

3/7 799 HIGH LOW LOW NEAR near DIFF DIFF DIFF
1/9 168 far far
8/8 159 LOW high FAR NEAR NEAR DIFF diff diff
7/9 145 long high HIGH HIGH FAR far EASY EASY EASY
2/8 87 short LOW LOW low far DIFF
4/4 87 high HIGH high FAR FAR EASY EASY EASY
4/6 77 high near far far EASY EASY EASY
1/3 58 short near
2/2 50 low low FAR near DIFF DIFF DIFF
6/7 31 high near near EASY EASY easy
4/5 31 short high EASY EASY EASY
7/8 24 far NEAR near
5/5 23 long low FAR FAR easy
n
C)

0
::D
C
::D
-<
THE HAND-MACHINE INTERFACE 209

10

Move I 1\11 .!d)llSl


Corry n

Fig. 7. Hand Positions in Box Handling Survey

that lighter boxes have 4/4 more frequently. OVerall, box width and
distance carried have the least significant effects on hand position
with all other variables showing a complex pattern of significant
effects. It appears that the frequency of hand positions is greatly
affected by subject size and box size. Table 3 shows generally that
there are more significant relationships at the start and mid stages
of the movement than at the end, suggesting that humans determine
their hand positions based on factors early in the movement rather
than factors which will be appropriate at the end.

In order to elucidate these results a second study of handle


positions, Drury and Coury (1982), needs to be considered. This was
a laboratory study in which subjects held boxes using different
pairs of hand positions while biomechanical, physiological and psy-
chophysical measures were taken.
210 C.G.DRURY

In this study, ten male subjects each performed one hundred


trials of a static holding task with box held at waist height using
all possible combinations of hand positions 3, 6, 8 and 9 from Fi-
gure 6. The ten hand positions (3/3, 3/6, 3/8, 3/9, 6/8 ••. 8/9,
9/9) were performed once each on ten boxes. The ten boxes were either
10 kg or 15 kg in weight and measured from 400 x 400 x 400 mID in five
steps to 500 x 500 x 400 mID.

The handle was free to pivot in each position so that the


'natural' handle angle could be determined at each position. The
handle forces were measured, and the reaction and friction forces at
the torso/box interface calculated. Heart rate during the task,
Rated Perceived Exertion (Borg, 1962), and body part discomfort
(Corlett and Bishop, 1976) were measured in each trial.

The results showed that handle position had a very large effect
on all measures (p< 0.001) and box weight affected everything except
handle angle, heart rate and body part discomfort (p< 0.001).

Of particular importance is the size of the effects of handle


position. Table 4 lists for each variable the magnitude of the dif-
ference between the best and worst handle positions and the magni-
tude of the difference between a 10 kg box and a 15 kg box. For
every measure except one (force exerted by left hand) there was a
greater difference between handle positions than a 50 % increase in
box weight. Clearly handle position has an enormous effect on all
variables studied.

As important as the significance of the results was the actual


pattern of changes induced by changes in handle position. The vari-
ables fell into two groups with a separate, but internally consi-
stent, pattern in each group.

The biomechanical variables, forces and angles, showed a smooth


decrease in force (and angle from the vertical) as the handle posi-
tions moved from the top of the box to the bottom. When the hands,
one or both, were in positions 3 or 6, high forces were needed to
pull the box against the body, generating high reaction forces and
therefore high friction forces to help support the box. When the
hands were under the box (positions 8 or 9) the forces on hands and
body were minimized. Table 5 shows a typical result from this group.

But low forces do not mean low stress. Heart rate and the other
'cost to the operator' measures showed a very different pattern, for
example, that shown in Table 6. Subjective and physiological costs
were higher at both positions 3 and 8, with lower values at 6 and 9.

In terms of recommending optimum handle positions both 3/8 and


6/8 appear to be strong contenders over the whole set of variables
studied as both provide horizontal and vertical stability. Also,
-t
::t
m
::t
>
Z
Table 4. Size of Handle Position vs Size of Weight Effects in 2 Handed Holding Experiments 0
I
3:
Handle Weight Minimum >
n
Position (10-15 kg) HP Range as Sample ::t
Range Range % wt. Range Size Z
m
Z
-t
m
Heart Rate 6.6 2.8 236 % 11.5 11.7 :II
"T1
>
n
Rated Perceived Exertion 1.9 1.3 146 % 1.2 1.5 m

Body Part Frequency 1.7 0.3 566 % 2.9 11.2

Body Part Severity 0.39 0.2 195 % 0.66 11.0

Friction Force 5.7 1.7 335 % 3.7 1.6

Reaction Force 14.0 9.3 151 % 3.7 1

% MVC Left Hand Force 7.8 7.8 100 % 7.6 3.6

% MVC Right Hand Force 10.6 7.4 143 % 7.3 1.8

Left Hand Force 4.6 4.7 98 % 3.5 2.2

Right Hand Force 6.4 4.6 139 % 3.4 1.1

N
212 C.G.DRURY

Table 5. Hand Forces as a Function of Handle Positions

Right Hand Position


kg. 3 6 9 8

3 13.4 11.7 10.5 8.9


Left 6 10.2 9.2 8.3
H~d
9 9.0 8.0
Position
8 8.9

Table 6. Heart Rates as a Function of Handle Positions

Right Hand Position


b/rnin. 3 6 9 8

3 98.8 97.0 96.2 95.4


Left 6 94.6 95.6 94.1
Hand
9 95.8 96.4
Position
8 100.7

positions 6 ~d 9 both lead to reasonably low stress. A further con-


clusion of this study was that for static holding with a waist-high
container, the body/box interface plays an important part in the MMH
task. Calculated reaction forces were always larger than the con-
tainer weight while body/box friction force averaged 25 % to 75 % of
box weight. The h~dles are being used to help the worker hold the
box against his body and so relieve the hands and arms of upward
lifting forces at the expense of horizontal 'hugging' forces.

The final conclusion of the laboratory study was that there


were no effects of box size except for a small change in tilt ~gle
of the box. There were similarly no interactions between any of the
independent variables and box size. Hence future work can be per-
formed with a single box size as the effect of this v~riable is ne-
gligible compared with the effects of handle position, box weight
and subject differences.

To return to the survey of Manual Materials R~dling in in-


dustry, the h~d position results can now be more fully analyzed.
THE HAND-MACHINE INTERFACE 213

The 3/7 and 1/9 hand positions, used on over 50 % of all movements,
fulfilled the conditions derived from the laboratory experiments,
providing both horizontal and vertical stability. These two diago-
nally opposite hand positions took this principle as far as pos-
sible. Hand positions were not random, as relatively few positions
accounted for the great majority of movements and hand positions
tended not to change after the pre-grasp stage until the adjust
stage. In this latter stage, hand positions on the subject's side of
the box were favored.

If the diagonally opposite hand positions were the most favored


in both laboratory and field, the position 8/8 also rated highly in
both. This position was found to minimize forces on the arms in the
laboratory and also be among the five most frequent positions in the
survey. It was associated (Table 2) with tall heavy people moving
compact but heavy boxes. Such a 'both hands and under' position ob-
viously hat its specialized uses, especially on heavy boxes where
hand forces need to be minimized and narrow boxes where lack of ho-
rizontal stability is relatively unimportant. The only other posi-
tion associated with heavy boxes was 2/2, also a central, symme-
trical position.

A low box height was associated with hand positions 8/8, 7/9,
4/4, 4/6, 4/5, 7/8 and 5/5, all in the lower half of the box. Unless
a box was tall, the subject was unlikely to benefit from having a
steadying hand near the top. Almost the same set of hand positions
were associated with narrow boxes, that is, those which are compact
fore-and-aft.

Task variables (Table 3) give interesting effects on hand posi-


tions. The Biomechanical Lifting Equivalent effects were quite S~mQ­
lar to the box weight effects, with easy BLE values confined to the
lower half of the box and difficult BLE values having either hori-
zontal and vertical stability (3/7) or being centrally symmetric
(2/2, 8/8 and 2/8). With the height and reach-over variables, again
clear patterns emerged. First, hand position was determined more by
the situation at the start of the movement and during the movement
than by conditions at the end of the movement. Second, hand position
3/7 seemed to be associated with lowering (LOW to HIGH) while posi-
tion 8/8 was associated with lifting (HIGH to LOW). High movements,
and movements far out from the body used low and far-back hand posi-
tions (7/9 and 4/4) while low movements used high hand positions 2/2
and 2/8. Movements towards the subject (FAR to NEAR) tended to use
8/8 while movements away from the body (NEAR to FAR) showed more use
of 4/6. Many of these effects look 'sensible' in that they extend a
person's reach or introduce appropriate stability.

Handle position in pushing and pulling tasks has been studied


biomechanically and anthropometrically. In terms of maximum forces
exerted, Kroemer (1974) found that a bar height of 70 % of shoulder
214 C.G.DRURY

(acromion) height gave maximum push forces. A biomechanical analysis


by Ayoub and McDaniel (1974) showed that a bar height of 70 - 80 %
shoulder height gave maximum pushing forces and maximum efficiency
of work. In terms of height above the floor, this gives a range of
860 mm to 1230 mm for 5th percentile female to 95th percentile male.
This is in line with the usual textbook recommendations of a pushing
height of 'between elbow and shoulder'.

It is interesting that a study of 57 pedestrian-powered ve-


hicles in a hospital (Drury, Barnes and Daniels, 1975) found only
37 % with handles of any type. Of those with handles, 71 % had
handles within the 5 % female to 95 % male range but only 43 % had
handles of a reasonable diameter (25 mm to 38 mm). The main problem
with all vehicles was lack of footspace behind and under the ve-
hicle.

Pulling forces can best be exerted at a lower height than push-


ing forces. For example, Ayoub and McDaniel (1974) found 40 % shoul-
der height (490 mm to 620 mm) optimum for pulling. Snook's tables of
maximum acceptable pushing and pulling forces (Snook, 1978) show
that for males a 640 mm height was better than higher bar height for
males and for females a 570 mm height was better than higher bar
height. This is contrasted to push forces where 950 mm was optimum
for males and 890 mm for females.

The Coury and Drury laboratory two handed holding task used
freely pivoting handles so as to measure the most natural angle of
each handle. Not surprisingly, the handle angles measured were near-
er horizontal on the bottom of the box (position 8) but became al-
most vertical along the front of the box (position 3). The actual
mean handle angles to the horizontal axis of the box were

Position 3 83°
Position 6 75°
Position 9 65°
Position 8 55°

While these angles may be optimum for a box holding task at


waist height, Nielson (1978) points out that any changes in the
height at which the box is held or moved will have a large effect on
the handle angles. If the angle between the handle and the subject's
forearm long axis is anything but 90°, then the wrist and hand must
be angulated to accommodate the difference. The angulation is ex-
pected to be radial/ulnar deviation of the wrist, a condition with
an established pathology and an association with worker complaints.
THE HAND-MACHINE INTERFACE 215

In order to test the effects of wrist deviation in a static


holding task, an experiment, Ulate, 1979, was run at SUNY at
Buffalo. Each of nine subjects (7 male and 2 female) held a rect-
angular box by a handle on its upper surface for 30 seconds in each
trial. The handle was adjustable in angle to require 20°, 15°, 10°,
5° and 0° of both radial and ulnar deviation with a horizontal box
and a vertical forearm. Side view still photographs were taken dur-
ing each trial to obtain anatomical measurements of elbow angle,
actual wrist deviation, hand angle (i.e., angle between the hand and
the box handle) and angle of tilt of the box. As in the two handed
task, Borg's RPE and Corlett and Bishop's body part discomfort were
measured in each trial. Also measured were integrated EMG in finger
flexors and loss of grip strength after each trial.

All of the anatomical measures varied significantly with handle


angle but only two 'cost' measures, EMG in the finger flexors and
Pain/Discomfort in the forearm. Both of these measures showed optima
near the neutral point. The main result was that actual wrist devia-
tion was only about one quarter of handle angle, ranging over + So in-
stead of ~ 20°. Subjects tilted the box and bent their forearm-Slight-
ly to reduce the required deviation by about 20 %. However, the major
effect was that the subjects relaxed their grip on the box to change
their hand angle and so reduced wrist deviation by almost 60 %.

This pilot study showed that wrist deviation is seen as enough


of a problem even in a 30 second holding task for subjects to dis-
regard experimental instructions so as to avoid large deviations.

CONCLUSIONS

Whether for moving an object, steadying an operator or control-


ling a machine, the requirements on the hand/machine interface are
quite similar. Handles 25 - 38 rom in diameter, 115 rom long and with
a 30 - 50 rom clearance all around are favored in many different
studies. Handle position is more specific to the task performed by
the operator. In manual materials handling of boxes, handle posi-
tions which allow both horizontal and vertical stability should be
used. This means a 'diagonally opposite' hand position such as 3/7
or 1/9 in Figure 7. Hand positions such as 8/8 in Figure 7, where
the hands are both under the box, minimize forces exerted and are
used in special box movements, particularly with heavy boxes. For
pushing and pulling tasks, handles should be 70 % to 80 % of shoul-
der height for pushing and 40 % of shoulder height for pulling. The
handle should again be of appropriate diameter and positioned so as
to allow free foot movement in walking.
216 C.G.DRURY

ACKNOWLEDGEMENTS

The author wishes to acknowledge both NIOSH and the Department


of Industrial Engineering at the State University of New York at
Buffalo for their support of the series of experiments described
here.

REFERENCES

Ayoub, M. M., and LoPresti, P., 1971, The determination of optimum


size cylindrical handle by use of EMG, Ergonomics, 14:509.
Ayoub, M. M., and McDaniel, J. W., 1974, Effects of Operator stance
on pushing and pulling tasks, AIlE Transactions, 6:185.
Bobbert, A. C., 1960, Optimal form and dimensions of hand-grips on
certain concrete building blocks, Ergonomics, 2:141.
Borg, G., 1962, "Physical Performance and Perceived Exertion," Lund:
Gleerups, Copenhagen.
Brooks, B. M., Ruffell-Smith, H. P., and Ward, J. S., 1974, An in-
vestigation of factors affecting the use of buses by both
elderly and ambulant disabled passengers, BL/TRRL Contract
Report No. CON/3140/32, TRRL, England.
Corlett, E. N., and Bishop, R. B., 1976, A technique for assessing
postural discomfort, Ergonomics, 19:175.
Coury, B. G., and Drury, C. G., in press, Optimum handle position on
boxes: a multi factor approach, Ergonomics.
Damon, A., Stoudt, H. W., and Mc Farland, R. A., 1966, "The human
body in equipment design," Harvard University Press, cambridge
(Mass.).
Drury, C. G., 1980, Handles for manual materials handling, Applied
Ergonomics, 11:35.
Drury, C. G., Barnes, R. E., and Daniels, E. G., 1975, Pedestrian
operated vehicles in hospitals, AIlE Annual Conferences Pro-
ceedings, 184, Washington, D.C.
Drury, C. G., Law, C. H., and Pawenski, C. S., in press, A survey of
industrial box handling, Human Factors.
Garret, J. W., 1971, The adult human hand, some anthropometric and
biomechanical considerations, Human Factors, 13:117.
Grandjean, E., 1980, "Fitting the task to the man," Taylor and
Francis, London.
Greenberg, L., and Chaffin, D. B., 1979, "Workers and their tools,"
Perdel Publishing Co., Midland (Mich.).
HEDGE, 1974, "Human factors data guide for evaluation," U.S. Army
Test and Evaluation Command, Aberdeen Proving Grounds (Md.).
Hertzberg, H. T. E., 1955, Some contributions of applied physical
anthropometry to human engineering. Annals of the New York
Academy of Science, 63:616.
Khalil, T. M., 1973, An electromyographic methodology for the evalu-
ation of industrial design, Human Factors, 15:257.
Kroemer, K. H. E., 1974, Horizontal push and pull forces, Applied
Ergonomics, 5:94.
THE HAND-MACHINE INTERFACE 217

Napier, J. R., 1956, The prehensile movements of the human hand,


Journal of Bone and Joint Surgery, 38-B:902.
Pheasant, S., and O'Neill, D., 1975, Performance in gripping and
turning - A study in hand/handle effectiveness, Applied
Ergonomics, 6:205.
Rigby, L., 1973, Why do people drop things?, Quality Progress,
Sept.: 16.
Rubarth, B., 1928, Untersuchung zur Bestgestaltung von Handheften
fur Schraubenzieher und ahnliche werkzeuge, Industrielle
Psychotechnik, 5:129.
Saran, C., 1973, Biomechanical evaluation of T-handles for a
pronation-supination task, J. Occupational Medicine, 15:712.
Steinfeld, E., Czaja, S., and Beer, J., 1981, "Human factors re-
search with disabled children," Report of School of Arch-
itecture and Environmental Design, SUNY, Buffalo (N. Y.).
Snook, S. H., 1978, The design of manual materials handling tasks,
Ergonomics, 21:963.
Ulate, C., 1980, "Ulnar and radial wrist deviations in a one-handed
static holding task," M.S. thesis, State University at
Buffalo (N. Y. ) .
VanCott, H. D., and Kinkade, R. G., 1971, "Human engineering guide
to equipment design," U.S. Government Printing Office,
Washington, D.C.
Woodson, w. E., 1971, Godd human engineering is possible using off-
the-shelf component products, Human Factors, 13:141.
Woodson, w. E., and Conover, D. W., 1964, "Human engineering guide
for equipment designers," University of California Press,
Berkeley and Los Angeles (Cal.).
ERGONOMIC DATABASE FOR VISUAL DISPLAYS AND VDUs

Harry L. Snyder

Virginia Polytechnic Institute and State University

Blacksburg, Virginia/USA

INTRODUCTION

This paper will survey the current data and knowledge in the
human factors/ergonomics literature pertaining to the design and
evaluation of visual displays and visual display units (VDUs). The
discussion will be broken into several critical areas of current
concern and activity. For each such area, existing data and know-
ledge will be generally summarized, data gaps will be indicated,
current activity in research will be defined, and future problems
or requirements will be suggested. Much of the discussion will per-
tain to activities within the United States, although some techno-
logical developments from other countries will be noted. In addition,
since much of the emphasis on VDUs emanates from European countries,
these driving considerations will also be described.

It is certainly beyond the scope of this paper to define all


potential problem areas and existing data, nor would the author be
presumptuous enough to suggest he is capable of doing so. At the
same time, there appear to be consistent threads of common issues
across many nations, and these elements are the ones considered to
be most critical for future display design und development, as well
~s for future applied visual research.

No emphasis will be placed upon the standardized databases


which relate to the design of simple printed materials, individual
alarm indicators, simplified alphanumeric readouts, and the like.
It is considered that an existing database is quite adequate in this
regard, as represented by the usual human engineering data sources
(e.g., Van Cott and Kinkade, 1963; Woodson, 1981). Rather, emphasis
will be placed upon those areas considered more contemporary and

219
220 H.L.SNYDER

critical, especially insofar as newer technological developments


appear to be causing problems with human visual interaction. Ac-
cordingly, the areas to be discussed in this paper are the newer
technological display hardware areas, the emphasis upon color dis-
plays, the recently increasing emphasis upon interface units such
as touch panels, the entire area of VDUs and standards, 3-dimensio-
nal displays, and visual system research and quantification. Each
of these areas is discussed below, and final summary comments are
offered.

NEW TECHNOLOGIES

The traditional cathode-ray tube (CRT) has been in existence


for many years, is improving annually in its capability, is extemely
inexpensive, and can do virtually everything required in most dis-
play systems other than perform for lengthy periods of time with
small power supplies. For example, CRTs can be obtained in quantity
purchases for as little as $80, including all driving electronics.
Resolution can be as great as 130 lines per centimeter. Nearly full
color can be obtained, with the exception of only moderate saturation
in the green area of the spectrum. And fairly high intensities can
be obtained for outdoor viewing, although this is at the expense
of resolution. While most persons have expected various solid-state,
or flat panel, displays to replace the CRT gradually during recent
years, each year the growing use of solid-state displays is matched
or bettered by an increasing number of applications and sales of
CRTs. In one respect, the continued development of CRT technology
has in fact exceeded the rate of development of solid-state dis-
play technology.

Of the solid-state displays which have seen considerable deve-


lopment, only three seem to be significant contenders for the domi-
nant position of the CRT. Yet, it appears that these contenders
will fall far short of the wide variety of applications of the CRT
and will only substitute for the CRT in particularly limited ap-
plications and circumstances.

Gas Discharge (Plasma) Displays

The gas discharge display, available in both AC and DC versions,


is an orange-emittina, neon gas display which can be obtained in many
sizes and is only a couple of centimeters deep. It requires, how-
ever, voltage supplies on the order of 500 volts and uses a fair
amount of current. Thus, battery powered operation is not feasible.
As such, it is not a contender to replace the CRT in applications
that require large mobility and small power supplies. While this
technology permits good resolution (up to 50 or more elements per
centimeter) it is extremely limited in the amount of luminance it
can exhibit under any circumstances. Its maximum luminance is on
ERGONOMIC DATABASE FOR VISUAL DISPLAYS AND VDUs 221

the order of 100 cd/m2 , which is adequate in typical room environ-


ments but is totally inadequate out of doors.

Additional problems with the gas discharge panel deal with its
limited color capability. The most popular version, with a neon gas,
emits a narrow band orange color, which is totally invariant. While
experimental models have been made with several colors, the colors
are low in saturation and fairly dim in luminance. To date, these
have been considered unacceptable for any meaningful application.

The gas panel can be manufactured in many sizes, and develop-


mental models have been made up to two meters diagonal on a square
display. Typically available models have approximately 24 elements
per centimeter and are available with either 512 x 512 elements or
1024 x 1024 elements. These range in cost from approximately $5,000
to $35,000, including all electronics and depending somewhat upon
environmental protective requirements. Thus, they are not inexepen-
sive and do not in any way threaten the CRT for cost.

The main attractiveness of the gas panel at this point in time


appears to be its cosmetic appearance, extremely linear and stable
image geometry, and flat packaging. Its obvious drawbacks are limi-
tations in color and cost.

Electroluminescent Displays

The electroluminescent display, available in both powder and


film versions, and driven by either AC or DC power, is maturing
slowly. A few versions are available in the current marketplace.
As with the gas discharge display, the· electroluminescent (EL) dis-
play is limited in color, emitting a pale yellow color. Thus, it is
considered unacceptable for monoc~~me television (flesh tones are
unacceptable), nor is it usefu,l---(or multi-color displays. It can
be driven in a matrix addres;~ed mode, permitting both graphics and
alphanumeric characters. It can be obtained in fairly large sizes,
for example 20 lines of 24 characters each.

Problems dealing with the electroluminescent display relate to


its cost (approximately $2,000 for a 12-line panel) and the manu-
facturing yield related to the driving electronics. It is considered
appropriate to address each element in an EL panel by a thin film
transistor (TFT) matrix, which is difficult to manufacture in small,
accurate formats. Thus, the manufacturing cost at the present time,
is unlikely to drop substantially.

Obvious drawbacks to this technology, then, are primarily cost


and color limitations. While the EL display requires a fairly high
voltage to drive it, current requirements are low and battery op-
eration is feasible.
222 H.l. SNYDER

Liquid Crystal Displays


The major contender for the portable market is the liquid
crystal display (LCD), which has achieved great popularity in small
calculators and wristwatches. The technology is quite mature, the
cost per digit or character is extremely small (a few cents), and
the display has the inherent advantage of achieving greater con-
trast as the ambient illuminance increases. Thus, it is probably
the best display technology for use under high illumination, and
yet it can be used under very low illumination with an internal
light source.
Its problems deal largely with the directionality of viewing,
for the crystal form which has the greatest contrast also has the
most limited viewing angle. Other problems deal with matrix address-
ing in large sizes (thin film transistor problems exist here also)
and cost of large matrix panels. To date, no large matrix panels
exist, although several are promised for introduction in prototype
form during 1983.
This technology has the greatest promise to replace the CRT
for the following reasons. First, it presents good contrast under
high ambient illumination conditions. Secondly, it can be battery
powered for long periods of time. Thirdly, it promises to achieve
a reasonable cost in mass production if the transistor driving
matrix problems can be solved. Lastly, it is aesthetically pleas-
ing, as is the electochromic display (ECD), yet it has a longer
lifetime than the ECD and therefore should survive market require-
ments.

Ergonomics Problems
In each of the above technologies there are general problems
which exist and which have yet to be addressed adequately in the
research community. First of these is the overall size requirement
for such displays. Manufacturers have absolutely no understanding
of the combination of display size and information density which is
required for numerous applications. At the present time, there ap-
pears to be a competition to see which company can make the largest
of any given type of display, with no great concern for application,
human visual system requirements, system design requirements, or
other driving influences. Questions are often asked in an attempt
to define the total number of picture elements (pixels) needed for
a given display, without regard for its application. Similarly, it
is often aked how large a display the human visual system can "use".
Professional persons in the ergonomics field understand that these
are meaningless questions, but the enthusiasm of display designers
is not at all dampened by appropriate replies. Thus, some research
is necessary to attempt to define "rules" for display size require-
ments for various types of applications and for various display
technological types.
ERGONOMIC DATABASE FOR VISUAL DISPLAYS AND VDUs 223

A second problem is the overall tradeoff of contrast versus


resolution. It is well established that increases in contrast can
be combined with increases in resolution without loss in visual per-
formance. That is, the total effective image quality is a joint
function of both contrast and resolution of individual display ele-
ments. In the CRT, iricreases in luminance (through beam current in-
creases) typically cause the spot to grow larger and yield a de-
crease in resolution. While this is well understood, most specifica-
tions disregard this relationship. In the case of solid-state dis-
plays, however, changes in luminance (where available) have no in-
fluence whatsoever upon resolution. That is because the resolution
of the display is set by the location of the individual pixels, and
these are invariant in size oder spacing with the intensity control.
Thus, we have a very limited amount of information and data on the
resolution required and the contrast required for various applica-
tions of such displays. Because of the often sharp edge gradiant of
such displays, it is necessary that these questions be addressed in
a research program that takes into account such displays. Some data
do exist (Snyder and Maddox, 1978), but the data are not generally
adequate for all possible technologies and applications.

The last major issue dealing with solid-state displays, and


which also exists in some cases with cathode-ray tube displays, is
the standardization of symbol and character design. While there are
standards that pertain to the font of stroke characters, as in print-
ed characters and text, no such standards have been offered for dot-
matrix or raster-written displays. Thus, the character style or font
is likely to be whatever the electronic or software designer thought
it should be, and one can find virtually anything in this market-
place. Some research has been conducted in this area (e.g., Snyder
and Maddox, 1978), but the research was necessarily limited in scope.
Future research programs should address this issue in order to yield
meaningful data for international standards, especially for upper
and lower case characters and for standard ASCII symbols. Foreign
language (national) characters will require additional research.

COLOR DI SPLAYS

In the last few years, there has developed an increasing


emphasis upon color displays for many applications in many environ-
ments. Office equipment is offered with full color CRTs, air-
plane environments now use color CRTs in the cockpit for commercial
flights, the military aircraft manufacturers are considering color
CRTs, and most companies doing research in solid-state displays
are attempting to incorporate color into those technologies. In
part, this popularity has resulted from the high resolution shadow
mask CRT, which gives an image quality almost as good as that'of
a high resolution monochrome CRT. In addition, the recent design
and perfection of narrow bandpass filters for tricolor CRTs has
224 H.L.SNYDER

permitted the display of high contrast images, in full color, under


high ambient illumination conditions. These filters are now avail-
able commercially at reasonable cost, thereby promising to expand
the color CRT market into less friendly environments.

Ergonomic Issues

While the general increase in emphasis upon color displays is


noticeable, there is simultaneously a distinct lack of appropriate
research and standard development for acceptability of such color
displays. In fact, careful examination of existing standards in-
dicates no general guideline for when or when not to use color dis-
plays, although it has been demonstrated in dozens of experiments
over the past several decades that color coding of displays is not
always beneficial, and can sometimes be harmful (Christ, 1975). Thus,
the general movement toward greater use of color displays appears to
be due more to aesthetic desires and marketing capabilities than to
functionally valid scientific or engineering requirements. Further
research is clearly needed to define the appropriate uses of color
displays, as well as to define the specific characteristics and re-
quirements for color display in those applications.

Standards define the appropriate CIE x,y coordinates for inter-


national colors and for discriminability in reflective displays.
However, there are no equivalent standards or recommendations for
color coding of emissive displays which are compatible with long-
term visual viewing. That is, the existing literature recommends
appropriate co~ors for optimum display coding based upon six or
eight simultaneously displayed colors. No mention is made of the
necessary alterations in both luminance and saturation to achieve
long-range compatibility with the visual system. It is well estab-
lished that highly saturated blue and red colors are viewed as
having much greater brightness than equally saturated greens or
equally luminous whites. In addition, purples are viewed as having
even greater brightness than white or green colors of the same
luminance (Booker, 1981). Recent research has shown that i t is
necessary to reduce the luminance and saturation of blues, reds,
and purples to achieve a.uniform brightness display which observers
can use for sustained periods of time without significant visual
fatigue (Costanza, 1981).

In addition, there is no adequate metric of color contrast


which takes into account the magnitude of perceived differences
between colors when the colors vary in luminance, dominant wave-
length, and saturation. Recent research along these lines has in-
dicated that the brightness prediction of monochromatic stimUli
(Booker, 1981; Costanza, 1981) can be used in a complex fashion to
predict the magnitude of total color contrast between colors vary-
ing in saturation, dominant wavelength, and luminance. Modifications
ERGONOMIC DATABASE FOR VISUAL DISPLAYS AND VDUs 225

of the current eIE uniform chromaticity scales are necessary for


the display designer to make use of such data.

One of the major difficulties in using nonequal brightness


colors is the production of visual fatigue, as the eye attempts to
adjust its "gain" to compensate for varying brightness differences.
Related to this, especially under low ambient illumination condi-
tions and with a black display background, is the phenomenon of
chromostereopsis. Under conditions yielding chromostereopsis, the
chromatic aberration of the visual lens causes a perceived differen-
tial distance of the colors ranging from blue through green to red
in the spectrum. As a result, highly saturated blues and reds are
perceived as being in front of or behind the display plane, thereby
causing a false threedimensional perception by the observer. Since
the lens of the eye attempts to correct for this focusing error,
visual fatigue results from the continued accommodative reaction,
during which the observer believes the reds and blues (and purples)
to be "swimming" in front of or behind the display. This is parti-
cularly distressing, often causing feelings of nausea or dizziness,
in addition to inducing visual feelings of fatigue. Increasing the
luminance of the display background or reducing the saturation of
the blue and red colors can alleviate the problem. Display designers
are not generally aware of this phenomenon or of its solution.
Neither are most users.

Lastly, there is totally inadequate information available per-


taining to metrics of color image quality. While recent research
has generally produced acceptable measures of monochrome display
image quality (Snyder, 1980; Task, 1979), extrapolation to chromatic
images is beyond our present ability. To the best of this author's
awareness, no current research exists on quality metrics of chroma-
tic displays. Again, the technology appears to be leading the neces-
sary research to define requirements for the technology.

INTERFACE UNITS

Wi th the increasing desire to make computers more" £riendly"


to persons who have not had extensive programming experience or who
have no hardware computer knowledge, manufacturers have resorted
to implementing techniques which reduce the need for programming
skills and usage of the keyboard on the computer. Thus, we see in-
creasing emphasis for peripheral control devices such as a track-
ball, a data tablet, a force or displacement stick, and a "mouse".
These devices permit natural up-down and left-right movement com-
patible with the display coordinates. More recently, several types
of display overlays or touJh panels have been developed to permit
the user simply to pOint to or touch the display at a location in-
dicated by the displayed menu items behind the panel. The display
is capable of sensing the position touched and changing the dis-
226 H. L. SNYDER

play content in accordance with the commanded information. Touch


panels are considered very "friendly" inasmuch as the responses are
thought to be natural and not unique to computer interaction. While
these devices are increasing in popularity, the necessary research
to optimize the design of these panels is just beginning.

Ergonomics Issues

Research is definitely needed to directly compare the user per-


formance and user subjective utility of these various types of de-
vices, both display mounted as well as peripheral devices. Some data
do exist comparing various dynamics and characteristics of displace-
ment sticks and force sticks. However, this' author knows of no data
to support the selection of a mouse versus a trackball versus a data
tablet versus a stick of any type, for example. When lightpens and
lightguns are added to this array of techniques, the designer is at
a loss to select one from another and usually bases his decision upon
pure cost or subjective judgement, rather than upon engineering de-
sign data. Comparative performance data are needed.

The touch panel or display overlay causes additional problems


related to the quality of the image and the continued use of such
techniques. Many such displays have a finite thickness and notice-
ably degrade the quality of the image. Since some techniques require
an air separation between the display and the panel (e.g., the
acoustic wave panel), the optical refractive surfaces cause even
greater degradation of display image quality. Some devices cause
significant refractive index differences (e.g., resistive and capa-
citive panels) while others have no physical overlay on the device
itself, sensing the position of the observer's finger by peripheral
mounted detectors (e.g., LED detector arrays). Necessary research to
evaluate the functional utility and to specify requirements for such
display panels should include both optical measurements as well as
observer performance measurements. Research of this form is under-
way in our laboratory, but the results will not be conclusive as
the studies are only exploratory in nature. The hardware technology
seems to be leading, by far, the research defining the requirements
for the hardware.

A long-term problem with many of the overlay panels is, in


addition, the buildup of scratches and finger oils from constant
use. Professionals in the field have noticed that users of CRT
terminals tend to place their fingers on the screen, even when no
functional usage can be ascribed to the touch interaction. Thus,
one can expect touch panels to be used in excess of the actual re-
quirements for touching the panel. Many such panels are soft in
nature, scratch easily, and easily accept finger oils, which ulti-
mately will significantly degrade the image. Some panels on the
market have been observed to be essentially opaque after a few
months of usage, due to the buildup of oils and scratches. Other
ERGONOMIC DATABASE FOR VISUAL DISPLAYS AND VDUs 227

panels change their electrical characteristics as oils and scratches


are generated, thereby yielding lower accuracy and ultimately unac-
ceptable usage.

Lastly, many touch panels are highly reflective in nature and


therefore need to be integrated with glare reduction filters or anti-
reflective coatings. Such integration has yet to be addressed ad-
equately by the industry or by research related to visual image
quality as a function of such combinations.

STANDARDS AND THE VDU

The last few years have seen strong international emphasis upon
the development of standards for VDUs and VDU operators, particular-
ly under heavy usage conditions. West Germany has generated DIN
66 234 and a similar standard from the Trade Cooperative Association
(Berufsgenossenschaften). It is generally assumed that terminals
sold for commercial usage in West Germany and in much of Western
Europe must meet such standards. These standards exist because of
great concern over the visual and health problems of the worker, and
reflect the very strong union involvement in hardware design for the
workplace. Internationally, it has been estimated that approximately
15 million workers currently use VDUs, and this number is likely to
double approximately every two or three years. Because of the many
complaints of visual fatigue and other physical difficulties, it is
extremely critical that VDUs be designed to be compatible with human
visual requirements.

Ergonomics Issues

There are two major categories of areas in which better data


must be obtained for adequate VDU design. In the first case, the
hardware itself is of a critical nature. Various design parameters
which are largely disregarded or treated in a convenient fashion
must be adequately addressed in research and in the resulting stan-
dards. Several of these issues have been mentioned above. For ex-
ample, current philosophies and standards permit the use of various
color phosphors ranging from the green through the orange parts of
the spectrum, yet state nothing about the narrowness or distribution
of the spectral emission required. These standards also permit the
use of one color phosphor as the character or symbol color against
the background of another color, thereby permitting the possibility
of chromostereopsis or of defocusing due to the chromatic aberration
of the visual lens. Clearly, existing data suggest that phosphor
colors should be in the white or green range and that color contrast
should be avoided for prolonged usage. Another characteristic of
the display which is largely desregarded, and was noted above, is
the character style or font. Various displays use different fonts.
228 H. L. SNYDER

No standardization of font exists, yet legibility and readability


data show conclusively that some fonts are clearly better than
others.

Perhaps most critical in the area of display design is the


issue of spot or element size versus element spacing or raster pitch.
Many displays have dots which are separated sufficiently to permit
visibility of a blank or inert space between adjacent dots, causing
the characters to appear very definitely as a dot formatted charac-
ter. It has been shown that dot characters become less legible as
the spacing between dots increases (Snyder and Maddox, 1978), and
that the most legible character is in fact a stroke character
(Figure 1). It has further been shown that greater separability bet-
ween adjacent raster elements yields poorer visual performance
(Keesee, 1976), yet no specification currently exists for either
raster pitch or spot size to prevent such "breakup" of the image.
Future standards and display design requirements should reflect
these existing data and understandings.

Another related hardware issue is that of contrast polarity,


with positive contrast (negative display) being light characters on
a dark background and negative contrast (positive display) being
dark characters on a light background. Dark characters on a light
background should have approximately 20 percent greater strokewidth
to appear to have equal contrast with the opposite polarity
(Cavonius, 1981), yet vi-oeo inversion designs do not take this into
account. The advantage of negative contrast is a reduced suscepti-
bility to glare from the front of the display or to reflected
images. On the other hand, a liability of the negative contrast is
a requirement for a higher refresh rate to avoid flicker. Few manu-
facturers are aware of this requirement. Thus, while some standards
prefer a negative contrast, adequate research to firmly demonstrate
the importance of this requirement has yet to be done to the satis-
faction of the research community.

The second major category of ergonomics issues deals with the


interaction between the display surface and the ambient illumination
or environment. Many environments are incompatible with the heavy
usage of visual displays, due to high ambient illumination levels,
direct glare sources., highly luminous window areas, and the like.
Techniques for integrating VDUs into the working environment, in
terms of adequate illumination and glare control, are well estab-
lished. However, in many installations, such control cannot be
achieved adequately and users resort to various types of glare re-
duction or antireflective coatings on the displays. Existing stand-
ards suggest alternative filters for this purpose, and some stand-
ards suggest a rank ordering of the effectiveness of such filters.
It must be recognized, however, that any filter placed over the
display must necessarily reduce the amount of transmitted energy
through the filter, thus achieving increases in contrast at a cost
ERGONOMIC DATABASE FOR VISUAL DISPLAYS AND VDUs 229

-;;;4r----r----,-----r-----,
~~
~i= 3
~~
a::...J
WW
Il.!a 2
WID
~Z
1-«
«Z~
W(f)
1

:;:!;~
~
O~----'------'------'-----.l
0.0 .05 1.0 15 2.0
INTERELEMENT SPACE/ELEMENT SIZE
(EDGE-TO-EDGE)

Fig_ 1. Effect of dot spacing upon text readability

of reductions in luminance. In some cases this tradeoff may not,


in fact, be beneficial. That is, one very popular filter on the
market improves the contrast rendition, under diffuse glare condi-
tions, by approximately .5 percent but is accompanied by a loss of
45 percent in luminance! That tradeoff is not beneficial to the
visual system.

Other filters considerably soften the image, thereby reducing


the overall image quality although achieving some increase in con-
trast. Yet still other filters cause a decided texturing effect to
the display in a plane other than that of the displayed image,
thereby causing focusing problems on the part of the observer which
can lead to visual fatigue. No research has adequately explored the
differences among these filters, has measured the image quality
degradation through the various filters, and has related these to
performance and subjective feelings of visual fatigue. Pilot re-
search of this form has recently been discussed by Jacobsen (1982),
but additional research is urgently needed.

THREE-DIMENSIONAL DISPLAYS

Another area which has recently seen renewed emphasis is that


of three-dimensional (3-D) displays. While 3-D photography has
existed for many years, using polarized glasses or red and green
glasses, this technique has been considered unacceptable for
sustained display. usage. However, in the last couple years, pro-
ponents of computer graphics have wished to display true 3-D pro-
jections on single display surfaces, and have resorted once again
230 H. L. SNYDER

to the use of the green and red glasses and putting one image on a
red phosphor and the other on a green phosphor of a traditional
color display. Other techniques to achieve 3-D capabilities in a
two-dimensional display include mechanical shutters, one for each
eye synchronized to alternate fields of the display, each field re-
presenting one of the two images necessary to achieve the third
(depth) dimension. Piezoelectric shutters are also used in place of
mechanical shutters for the same synchronized effect.

Holographic images have been around for some time, achieve a


3-D image, permit the observer to view the image from various di-
rections in 3-D space, but suffer the problem of "speckle" of the
image. For some conditions, this speckle is considered to be un-
desirable and in fact disturbing to the eye.

More recently, a true 3-D display has been achieved by the


synchronized vibration of a parabolic mirror with the generation of
a two-dimensional display on a standard CRT. Thus, the high fre-
quency change of the focal plane with movement of the parabolic
mirror is synchronized to the painting of the image on the CRT, per-
mitting the viewer to look directly into the mirror and see a vir-
tual image having true 3-D content. This device has recently become
available in a commercial version, and is financially compatible
with moderately expensive computer systems and computer graphics.

Ergonomics Issues

Major ergonomics issues remain to be addressed and settled in


the area of 3-D displays. Among these are questions of the necessity
of true 3-D displays for many applications. Careful experimentation
is required to determine whether or not additional information con-
tent is obtained from 3-D displays as compared to standard 2-D dis-
plays with appropriate depth and coding in some symbolic form. Ade-
quate research in this area has yet to be undertaken.

Assuming 3-D displays are desirable, questions that must be


addressed in research relate to the resolution and quality of such
displays. For example, is it necessary that there be ~pth resolu-
tion comparable to the height and width resolution of the display?
How does one measure image quality in a 3-D display? What are small
increments in image quality measurement and how do they relate to
performance? Is color necessary in 3-D displays? Must the dynamics
of 3-D displays be as good as those of 2-D displays? What are the
limitations of data update rates in 3-D displays, and how do these
relate to perceived flicker and blur?

Each of these areas must be addressed in a carefully construct-


ed research program, incorporating both the visual issues as well
as the necessary controls of the 3-D display hardware and software.
This is not a small research task, but because of the rather ex-
ERGONOMIC DATABASE FOR VISUAL DISPLAYS AND VDUs 231

pensive and complicated nature of 3-D displays, the research certain-


ly should be conducted prior to the introduction of very expensive
and complicated 3-D display products.

VISUAL SYSTEM QUANTIFICATION AND DISPLAY DESIGN/EVALUATION

During the past decade, new approaches have been taken toward
the quantification and characterization of the human visual system.
This research has yielded concepts and measurements of visual system
performance which are in a form that is compatible with the analysis
and specification of visual display products, thereby permitting the
visual system and the display product to be analyzed compatibly
through linear systems analysis. The fundamental concept of interest
in the spatial domain is that of the contrast sensitivity function,
which relates threshold contrast capabilities of the visual system
to the spatial frequency (e.g., cycles per degree) of a sinusoidal
stimUlus pattern. The general result of such research is that the
visual system is essentially a bandpass filter, with the greatest
sensitivity (requiring least contrast) at approximately 3-7 cycles
per degree and requiring greater contrast at either greater or lesser
spatial frequencies. Changes in this contrast sensitivity function
have been related to such parameters as retinal illuminance, ex-
posure time, display size, observer age, grating orientation, chro-
matic grating characteristics, and the like. Further, combining the
contrast sensitivity with measures of contrast rendition (modulation)
by the display, using the same coordinates, has resulted in accept-
able measures of overall display image quality '(e.g., Biberman,
1973; Snyder, 1973; Snyder and Maddox, 1978; Task, 1979).

Using these techniques, the spatial contrast sensitivity func-


tion has achieved a meaningful role in the evaluation and specifi-
cation of display quality. In addition, a similar contrast sensi-
tivity function exists in the temporal domain, relating threshold
contrast to time varying characteristics of grating patterns. In
this fashion, the visual system has been demonstrated to operate
as a bandpass filter in the time domain, with maximum sensitivity
depending upon the overall display luminance level. Increases in
display luminance yield more sensitivity, or greater susceptibility,
to perceived flicker. Again, the temporal contrast sensitivity func-
tion can be used to assess the requirements for display refresh and
flicker avoidance.

In both the temporal and spatial domains, it has been shown


that linear systems analysis (i.e., Fourier analysis) can be used
to predict visible displayed content and visual susceptibility to
flicker. These techniques are reasonably well established, and
contemporary display designers make use of these concepts in many
cases.
232 H. L. SNYDER

1.0

0.5

FIELO SIZE • ~5 DEG


Z
0 LUMINANCE· ~50 CD/M2 / I
I
i= 0.1 I I
<[ I I
...J I I
:::> I I
0 0.05 ESTIMATED I
0 I
I
::;: 90% I

C~"~""
I
UMITS I I
I I
l- I
I I

I'" /
(fJ I I
<[ I
0.01 I
0::: I
l- I
Z
0
'.... ....... _-", ,;/ I
0.005 I
U I
/
"
" " ----" ... '" '"

0.001 L--'---L--'-J....L.LJ...1J_-'--'---LJ....L.JUi.1_-'---'---LLLl-U.J
0.1 0.5 I 5 10 50 100

SPATIAL FREQUENCY. CYCLES PER DEGREE

Fig. 2. Typical contrast sensitivity function, plotted as a thre-


shold function

Ergonomic Issues

While the above discussion, necessarily of a simplified nature,


indicates appropriate application of the contrast sensitivity func-
tion to both the spatial and temporal domains, it has yet to be
proven that the contrast sensitivity function can be meaningfully
used with displays varying in chromaticity. While chromatic sensi-
tivity functions exist, inadequate data thus far have prevented the
useful extrapolation of such functions to the specification of chro-
matic image quality. Considerable research is needed in this area,
and could be done in conjunction with research exploring uniform
color contrast scales, as eluded to above.

The more ciritical ergonomic requirement at this time is to


make ergonomics experts aware of these concepts of display quality,
to make display designers aware of such concepts, and to force the
integration of these metrics into display standards. Current stand-
ards appear to be woefully naive of any such related measures of
display quality or visual system characterization.

SUMMARY

From the above discussion, several generalizable results appear


warranted. First, in many cases, hardware and software technological
ERGONOMIC DATABASE FOR VISUAL DISPLAYS AND VDUs '233

developments are moving at a rate much more rapid than is that of


the research which generates and reinforces the development of such
technologies. That is, requirements for displays should be specified
from appropriate ergonomics research. The requirements are totally
missing, the technologies themselves having become the driving force
in the visual display area. More research is needed, and this re-
search should be of a forward-looking nature, rather than merely
the exploration of old problems.

Secondly, there exists a good quantitative, applied vision


database which can be used in the design and evaluation of displays.
Unfortunately, display and system designers have thus far been un-
able to combine the quantitative visual database with the quantita-
tive analysis of display designs and display content to produce
compatible systems. This is largely an educational problem, rather
than a technological one.

Thirdly, there is currently inadequate research to support even


those VDU standards which have been offered. Because of the immense
nature of this problem, research on VDUs to produce justifiable de-
sign standards is urgently needed. It must be done in a controlled
laboratory environment, and it must be verified in appropriate field
settings using controlled measures with appropriate subject popula-
tions. Until such research is done, VDU standards may in fact cause
the introduction into the working environment of display designs
which are not beneficial or not even an improvement over existing
designs, yet they might make it more difficult for the subsequent
introduction of modifications to those standards, based upon ap-
propriate research work. This is an urgent, contemporary problem
which needs to be addressed on an international basis.

REFERENCES

Biberman, L. M., ed., 1973, "Perception of displayed inf<;>rmation,"


Plenum, New York.
Booker, R. L., 1981, Luminance brightness comparisons of separated
circular stimuli, J. Opt. Soc. Am., 71:139.
Costanza, E. B., 1981, An evaluation of a method to determine supra-
threshold color contrast on CRT displays, Master's Thesis,
Virginia Polytechnic Institute and State University.
Cavonius, C. R., 1981, Personal communication.
Christ, R. E., 1975, Review and analysis of color coding research
for visual displays, Human Factors, 17:542.
Jacobsen, K. D., 1982, A comparison of optical measurements of dis-
play products, SID International Symposium Digest of Technic-
al Papers, XIII.
Keesee, R. L., 1976, Prediction of modulation detectability thre-
sholds for line-scan displays, USAF Technical Report AMRL-
TR-76-38, Wright Patterson Air Force Base.
234 H. L. SNYDER

Snyder, H. L., 1973, Operator performance and image quality, in:


"Perception of displayed information," L. M. Beberman, ed.,
Plenum, New York.
Snyder, H. L., 1980, Human visual performance and flat-panel display
image quality, Virginia Polytechnic Institute and State Uni-
versity Technical Report BFL-80-1.
Snyder, H. L., and Maddox, M. E., 1978, Information transfer from
computer-generated dot-matrix displays, Virginia Polytechnic
Institute and State University Technical Report HFL-78-3.
Task, H. L., 1979, An evaluation and comparison of several measures
of image quality for television displays, USAF Technical Re-
port.AMRL-TR-79-7, Wright Patterson Air Force Base.
VanCott, H. P., and Kinkade, R. G., 1972, "Human engineering guide
to equipment design," U.S. Government Printing Office,
Washington, D. C.
Woodson, W. E., 1981, "Human factors design handbook," McGraw-Hill,
New York.
ERGONOMICS OF SOFTWARE FOR VISUAL COMMUNICATION

A. van Meeteren

Institute for Perception TNO

Soesterberg / The Netherlands

INTRODUCTION

Parallel communication and parallel processing

Visual displays for communication are very old. We all know


the prehistoric paintings on the walls of the caves in which our
ancestors forgathered. Also writing and reading have existed long
before the first "write" and "read" instructions were incorporated
in computer programs. So, in considering the ergonomic approach of
man-equipment communication through the visual channel, it will be
wise to first survey the basic ergonomic experiences with hard copy
books, figures, maps, tables, forms, slides and the like. Admitted-
ly, most of these visual aids played an educative rather than a
"job aid" role and this requires reconsideration. Next to this
change of task, the background of computing power provides a number
of new facilities, such as animation and the addition of more di-
mensions.

In this paper written or printed messages as such will be con-


sidered to be more verbal oriented than visual, and, I refer to Dzida
(1984) for a treatment of the cognitive ergonomics of man-equipment
conversations. However, there will be a number of examples of text
lay-outs, made so intentionally in order to apply for the special
properties of visual perception and cognition.

What is it that makes visual communication so special particu-


larly when compared with verbal communication? One may speculate
that verbal communication is related with a limited form of thinking,
known as logical, whereas visual communication is related to a higher
and unfortunately more magic form of thinking, unknown by definition,

235
236 A. VAN MEETEREN

perhaps adequately called "imagination". Wi thin the con text of this


paper it is more relevant to describe the difference in terms of
serial and parallel processing.

The intake of information in a temporal coding sequence is


limited to about 7 bits per sec at the human side and in this way
the computer is speaking to a very slow partner. However, when we
enter a dark room and switch on the light our impression is that we
can see all there is at once. Of course this does not mean that we
can answer much more than 7 one bit questions when we are inter-
rogated after the light is switched off again one second later. On
the contrary, it will be difficult enough to answer 7 unpredictable
questions in this case, except for those among us, who have a so-
called iconic memory. The point about the parallel communication is,
that we can start answering questions with regard to all pixels im-
mediately, i.e. there is no speed limit other than imposed by the
processes of understanding and answering the questions. In other
words there is no perceptible serial order in the intake of pixel
information by the visual system. In fact it often occurs in image
processing laboratories that observers can see how the images are
processed and up-dated serially by the computer, which means that
the computer in this case is slower than the human.

Next to parallel communication there surely is parallel pro-


cessing in vision. The threshold for the detection of a vertical
grating modulated in a random dot display (see Fig. 1) was found
by Flaherty and van Meeteren (1980) to be the same for gratings ex-
tending over 1° x 1 ° and 10° x 10°. Thus, it appeared that infor-
mation, at least in this particular case, could be distributed over
the whole screen area without a loss of integrative power within the
time of one eye-fixation.

The immediate and integral accessibility of visual images fur-


ther allows pattern recognition, usually without noticeable delay.
Perhaps, the magic form of thinking, called "imagination" is based
upon this capacity, as it helps us to relate any image with the
whole body of visual experiences that we have, and all the relations
between objects and events that we have abstracted unconsciously,
by-passing the logic chain of verbalisation. Pattern recognition
allows us to directly recognize relations between dependent and in-
dependent variables plotted in graphs and to use the organizational
structures of tables. Also the illustrative power of drawings stems
from this human faculty.

An important issue in man-machine systems is the "who is in


control" issue. I feel very strongly that men must be in control.
The very reason one has, to still reserve one or more functions to
some operator, logically implies that this operator must be in con-
trol, and that he must know more about his task than the machine.
A prerequisite then is to have a clear mental representation of the
ERGONOMICS OF SOFTWARE FOR VISUAL COMMUNICATION 237

.... ; . , ... .
' :"
". j.~.
. 0:1
'- :i . ..>'-:.:.'..." " , ' .;"
• -'7>- '! ~ '.
...
.;":,
'J. .. ~~~ ..
';
..
. ~

..\0
' :\.
:.. :
~.~ ,.

:
"} : . 'j' •
· .!. .....
f ·.~ .,..~-:"
, ,•
.......~:..; · . I~
.
.;..,.~ :
... ~
.....:- .
....'".
,. . .. . .
~
.\

·.. ·;i · Xt::: ?f-:' ! :' " ~ .


,.,;,. i \. :.).<':, 'i.~"
"
. \,.~ .. · 4~

'\.,. .:\~~ "~n.:·

.... <K
... .''
.\~:,
",.

..... -:' . .
"~~ ".
' , " ~F ,,\. ; I .~
.
~

.. ;;~~ :.
~. ... .:,.. ",, . ..: ;~..;'. ., ' J . :~\ ' ~

.'
:~.~. ;V , :P'i~.~ "-' " .t• .
: .;'.;:'
:...... .... ..-,.,.
.. .~ ~~~.
, ..
-
~
.
"
~"; -: .
.~! .~ . ....:. . '~~
,--..,1.", . ~;:4<0' . ....:!-.. . .
~' . ,'

.:;...- " i' . .. . . .;a,..., ':j~;~


_ J't_

·.:",r· 0
#
r'.1'.'
'; ' "",
,!f::
",'1 :
...~:.. :{~~,..
.~:~. :

."'........ .~:~.~~ .

Fig. 1. Square wave grating modulated in dot-density. Gratings are


equally well detected when covering a small angular area
of 1° x 1° or a large one of 10° x 10°.

processes one has to control, or of the data-base one has to handle.


In order to build up and maintain such representations visual dis-
plays can be designed as a kind of open window, presenting a view
into the structure of the data, or the machinery of a process. Also,
the presentation of menus for man-machine interaction, as an alter-
native for the more machine controlled sequential dialogues is a
powerful option of visual communication.

Search

The size of the immediate field of view of the human eye is


limited and depends strongly on the nature of the visual task. If
maximum resolution is a prereqUisite the field is limited to the
central fovea covering only 1° in diameter. On the other hand, ac-
cording to Engel (1976) conspicuous objects may be found immediately
in fields as large as 50° in diameter. In general, the limitations
of the immediate field of view are compensated by a natural search
process. Discrete eye-movements are registered 2-3 times per second
when observers are inspecting a display. Of course, the search pro-
cedure introduces a sequential element in the display-to-brain
communication link. The scan-paths can be very complicated (Noton
and Stark, 1971). They are best described as sensible dialogues with
the pictures. Eye movements are triggered by conspicuous spots as
well as by relevant cues. Apart from this these dialogues are equal-
ly unpredictable as verbal conversations.
238 A. VAN MEETEREN

The question may now arise whether the sequential nature of


image-inspection by eye-movements does not imply, that the advantages
of parallel presentation are restricted to the immediate field size
of single fixations? This may be so to a certain extent. However, it
appears that the hybrid system of parallel and serial processing suc-
cesssfully maintains the integrity of the image as a whole. In fact
we are not aware of the eye-movements we make. In other words, it
certainly is much better to arrange displays larger than the im-
mediate field of view, than to present a sequence of smaller dis-
plays paced by the machine. In the latter approach the advantage of
showing the spatial (or spatially coded) relations between the blocs
would be lost.

PRINTED VISUAL DISPLAYS

Tables

According to Ironman (1960) tables are the easiest and most


effective way of presenting a mass of data to the reader accurately.
Ironman presents a convincing example of a small table of 3 lines
and 3 columns replacing a paragraph of text of 7 lines, that must be
read a couple of times in order to be digested. Tables do require
careful design. A number of hints for good table design can be found
in textbooks on writing and publishing. The basic principle is func-
tionality: What use will be made of the table concerned? From
Wagenaar's (1982) recent evaluation of a number of tables for the
Dutch Consumer's Association I take the example of a 45 lines by
6 columns table, presenting prices, technical test results, field
trial results, subjective appreciations, and final judgements, in
this order, of 45 different makes of skis in alphabetical order.
The alphabetical order is only functional for readers interested in
one special make, known a priori, for instance, because they just
bought skis of that make. Readers looking for the cheapest make pro-
viding good quality, or the ones that are interested in all skis
that passed field trials successfully have a hard job. Wagenaar
suggests that the order of the lines should be subservient to "do
it yourself" conclusion drawing by the reader. This was clearly
improved be reordering the lines according to prices, or, as an
alternative, according to final judgements. Table 1, indeed, makes
it very easy to conclude that:

I} the cheapest makes are good

2} there is only one very good make below f 300,--

3} there are six good makes below f 300,--

4} there is a (weak) price-quality correlation.


ERGONOMICS OF SOFTWARE FOR VISUAL COMMUNICATION 239

Table 1. Test results of 45 makes of skis. This is an example of


a table arranged in order to help the reader draw his own
conclusions very quickly. In the original table makes were
ordered alphabetically and it was difficult to find that
6 good makes are available below a price of f 300,--

Final Price Technical Field Subjective


judgement Make (guilders) test trial evaluation

Ravenswaaij 305 xx o
FAIR Rijswijk 309 x o o
Maurik 325 x o o

Eck 229 x x x
Lienden 229 x x x
Ingen 259 x x x
Ommeren 269 xx x o
GOOD Echteld 270 x x x
Buren 299 o x x
Zoe len 329 x x x
(20 makes)
Drumpt 655 x x x
Beusichem 660 x x x
Erichem 790 x x x

Zoelmonc 279 xx x x
Deil 360 x xx x
Geldermalsen 495 xx xx x
EXCELLENT
(6 makes)
Kerkdriel 585 x xx x
Avezaath 620 xx xx xx
Wadenooijen 605 xx x x

A well-considered ordering of the columns of a table is equally


imperative. Columns with high mutual correlations may be placed side
by side. Categori.es of properties (for instance "primary" and "secon-
dary") may be grouped deliberately by the format of the table.

More rules for good table design may be found in textbooks. How-
ever, if you take the functional approach, your own commson sense
will tell you how to proceed.

Illustrations

Illustrations like graphs, drawings and photographs are every-


day tools in science and technology. A number of rules for good de-
sign have evolved from experience, dealing with the use of linear
vs. non-linear scales in graphs, the intervals along the scales,
240 A. VAN MEETEREN

Fig. 2. Illustration taken from Descartes' "Traite de l'Homme.


Descartes' idea of a material memory in the brain is clear
at once.

the ranges of the scales, the frame of the picture, the symbols and
the letterings, to mention a few issues, treated adequately in text-
books. Here I would like to emphasize the basic ergonomic principle,
i.e. to ask what use is to be made of the illustration? Literally an
illustration is made in order to "enlighten" something, like Fig. 2
illustrates. Graphs, for example,should not be used to present only
the data. The gain of graphical presentation is in the ordering of
the data, the quick communication of preprocessed data, in order to
facilitate "do it yourself" conclusions by the reader as mentioned
above with regard to tables. I hope the difference between the two
alternative presentations of the same data in Fig. 3 may illustrate
this point sufficiently. The original example in Fig. 3 was made
after a real graph in an otherwise excellent paper in one of our
respectable scientific journals.

Maps probably are the oldest printed visual displays for job-
aid purposes in travelling and geo-poli tics. No wonder that the
functional approach is nothing new in map design, witness the fact
that different maps were made for different purposes early in
history.

Some time ago I was asked by the Belgium Consumers' Association


to evaluate 12 road maps of Belgium and 8 road maps of France
(Van Meeteren, 1974). According to Sheppard and Adams (1971), who
investigated driver's opinions on maps for route finding, preference
ERGONOMICS OF SOFTWARE FOR VISUAL COMMUNICATION 241

10

,
single double single double

-°e
...0
u 8 ~ 0---0

.=.
0

~
c 6
~ ~

"0.s::.
;:
III
...
QI
4

2
~ ~ -.
~ ~

0
~ 002 sec
~B to sec
A B A B A A B
~
o
10 10eo
[] '0
c 8 [] 8 c
°e [] []
°e
" 6
[]
... . 6 6:2
~III []
• !e

6

[]
[]
• • 4 III
~ 4 • 0
0 ...
;:
.
6 6 0
:= •
!c
0
o
! 2 o 2
QI QI

~O
°iii
o o
A B
O:go
A B A B A B
02sec 02sec 10sec 10sec 02sec 02sec 10sec 1osec
"
Fig. 3. Examples of a wrongly designed graphical presentation of
e~erimental result (below) and its improved version
(above). Different symbols refer to different subjects in
a threshold experiment. The difference between A and B in
3 of 4 conditions is better illustrated in the improved
version. Moreover, we now see immediately that all sub-
jects contribute to this difference.

was given to maps being "clearer, less detailed, easier to read,


with plain backgrounds, roads shown more clearly, contours dis-
concerting". In addition to this we concluded from the preferences
of our subjects that there should be a clear hierarchy of road net-
works on the map. A course net of main roads in a conspicuous colour
to be surveyed over a great distance, a fine ne~ of local roads not
interfering with the perception of the course net, only used for
final adjustments of begin and end stages, and if possible an inter-
mediate net of provincial significance.

Forms

Forms, as used by public services, are good examples of modern


"job aid" visual hard copy displays, or rather, at least in The
242 A. VAN MEETEREN

Netherlands, they are bad examples more often than good. Fig. 4
shows how easily Wagenaar (personal communication) could improve
the form used by Utrecht's Electricity Company for billing the
energy delivered to domestic customers. These bills should help the
customer to monitor his expenditure and to check whether or not he
agrees. However, these forms are frequently designed to satisfy the
needs of the company. This must be kept in mind as a general issue:
who's job is made easier, the laity's job or the professional's
job?

FUNCTIONAL DIFFERENCES BETWEEN TRANSIENT DISPLAYS AND HARD COPIES

The "closure"-function of hard copies vs. the t'ransient character of


VDU-displays

In the above paragraphs we have looked at the ergonomic ap-


proach of classical hard copy examples of visual displays for com-
munication, such as tables and illustrations in educational publica-
tions, and maps and forms as "job aid" devices. Many of the transient
displays presented nowadays on VDU screens with computer support are
imitations of their printed ancestors and, as a matter of fact, should
be subject to the same design principles in first approximation. Next
to this we have to realize that computer supported VDU screens offer
exciting new facilities for visual communications. Some of these will
be discussed below, together with some limitations of VDU screens re-
lative to printed hard copies, but before doing so the functional
differences between transient displays and hard copies deserve our
attention. In my opinion, there is one very distinct functional
caesura between hard copies and VDU displays. Hard copies may not
be made for eternity, but definitely have some durability, while VDU
displays come and go. Hard copies have a "closure" function: after
careful preparation and contemplation a decision is made on how to
finally phrase the statement or present the data. From then on what
is printed is closed and "true". VDU-displays function in an ambience
of change. Word processors, for example, are advertised for their
option of easy and endless corrections, insertions, omissions, re-
placements and modifications. There is a great danger, that transient
habits penetrate into the hard copy territory and so cause an in-
flation of "closure".

Hard copy data presentations: more than setting soft copies hard

The ease of making hard copies of data filed in computers has


become a scourge to many innocent users. The need to summarize and
comprise, in general to process the data, seems to have lapsed. One
yields to the temptation of providing all data and leaving all work
to the reader. Reports are produced simply by stapling together a
number of hard copy computer output pages. We will discuss one ex-
ample: the periodical surveys on sick-leaves in my own organization
ERGONOMICS OF SOFTWARE FOR VISUAL COMMUNICATION 243

c
... --.. --
o
--- -,
E F I B A G
!
.
.- -
.,."'''',..~

34! 1~4 160


'
1903 12205 3394!
2368E
34953
2431C
100
624
"'~ ~-- ;
16112
131 Q4
.
1 84 12 210

I I
I
---
!
~I 1'1 ~tW'K-':~ -1 v.-r::_ -1. !l. o....a.-"" ~ tOl-... . .

,I
" II ,.1 1032 1 .I lZ:6()i 1 2292

~
A V M£ETEREN H
3769 DJ KAMPWG 102 SOESTERBERG 5671

~. JM:. percee:som!.C"." vlng


,.
fMm~':~~D~';~Ob-8111 0:;1 1118';'-;~'07 I~
.e""'_~

04555 bC 37179!
I

A B c o E F G H

\
meter readi ng I Amountl kWh price
Amount owed VAT
Cat. last previous I used 'basic fuel total

night 34953 33946 1007 3.4 12.6 16.0 f 161.12 f 29.00


} day 24310 23686 624 B.4 12.6 21.0 f 131.04 f 23.59
kWH ct ct ct
'IFixed charge 180il er rent !Other charges Total charges VAT I
I I fl0.32
I f 12.60 I f !f 22.92 f 4.12
I
8i11 over period
19 March 1981 - 22 May 1981
Sum If 315.0B + f 56.76
I
A. van l1eeteren
Kampweg 102
3769 OJ SOESTER8ERG
Cons. code: 18012007
To be paid
I f 371.79
I
Payment is requi red before:
16 June 1981

Fig. 4. Example of wrongly designed form (above) with improved


version (below). In the improved version it is easy to
follow how the amount used (C) is obtained from the
meter readings (A-B), and how the amount owed (G) is
found by multiplication of amount used with price (CxF).
Further, there is no difficulty in checking sums and VAT.
In the original form the logical chain is broken by two
reversions and one interruption (I). Finally codes, "0"
and "1" are replaced by "night" and "day" in the improved
version, all as a service to the public.
244 A. VAN MEETEREN

(The Dutch Organization for Applied Scientific Research, TNO)). This


report originally contained 13 pages as follows:

ORIGINAL

1. Title

2. Contents

3. List of definitions and code-names of sick-leave variables

4. List of code-numbers of industrial sections

5. List of code-numbers of laboratories

6. Table of sick-leave data of industrial sections in 1978.


Lines haeded by code-numbers; columns headed by code-names

7. Table of sick-leave data of laboratories in 1978. See 6

8. Table of sick-leave data of TNO in 1978. See 6. Only one


line

9. As 6 for 1977

10. As 7 for 1977

11. As 8 for 1977

12. Discussion

13. Discussion

Pages 6-11 are direct copies of computer print-outs and this


undoubtedly is the reason for the irritative wrong design. No table
can be used without inspection of two other pages for code-names
and code-numbers. The data for 1977 and 1978 cannot be compared di-
rectly. The data of the industrial sections are presented for com-
parison, as is told on the very last page of the report, but no
direct comparison can be made between laboratories and industrial
sections. Eleven pages can be gained together with a much more
functional presentation by spending a couple of hours resulting in
Table 2 accompanied by one page of introductory text and discussion.
The result may further speak for itself:

ALTERNATIVE

1. Title, introduction and discussion


ERGONOMICS OF SOFTWARE FOR VISUAL COMMUNICATION 245

Table 2: Sick-leave data for laboratories of the Defense Research


Organization in 1979. For comparison corresponding data
are given for TNO (Dutch Organization for Applied Scient-
ific Research) and some branches of industry. Between
brackets: data for 1978. This table adequately replaces
7 pages of the original report (see text).

unit Sick-leave Frequency Number of


days per head employees

Electronics Lab. 11.3 ( 7.3) 0.9 11


Inst.for Perception 14.6 (13.1) 1.7 83
Physics Lab. 28.5 (30.7) 2.1 281
Techn. Lab. 33.6 (31.0) 1.9 227
Defense Res. Org. 28.1 (28.1) 1.9 604

TNO 24.1 (23.7) 1.9 4.030


Banks and Ins. Cy's 20.4 (23.7) 2.1 11.826
Transportation 31. 7 (31.0) 1.9 16.853
Metal industry 60.9 (48.9) 2.5 5.324

2. Sick-leave data presented in two tables for men and women


separately. Headings directly understandable. Required com-
parisons can be made immediately without reference to other
pages.
Summarizing, the transient style of computer products badly
needs a finishing touch in order to satisfy the "closure" character
of printed reports.

Task analysis of word-processors


The use of a word-processor as a tool for the preparation of
reports and publications should be looked at with some suspicion.
Here again there is a chance for the transient, unfinished style
of man-computer interaction to penetrate into the world of purposive-
ly pursued "closure".
Indeed, the identity of a report gets fuzzy with 10 or more
versions on file. More importantly, the pressure to really think
hard about the contents of the report may sink below the required
critical level, because it is supposed to be easy to improve the
text later. Is it possible to replace one turn of thorough struggling
by five turns of superficial typing? For the time being I am inclined
to recommend that -authors themselves stay away from word-processors.
I will not be surprised when a new generation will learn by early
246 A. VAN MEETEREN

education to use word-processors just as a more powerful tool than


pen and paper, not interfering with the basic character and function
of writing a text for "closure" communication.

The preparation of reports or other kinds of pUblications in-


volves a number of different tasks:

1. Conception of text and design of tables and illustrations

2. First concept input and printing

3. Discussions. Conception of corrections

4. Implementation of corrections

5. Lay-out

6. Printing.

The advantages of word-processors are most functional in tasks


4 and 5. Task 5 requires a page-sized VDU presentation, preferably
with black symbols upon a white background. It is not imperative in
a functional approach to have task 2 executed at a VDU-terminal. Con-
sidering costs and eye-strain complaints it may be best to use the
VDU-screen exclusively for tasks 4 and 5, together with a number of
corresponding keyboards (perhaps just normal type-writers?) for in-
put. As argued above,one should be careful with regard to the com-
bination of tasks 1, 2, 3, and 4 executed by authors in a dialog
with the word-processor system.

VISUAL DISPLAYS FOR MAN-COMPUTER INTERACTIONS

Towards "Utilogs" rather than dialogs

There are two objections against the concept of a dialog between


men and computers, perhaps mutually related. The first is that the
illusion of speaking on equal terms may be evoked, followed by irri-
tation about the stupidity and rudeness of the partner (Van Katwijk
et al., 1979). The second is that computer systems should essential-
ly be considered as tools to be used by an operator in complete con-
trol. Shneiderman (1980) has proposed to speak of "utilogs" rather
than dialogs. This is very attractive indeed from an ergonomic point
of view.

Visual displays are very helpful in the design of utilogs, be-


cause of the parallel processing capabilities of the visual system.
As discussed above, they may help to reduce the speed-discrepancy
between the operator and the computer and so virtually camouflage,
that the latter is practically always waiting for a response from the
ERGONOMICS OF SOFTWARE FOR VISUAL COMMUNICATION 247

first. Secondly, they may help to build up and maintain a clear


mental representation of the data-structures stored in the computer.
In the next paragraphs this will be illustrated by a few examples.

Speeding up interactions

Each step of an interaction procedure involves a choice of the


operator from a number of alternatives. Naive operators may be pa-
tiently guided by a sequence of questions to be answered simply with
"yes" or "no".

As expertise increases, operators do not like to go through,


say, 5 steps in order to arrive at a well-known destination. It is
desirable then to directly present a 5-bit choice, assuming that the
operator knows the alternatives by heart. An intermediate solution
is the presentation of a menu, i.e. a numbered list of alternatives
as an aid to memory, from which a choice can be made by typing in
the corresponding number, or by ticking clearly marked locations
with a lightpen or a cursor. It is a prerequisite that the operator
is familiar with the meaning of the alternatives presented in the
menu.

This familiarity, or rather the lack of it, turned out to be


an awkward drawback of the query-procedure of a view-data system in
The Netherlands (Van Nes and Tromp, 1979). The procedure was a hier-
archical search tree procedure, with a main index page, followed by
second level index pages etcetera. The data base contained only
1000 pages (the British Prestel has about 156000 pages),. and yet sub-
jects viewed about twice as many pages as was strictly necessary,
because they frequently did not even know to which of the 10 items
on the main index the target belonged! This emphasizes the need of
well-defined alternatives in menus.

Menus can also be used in order to define function-keys with


the aid of a visual display. This can be particularly useful in con-
nection with representations of data upon the same display. This
display then is divided into a menu area and a representational
area. In the menu area a number of functions can be defined in order
to execute operations on the data in the representational area, as
Fig. 5 illustrates.

Forms may serve as another visual display for parallel communi-


cation with a data base, next to menus, particularly for writing,
modifying and reading groups of cohesive data, such as personal
names and addresses followed by the corresponding registrations. The
form contains a number of fixed headings followed by fill-in loca-
tions.
248 A. VAN MEETEREN

@) .A
....0
0-

D Is

Fig. 5. VDU-display with menu operation for a hypothetical traffic-


control center monitoring ships moving along a river. The
operator can "set up" new ships by pOinting SET UP in the
menu. The ship-symbol appears in the window S. With LOCATE
this ship can be moved to its proper position. Ships can
also be removed and radio-links between ships and stations
A2 etc. along the banks can be opened and closed. Alpha-
numeric ship data can be called for by TABLE and these
data then appear in the table-area. All commands are exe-
cuted by pOinting ENTER. Finally, SWITCH can be used to
switch to a totally different display with its own menu.

Support of representations

The most important potential of VDU-dis')lays perhaps is the pre-


sentation of mutually related data. Fig. 5 gives an example of a
topographical nature. This can easily be extended more in general
to symbolic representations of orderly structures. Such representa-
tions appear to be very useful job-aid devices. The display of Fig. 5,
for instance, allows the operator to decide what links between ships
and radiostations should be made.

However, as we have argued, it is also worth consideration to


use such representations simply to make the data-base more transpa-
rent, and so help the building-up and maintenance of a mental re-
presentation in the brain of the operator. Fig. 6 presents a table
calling attention for the "Gilian Everhard", but the information
concerned is presented in the context of all other ships, practical-
ly without any loss of search time and with the advantage that once
ERGONOMICS OF SOFTWARE FOR VISUAL COMMUNICATION 249

Fig. 6. Example of a VDU-table of ships moving upstream (upper


half) and downstream (lower half) along a river. The lines
divide the river into a lower and a higher sector. Note
how attention is called for the "Gilian Everhard" moving
upstream in the higher sector. This table could be one of
the presentations in a traffic-control system, manned by
experts.

more the operator is aware that there are more ships in the sector.
Moreover, he may see almost immediately that there are ships in
four different categories. VDU-screens offer a number of tricks to
call attention for data in the context of other data, as well as to
group data in the context of other data. One can apply different
light intensities, reversed contrast, underlining, flicker, or a
different colour.

Bouwman (1978) reviewed the literature on the use of colour.


Colour is very effective for grouping spatially distributed data.
This effect can be so strong, that the use of colour for other pur-
poses must be abandoned (Farrell and Booth, 1975). It must be rea-
lized that calling attention for a group of data in the context of
other data must be limited: there is no use in calling specific
attention for a majority of the data. No more than 3 or 4 colours
should be used for grouping data. It is a serious misunderstanding
250 A. VAN MEETEREN

that colour can be used for coding the value of a variable along a
continuous scale. It only helps to distinguish at best 7 categories.

Care should be taken that structured presentations do not lead


to confusion. In Fig. 6 one might consider to order ships going up-
stream from low to high and ships going downstream from high to low.
However, this would confuse the spatial representation. It is imper-
ative therefore to order ships going upstream and downstream both
from low to high. In general, a visual display should be internally
consistent.

Fig. 7 shows a display with mixed representations of ship


traffic congestions in space and time. Traffic streams are shown be-
low in a spatial ordering, while the time course of traffic density
is predicted for three locations in the upper half of the display.
In this case it is imperative to orientate the time-scale vertically
in order to avoid misleading suggestions of density movements in
space. This kind of display is proposed by our Institute to the port-
authority of Rotterdam to support a permanent mental representation
of traffic density along the river, and its history.

The small window problem

For a number of applications the field size and resolution li-


mitations of VDU's present a problem. For map inspection, for in-
stance, it turns out that the VDU window is but small. One really
would prefer a bigger display with preservation of local detail.
True, the field of view of the eye for detailed vision is easily
matched by VDU's, but the eye has a natural way of covering much
bigger fields by eve-movements. It will be hard to develop an ad-
equate substitute for machine controlled map displays. Maybe the
hierarchical approach mentioned above offers a solution: different
maps for course structures in a large area, and for detail in sub-
areas. An alternative approach is to arrange set-ups of, say, 3 x 3
VDU's, but this seems not to be an elegant way out. One wonders
whether eye-movements may be used to control the window-movements
of a VDU. With this problem in mind one must be careful with inte-
grated VDU-displays replacing multi-clock-panels such as in airplane
cockpits. The advantage of free eye-wandering paced by the needs of
the pilot may not be outbalanced by the advantages of integrated
displays with complicated machine-controlled inspection procedures.

SUPPORT OF PERCEPTION AND COGNITION

Signal noise discrimination

The visual system is quite good in discriminating signals em-


bedded in noise (Chambers and Courtney-Pratt, 1969; Van Meeteren
and Barlow, 1981). Regression lines drawn by eye through scattered
ERGONOMICS OF SOFTWARE FOR VISUAL COMMUNICATION 251

13 4~

17 4~-

m --- --(]IDIj:::=::::::S.1~::==::(!1]2~==:::~!t:==~1~0---- - 1< 101


m -----,fili121'::---+-C:9+--:::rn--~[2>-1-1IIIIIIIj~9 -----lL1l

COtl1AHOO

Fig. 7. VDU-display of traffic density data in the approaches of


the ports of Rotterdam. Traffic density is predicted for
three bottle-necks, four hours in advance, starting at
13.45 o'clock. Time-scales are deliberately oriented ver-
tically in order to avoid confusion with the spatial re-
presentation of traffic-streams in the lower part of the
display. The latter representation shows the details of
the congestion near "Botlek" at 15.45 o'clock. Near
"Botlek" two side-ways branch off to "Botlekhaven" and
"Oude Maas", and here ships may cross and collide. This
display helps operators to build up and maintain a con-
tinuous internal representation of the traffic situation,
its origins and the expected developments in the hours
ahead.

data in a graphical plot agree satisfactorily with computer calcu-


lated least square error fittings. This faculty can be useful for
modern target acquisition devices such as RADAR and SONAR. Fig. 8
is an example of a display for an active sonar-device. The reflec-
tions registered from 5 consecutive soundings in one sector are
presented here in one display, to be evaluated integrally. It is
practically impossible to detect a real target reflection among the
many non-target reflections in one sounding. It therefore is neces-
sary to integrate the registrations of, say, 5 soundings.
Van Meeteren and Sanders (1979) experimentally investigated target
detection by human subjects, i.e. by visual integration of the sound-
ings, and found very satisfactory performances. The alternative is
252 A. VAN MEETEREN

Fig. 8. The reflections of 5 consecutive soundings of an active


hypothetical sonar device in three different sectors.
Distance is read from left to right. Practically all re-
flections are non-targets. Targets may reveal themselves
as long dashes or as tracks of dashes, like in the indi-
cated part of the display. It appears that the integration
of repetitive soundings is done well by the visual system.

to integrate the registrations of the soundings mathematically, and


next print, or show otherwise, target probabilities. The problem is
that the statistics of the target and non-target reflections are
extremely complicated and a tidy mathematical treatment is not avail-
able as yet. So, here the computer is beaten by the human visual
parallel processor.

Process control

According to Bernotat (1972) "in most control tasks computer


predicted information results in a higher control quality". Jensen
(1980) further remarks, that "modern control systems frequently re-
quire information processing rates and prediction accuracies far ex-
ceeding human capabilities".

Again, VDU's offer excellent solutions to show the predicted


movements of ships and airplanes, or the dynamics of processes in
ERGONOMICS OF SOFTWARE FOR VISUAL COMMUNICATION 253

a plant. The reader is referred to Poulton (1979) and to the authors


cited above for some remarks about the design of prediction displays.

It is imperative to investigate the effects of prediction dis-


plays before implementation. The need of such evaluations was empha-
sized once more by a study of Truijens et al. (1974),.who investi-
gated by simulation an anti-collision radar system for ocean-going
vessels. The central component of this system was a display showing
present positions of ships (own ship in the center) together with
predicted positions 15 minutes later. This display helps to stay
away from other ships. The expected increase of passing distances
as a result of AC radar was clearly demonstrated. However, this was
achieved at the expense of a large number of course-corrections.
These are a source of uncertainty for other ships and one may well
question whether the risk of collisions will really be reduced in
this way.

Problem solving

VDU's with computer background finally may provide new thinking


aids. We all readily seek resource in pen and paper in order to pro-
ject some idea that came up in our mind before our eyes. I just won-
der whether this possibly is an attempt to speed up our stepwise
sequential logics by bringing our parallel processors more to the
fore. According to Craik (1967), thinking can be described as running
simulation experiments with mental models. If so, images made with
pen and paper might help to build such models and trigger such ex-
periments.

Another speculation may be that the drawings we make so willing-


ly in order to solve some problem are just means of communications
between our parallel processor faculties and our verbal logics, i.e.
that the eye is used in order to consciously read parallel processing
parts of our own brains.

However that may be, the important question now is, if VDU's with
computational powers may help us even more than pen and paper. I
have noticed at least five possible extensions:

1. The addition of accurate sizes, taken from a data base, can


be an important gain of VDU graphical try-outs compared with
pen and paper designs. Do people and things really fit in
spaces? Anthropometric data bases for working space design
do exist already.

2. Objects can be defined in 3D-spaces and looked at from dif-


ferent points of view. Pseudo 3D graphical plots can be made
to enable the interpretation of a variable dependent of two
other variables and such plots can be rotated in space to
be viewed from different directions.
254 A. VAN MEETEREN

3. Real 3D-displays can be made by presenting different dis-


plays to left and right eye. Thus, the pen and paper aid
is extended from 2D to 3D.

4. Projections of multidimensional objects can be calculated


and one may speculate on improved imagination in multi-
dimensional spaces.

5. Probably most helpful is the possibility of animations. The


visual system is sensitive to change. The growth of a flower
is easily seen when presented in a quick sequence. More
simply: What happens when valve A is opend? In many dis-
plays of control systems the effect of an action is shown
before execution, so that one may have a second thought.

REFERENCES

Bernotat, R. K., 1972, Prediction displays based on the extrapola-


tion method, in: "Displays and Controls," R. K. Bernotat and
K. P. Gartner, eds., Swets and Zeitlinger, Amsterdam.
Bouwman, A. G., 1978, Colour coding of information on displays, Re-
port IZF 1978-C14, Soesterberg, The Netherlands.
Chambers, R. P., and Courtney-Pratt, J. S., 1969, Experiments on
the detection of visual signals in noise using computer-
generated signals, Photo. Sci. and Eng., 13:286.
Craik, K. J. W., 1967, "The nature of explanation," Cambridge Uni-
versity Press, London.
Dzida, W., 1984, Towards an ergonomic design of software tools,
in: "Ergonomic data for equipment design," H. Schmidtke, ed.,
Plenum, New York.
Engel, F. L., 1976, Visual conspicuity as an external determinant
of eye-movements and selective attention, Thesis, Eindhoven
Technical University.
Farrell, R. J., and Booth, J. M., 1975, "Design handbook for
imagery interpretation equipment," Boeing Aircraft Company,
ed., Seattle.
Flaherty, R. T., and Van Meeteren, A., 1980, The efficiency of
detecting tribar patterns modulated in random dot-density,
Report IZF 1980-25, Soesterberg, The Netherlands.
Ironman, R., 1966, "Writing the executive report," Heinemann,
London.
Jensen, R. S., 1980, Ergonomics in modern cockpit displays, in:
"Human factors in transport research," D. J. Oborne and
J. A. Levis, eds., Academic Press, New York .
Noton, P., and Stark, L., 1971, Scan-paths in saccadic eye-movements
while viewing and recognizing patterns, Vision Res., 11:929.
poulton, E. C., 1974, "Tracking skill and manual control," Academic
Press, London.
Sheppard, D., and Adams, J. M., 197.1, A survey of driver's opinions
on maps for route finding, Cartograph. J., 8:105.
ERGONOMICS OF SOFTWARE FOR VISUAL COMMUNICATION 255

Shneiderman, B., 1980, "Software Psychology; Human factors in com-


puter and information systems," winthrop Publishers, Cambridge
(Mass.) •
Truijens, C. L., Bouwman, A. G., and Klerk, J. A., 1978, Evaluation
of anti-collision radar systems by means of simulation, Rep.
IZF 1978-C18, Soesterberg, The Netherlands.
Van Katwijk, A. F. V., Van Nes, F. L., Bunt, H. C., Muller, H. F.,
and Leopold, F. F., 1979, Naive subjects interacting with a
conversing information system, IPO Annual Progress Report
14:105, Eindhoven, The Netherlands.
Van Meeteren, A., 1974, Evaluation of some road maps of Belgium and
France (in Dutch), Report IZF 1974-C3, Soesterberg, The
Netherlands.
Van Meeteren, A., and Sanders, A. F., 1979, How to display digital
SONAR-signals: Some primary threshold measurements, IZF
Report 1979-16, Soesterberg, The Netherlands.
Van Meeteren, A., and Barlow, H. B., 1981, The statistical efficiency
for detecting sinusoidal modulation of average dot density in
random figures, Vision Res., 21:765.
Van Nes, F. L., and Tromp, J. K., 1979, Is view-data to use?, IPO
Annual Progress Report, 14:120, Eindhoven, The Netherlands.
Wagenaar, W. A., 1982, Tables in the Consumer's Guide (in Dutch),
Report IZF 1982-C, Soesterberg, The Netherlands.
TOWARDS AN ERGONOMIC DESIGN OF SOFTWARE TOOLS

Wolfgang Dzida

Institut fur Software-Technologie, GMD

St. Augustin/FRG

INTRODUCTION

During the last decade there has been much research effort in
the field of human factors of workstations with display terminals.
In Germany some of the research findings have influenced the ergo-
nomic standardization of display work stations. Strong recommen-
dations have been made, for instance, for the display contrast of
negative image displays, or for the relation between background
luminance of the display and the greater surrounding luminance of
the workstation. These human factors recommendations deal with hard-
ware characteristics of the visual display. However, it is question-
able whether an ergonomic design of the hardware-related features
will satisfy the goal of a user-friendly workstation.

One cannot judge the influence of a software system on human


task performance solely in terms of visual displays and keyboards.
It is necessary to study different dialogue sessions and the compo-
sition of diverse software tools to get a first impression of how
much the user will be kept dependent on the computer or to what ex-
tent the user retains control of planning and performing his tasks.

In order the evualuate software systems from the user's point


of view, one has to focus on a well-defined part of the whole system
in the scope of evaluation. Furthermore, one needs some criteria for
judgement. An overgeneralization of judgements should be avoided.
So, the judgement criteria introduced in this paper are suitable for
certain kinds of user/machine interfaces. Instead of an evaluation
the measurement of quality characteristics may be desirable. One ex-
ample from literature will be cited and discussed.

257
258 w. DZIDA
Evaluation criteria can also be taken as objectives in the
field of software design. Within this context they may inspire the
system designer to think aQout prospective advantages or disadvant-
ages of a design decision. The automation of human tasks by elec-
tronically controlled systems may have some undesirable human con-
sequences, for instance, deskilling of work. That means, vocational
experience might become useless due to the fact that essential parts
of a user's. tasks and duties have been delegated to a computer. How-
ever, some rationalization effects may vanish, if the user is not
able to adapt a computer to an environmental change. The user may
become totally expert-dependent. This may lead to cost-intensive
situations because of an increasing lag between the pace of the en-
vironmental change and the limited pace of user response.

With some objectives and some potential disadvantages in mind,


an ergonomic approach to the design of software tools is attempted.
This approach is called a 'cognitive ergonomic' one. Software tools
mainly influence cognitive processes, such as planning and mani-
pulating intangible objects, rather than tangible ones which may
challenge manual dexterity.

Cognitive Ergonomics lies somewhere between the traditional


ergonomic approach and traditional cognitive psychology, the first
dealing with the adaptation of hardware characteristics of working
environments to human processing, the latter concerned with problem
solving and related topics. However, the research on problem solving
is solely concentrated on rule-based problems, although in the early
days cognitive psychology also investigated tool-based problems, for
instance, the "monkey-and-bananas" problem paradigm experimentally
investigated by Koehler und published in "The Mentality of Apes",
1925. Cognitive Ergonomics has redetected tool-based problem
domains and deals with the adaptation of tools to cognitive pro-
cesses, such as learning and planning of task performance, the
arousal of curiosity, and initiative, and creativity. Since software
systems can be designed in such a way as to carry on a dialogue with
the user we are faced with quite a unique ergonomic situation: The
user's learning and planning, his initiative, and his creativity are
determined by more than one interface between him and the machine.
People sometimes use the term 'Software Ergonomics' in order to
distinguish between the new ergonomic approach and the traditional
one.

THE THREE MODE USER/MACHINE INTERFACE MODEL

In an IFIP workshop Hilary Williamson has presented a model


that can be interpreted as a three mode user/machine inferface
model. An interactive system, as perceived by the user, has always
three interfaces that define an actual state of user/machine inter-
action:
TOWARDS AN ERGONOMIC DESIGN OF SOFTWARE TOOLS 259

1. The 'terminal interface' determines the input/output features


of a display workstation. By this is meant, for instance, the
system's appearance to the user, the system's output in cer-
tain display areas (so-called windows), the grouping of
characters on the display, a device that allows you to move
an on-screen cursor and function keys. It depends on the
terminal interface whether a user is overwhelmed by a mass
of information or whether he can discriminate between rele-
vant and irrelevant messages. Colours and graphics can fa-
cilitate the user's perception. It also depends on the de-
sign of the terminal interface how many keystrokes a user
needs to accomplish a single task.

2. The 'session interface' determines the form of dialogue bet-


ween user and system. This kind of interface may allow the
user to interrupt a task and to resume another one. The user
starts a new form of dialogue if he requests explanations
from the system. The session interface also indicates whether
the system can manage decisions without consulting the user,
an whether the system can accept free formatted command in-
put. A body of user requirements for influencing the user-
oriented design of the session interfaces has been formulated.

3. The 'functional interface' determines the set of functions


available to the user. For example, does the system have an
editor or not? Generally speaking, the functional interface
provides access to a set of tools and defines the user's
privileges, for instance, the allowance for private filing.
The functional interface ,determines the user's influence on
the creation of new tools and of combined tools, such as
procedures. It depends on this kind of interface whether
the user is able to satisfy all the goals of a complex task
by the available software tools or whether he needs in ad-
dition some traditional tools.

It becomes evident from some of the above mentioned examples


that the three kinds of interfaces are partly interrelated. For ex-
ample, the number of keystrokes necessary to accomplish a task de-
pends
- on the number of freely programmable function keys (terminal
interface characteristic),
- on an actual menu-driven dialogue form, or on the command
language (session interface characteristic),
- on the composition of software tools, i.e., a few complex
tools or a lot of elementary tools (functional interface
characteristic).

Traditional tools or equipment cannot be described according


to Williamson's model, because they have no session interface. It
is not possible to get into dialogue with a camera, or a hammer, or
260 W.DZIDA

a pencil. From the user's point of view, the major advantage of


software tools is their dialogue capability. However, the major dis-
advantage of software tools is that they are intangible. It is up
to the system designer to compensate for this disadvantage by de-
veloping software systems that are highly efficient with respect to
the session interface. This ambitious objective is worth achieving
by interdisciplinary cooperation. As a first step, a set of evalua-
tion criteria for session interfaces should be analysed. These cri-
teria should also serve as objectives for the design of user-friendly
session interfaces.

EVALUATION AND MEASUREMENT OF INTERFACE CHARACTERISTICS

Recent investigations into the user-friendliness of interactive


systems have revealed the following factors to be important design
goals:

1. self-descriptiveness,
2. user control,
3. ease of learning,
4. problem adequate usability,
5. correspondence with user expectations,
6. flexibility in task handling,
7. fault tolerance.

Each of the.factors is defined by a set of session interface


properties. However, it seems difficult for a heterogeneous user
community to~ieve these goals in a satisfactory way. Casual users,
who know a moderate amount about the system, often prefer other
system characteristics than do regular users, who know the system
intimately and use it frequently. Therefore, the user should be en-
abled to define his level of experience by having for instance a so-
called user profile, i.e., an expert mode of dialogue or a special
mode for novice users.

Basic requirements which are associated with the above mentioned


seven factors are as follows:

1. Self-descriptiveness: The system should explain the range of


tasks that a user can do in interaction with the system. The
user should at least be informed about a specific part of
the user manual which explains it in detail. It is desirable
to supply help features pertinent to any dialogue situation.

2. User control: The user should influence the time slope


(chronological sequence) of dialogue steps. He should be in
a position to decide whether he will carry on a dialogue, or
whether he will interrupt it. Novice or casual users should
be allowed to proceed in easy steps (i.e. easy to under-
TOWARDS AN ERGONOMIC DESIGN OF SOFTWARE TOOLS 261

stand). A concatenation of such steps should be permitted for


expert users. In principle, the system should permit cancel-
lation of the last input (dialogue step) without detrimental
side effects.
3. Ease of learning: Novice or casual users should learn how to
use the system in the course of the dialogue. Learning of
the system should be facilitated by special dialogue forms
or help features. Assistance by user manuals or other users
should be an exceptional case. Basic knowledge about the
system should be acquired on a special training course.
4. Problem adequate usability: Software tools should be adapted
to the user's regular set of tasks (for example, a quick ac-
cess to functions by function keys). The user should not be
bothered with tasks which could be managed by the system,
for instance, data management, or memory organization, or
simply the correct positioning of the cursor.
5. Correspondence with user expectations: The course of the
dialogue should meet the user's expectations~ the system
should request analogous user actions to similar tasks. The
system should provide feedback so that the user can recognize
effects of the input. If a response time is longer than
usual the user should be informed. The output of intermediary
results is desirable.
6. Flexibility in task handling: The systeE should allow "the user
to extend the command language. The system should provide in-
put/output according to user's training level. The user
should be enabled to go beyond 'the one best way'. Shorter
ways to perform a task should be offered to trained users.

7. Fault tolerance: The system should tolerate typical typing


errors. If an in~ut was erroneous the system should insist
only on partial retyping. Error messages should involve both
the input that has caused the error and correction hints.

When a system designer attempts to satisfy all these require-


ments he will contribute to a user-friendly session interface. How-
ever, he will succeed only if he takes into account the interrela-
tions between the session interface and the functional interface.
At least problem adequate usability and flexible tool handling de-
pend on design decisions pertinent to the functional interface. As
a consequence, the designer who feels responsible for an ergonomic
design of the user/machine dialogue has to deal with the software
design. The traditional approach to the design of software tools is
that the responsibility for planning and design rests with a group
of EDP specialists. However, the human factors specialist is also
necessary engaged in the development of software tools from the be-
ginning.
262 W. DZIDA

One may argue that the above mentioned evaluation criteria are
vaguely formulated; the measurement of interface characteristics
should demand precisely defined dimensions or scales. I fully agree
with this argument. Nevertheless, I would like to illustrate some
of the difficulties which arise when the designer attempts to de-
velop a measurement method in the field of software ergonomics. Card
et al., 1980, have developed a model to measure the time it takes a
user to perform a task on an interactive computer system. User per-
formance time is simply measured by "the time it takes to do the
keystrokes" given a so-called unit task. This time is determined by
the sum of independent time components: the time for a button push,
the time for the positioning of the hand on the device, the time for
deciding which command to call, the system response time, etc.

The authors of the keystroke model have recommended their model


as a system design tool, neglecting the concern of what the design
tool should be useful for. They claim that the measurement model helps
to understand how the user interacts with the computer at the terminal
interface. This purpose is vaguely formulated. As a consequence, a
misleading application of this measurement concept is possible. A
system designer could attempt to minimize the amount of time for a
unit task, with an idea of unit task in mind which is not adequate.
One might define a whole prodecure as a unit task; then a user would
be able to execute a complex task by one keystroke only. Is such a
system really user-friendly, even for an expert user? That is doubt-
ful. Some of the following paragraphs deals with reasons for such
doubts.

System designers will have an important influence on the user's


work situations. If they ignore ergonomic user requirements the re-
sulting workstations could be effective with respect to the user's
execution time, but could be inefficient with respect to self-de-
scriptiveness, user control or flexibility. Evaluation criteria may
appear to be conflicting; it is difficult for a system to be both
time-saving and easily controlled by the user.

TOWARDS A FORMAL AND TRANSPARENT REPRESENTATION OF WORK

Conventional tools are des~gned so as to reduce human muscular


effort or to allow visual feedback. For each tool the task being
executed by the tool has been investigated in terms of job behavior:
for example, move something, press it, or synchronize it.

User/machine systems have been introduced in order to delegate


parts of traditional human work to machines. As already pointed out,
the human factors specialist is engaged in software design from the
beginning. For delegating human work to a machine it is required in
advance that the work and its conditions are properly described. Many
deficiencies of current interactive systemd may be due to an insuf-
TOWARDS AN ERGONOMIC DESIGN OF SOFTWARE TOOLS 263

ficient representation of human work. An ergonomic design of soft-


ware tools starts with a description of work. We ought to consider
some essential psychological features of work which have been so far
neglected during the problem analysis phase, at the very beginning
of software production.

However, system designers and human factors specialists must


not ignore the users. They have not only a right to influence their
own work situation, but they have also an excellent knowledge of
their work problems and can make useful contributions. The major
barrier that restrains system analysts from a user-participative ap-
proach is a communication barrier. The user is usually not capable
of understanding the formal description of the system designer's
specification.

People are more likely to understand a systems design and its


implications for the working environment if the system designer
would consider a conspicuous description. The user should be enabled
to check the representation of his work. He should examine the de-
scribed procedures to find out whether the typical conditions of
work are properly represented. Furthermore, the exceptional events
should be discussed, resulting in a decision how to consider these
events in the software design.

The software designer needs a formal description, the user


needs a transparent one. Nets can be both, formal and transparent.
c. A. Petri (1980) has illustrated that nets are suitable and proper
to be used for the representation of work and plans. Even lacking a
knowledge of the mathematical foundations of net theory, a user can
reconstruct the intended interpretation of a net, and he can verify
critical regions of a net. Structuring techniques for complex nets
can improve the graphic representation so as to bring critical re-
gions into focus (Oberquelle, 1981). The usefulness of net repre-
sentation.in the field of software ergonomics will now be illustrat-
ed. It is intended to describe some psychological relevant features
of work which may be changed in the course of an introduction of
software tools.

SOME STRUCTURAL FEATURES OF WORK

The structure of a working activity has been neglected in the


field of psychology as well as in the field of software engineering.
However, structural features are most relevant for the study of er-
gonomic principles of software deSign, particularly for the design
of the functional interface.

The structure of a working activity usually comprises the fol-


lowing components: a certain activity, conditions to be satisfied,
and a result (fig. 1). This definition does not distinguish between
264 W.DZIDA

Fig. 1. Components of a working activity

the result of an activity and the goal (or objective). I prefer to


point to the goal of an activity (and not to the result) when de-
scribing the flow of an activity (procedure), since if often happens
that a result only partially contributes to the goal of a procedure.
It is also typical for the procedural aspect of work (and not for
the structural one) that the worker has to evaluate the result with
respect to its goal directed contribution.

Furthermore, the definition of an activity structure is re-


stricted to the performance aspect of work. It is outside the scope
of concern that humans usually anticipate the outcome of a working
activity; i.e., they make a plan. So, the evaluation of a result and
its anticipation pertain to the procedural aspect of work. This is
described in detail elsewhere (Darlington, Dzida, and Herda; sub-
mitted for publication).

When users and system designers study a representation of work,


then they are interested in recognizing the execution conditions of
the activities described. The conditions to be satisfied for the
execution of an activity may be divided into three groups (fig. 2):
(1) tools, (2) objects (or material), (3) parameters which indicate
how to proceed. If a worker sets a certain parameter he intends to
achieve a desirable auxiliary result. Sometimes it is difficult to
understand what a parameter provides for. An example may explain it:
A secretary uses a typewriter (= tool) to print a column of numbers
(= result) on a paper (= object); she sets a tabulator key on a
certain position (= parameter), in order to type the digits more
reliably into the column (= desirable auxiliary result).

The conditions for the execution of an activity are by no means


exhaustively included. Conditions, for example, which arise from
TOWARDS AN ERGONOMIC DESIGN OF SOFTWARE TOOLS 265

Fig. 2. Components of a working activity


T-tool, o-object, P-parameter

division of labour or other or~anizational arrangements have been


neglected. We focus on such conditions as are likely to determine
the functional interface of a software system. It is important to
look at these conditions closely, since they may also influence the
session interface. The user who is concerned with both interfaces
can be enabled to understand the interaction with the system, or be
hindered in understanding it. That depends on the transparency of
some relevant structural relationships between the. results and the
execution conditions of work. The conviction on the part of the user
that he is in control of these relationships can contribute to sati-
sfactory performance.

Instrumental relationships

Interactive systems usually require the user to puzzle over


which result might be achieved by a certain command. The command is
merely presented by a list of commands lacking any hint as to the
typical effects of the commands. The user is restricted to trial and
error learning.

The connection between the tool (command) and the result of an


activity is called an "instrumental relationship". The user is in
control of this relationship, if he has an idea of the characteristic
effect of the tool (instrument). Such a relationship does exist also
between the parameters and the desirable auxiliary results. The user
does not control this relationship if there exist some default para-
meters unknown to him, or if he does not know how to alter them. It
is a basic requirement that the user should be enabled to get a sur-
vey about all existing instrumental relationships.

A special form of dialogue, for instance an excursion (Darling-


ton, Dzida, and Herda; submitted for publication) or a question-
answering dialogue should be implemented to provide a means whereby
users can learn which commands and parameters are appropriate for
the intended results. Such dialogue features would improve self-
descriptiveness and ease of learning.
266 W. DZIDA

Causal relationships

Between the threee groups of execution conditions (T, 0, P) of


an activity and a result exist "causal relationships". The user
should also be in control of this. He should be aware that it is he
who has caused a result. The user shoul'd regularly keep the initiative
upon the beginning or the interruption of a dialogue. If the course
of the dialogue is dominated by the initiative of the machine, then
the user feels himself restricted to the role of an observer. Conse-
quently, he does not behave as a causer, but as a control device in
a technical system. A passive user is not engaged in the planning
of a dialogue. It should be noted that there is some empirical evi-
dence for the fact that effective workers are superior in anticipat-
ing the results and side effects of their work (Hacker, 1978).

Productive relationships

The thiroimportant structural feature in fig. 2 is called "pro-


ductive relationship". It refers to the connection between the orig-
inal object (material) and the end object being transformed by a
tool. A productive relationship can appear to be easy or difficult,
depending on the effort required to achieve a result. The effort may
increase if some execution conditions are still unsatisfied: the
original object is not available, or a tool is not obtainable, or a
certain parameter is not yet adjusted.

System designers develop software tools for reducing the degree


of difficulty that pertains to a task. They attempt to relieve the
user from preparatory work: objects are administered by the system,
commands are available by function keys, parameters are adjusted in
advance according to the usual ~emands, etc. All these facilities
meet user requirements for problem adequate usability. The user need
not bother with tasks which could be managed by the system itself.
Thus, the user can concentrate on his intrinsic task.

It would be very useful to develop a measurement method that


provides estimates of the degree of difficulty inherent in a pro-
ductive relationship. For well-defined tasks one could measure prob-
lem adequate usability of diverse systems, probably in units of dia-
logue steps which are necessary for preparatory work to achieve a
defined result. In other words, one could measure which system among
different systems places a greater burden on the user.

Hierarchical-sequential relationships

The execution of an activity (A) requires that some subsidiary


activities have been accomplished. Three classes of subsidiary ac-
tivities can be distinguished:
TOWARDS AN ERGONOMIC DESIGN OF SOFTWARE TOOLS 267

1. activities which provide tools (A.t),


2. activities which provide objects (material) (A.o),
3. activities which set parameters in the required position
(A.p) •

These activities are subordinate in relation to a supraordinate


activity. On the other hand, they are related to subordinate or sub-
sidiary activities themselves which can also be classified according
to A.t, A.o, and A.p. A structure of working activities that is de-
scribed by these relations is called a "hierarchical-sequential
organized" action (Volpert; in print), illustrated in fig. 3.

Fig. 3 illustrates how to refine gradually an activity into


subsidiary activities up to those activities which can be taken as
"elementary" ones. This refinement process is necessary for a re-
quirement analysis in connection with a system analysis just at the
beginning of software production. After decomposition of complex ac-
tivities, the elements have to be reorganized in order to combine
those activities which make a suitable workstation, and which lead
to an appropriate tool design. Current methods in system analysis
ignore these aspects. Software Ergonomics could contribute to system
analysis in that the just described structural work analysis would
be consequently developed as a method for software design purposes.

The unique worth of a work analysis method could be that users


can better participate in the requirement specification for work-
stations, tools and organizations. The consequences of an implemen-
tation of software tools could become more transparent in advance.

Fig. 3 can be used to illustrate different kinds of work or-


ganization. A horizontal combination of activities (for instance,
A.t, and A.o, and A.p) is in accordance with the principle of "job
enlargement". A vertical combination is in accordance with the
principle of "job enrichment". The combined activities form a com-
plete unit so that the user could be responsible for a meaningful
unit of work completed.

The combination. principles job enrichment and job enlargement


have been conceptualized in solely organizational terms (Alderfer,
1976). The implementation of software tools has an impact on both
the combination of tasks in an organizational environment and the
combination of tools. So, combination principles should be analyzed
also in terms of tools.

PRINCIPLES FOR COMBINATION OF TOOLS

Workstation design and the design of software tools are likely


to be closely connected. The discussion may be made more concrete
by referring to some of the major tasks that a user may perform in
268 W.DZIDA

Fig. 3. Hierarchical and sequential organization of working acti-


vities

a German bank. Four combination principles will be investigated and


evaluated by ergonomic design criteria. This should also happen with-
in the system analysis phase, thereby obtaining an understanding of
the power of tools and the extent of reorganization in advance.

Uniform productive relationships

A typical task to be performed by a bank clerk is to "process


standing orders" for the remittance of money on certain dates, for
example, at the end of each month. An investigation into this acti-
vity reveals that the bank clerk has to deal with standing orders
in alternative ways as follows:

1. to file,
2. to alter,
3. to terminate,
4. to oversee.

These four activities result from a decomposition of a more


complex activity into elementary ones. The constitute uniform pro-
ductive relationships, since the elementary activities deal with
alternative modifications of the sam e object: standing
order (fig. 4).

When a designer is going to develop software tools to assist


these activities, it should be guaranteed that the already existing
uniform productive relationships will not be disturbed. The system
designer might take care of this state of affairs by combining the
tools (f,a,t,o). A powerful tool would result from that design. Power-
ful tools, however, may bring about some ergonomic disadvantages. Dif-
ferent productive relationships inherent in the four activities could
remain hidden to the user, and may become transparent to him only
after some experience with the complex tool.
TOWARDS AN ERGONOMIC DESIGN OF SOFTWARE TOOLS 269

Fig. 4. Elementary working activities of a bank clerk

The experienced user, however, should derive benefit from a


powerful combination tool. He should be allowed to combine elemen-
tary tools himself. In doing so, he abstracts from specific pro-
ductive relationships: to file, to alter, to terminate, and to over-
see will be subsumed in the more general activity which may be called
"process" (fig. 5).

Ease of learning could be improved by a set of simple tools,


Whereas problem adequate usability might accrue to the user-through
a powerful combination tool. This hypothesis could be tested empir-
ically.

Procedural relationships

Another principle for the combination of tools is derived from


the fact that a sequence of activities may make the impression of
closure. There may be a meaninqful start and a meaningful end of a
sequence. The term "procedure" has been introduced for such sequences.

Fig. 6 illustrates a bank clerk's activities which can be com-


bined so as to build a procedure. In order to "alter" a standing
order the clerk has to do the following in sequence: "show" account,
"aLter" standing order, "compute" banking charges, and "print" state-
ment of account.

The activities closely connected with the task to alter a stand-


ing order can be taken as a routine. Apparantly, there is "one right
way" to do the routine. In this case one might accomplish the routine
by a button push. So, the user does not need to reflect the different
working steps of a routine. He abstracts from intermediate results
of the sequence and subsumes the different activities, namely, show,
alter, compute, and print, in the most important step of the whole
procedure which is "alter" (fig. 7).

Software designers are inclined to build procedures when they


approach the problem of automatization. Procedures may have an enorm-
ous rationalization effect. Planning a procedural task in detail is
270 W.DZIDA

Fig. 5. Combination tool for processing standing orders

Fig. 6. Activities connected with altering a standing order

Fig. 7. Procedure for altering standing orders

no longer the user's duty. Thus, he saves much time. He can carry
out a lot of procedures when appropriate combination tools of this
kind are available.
One must take in consideration that the introduction of pro-
cedures might have some undesirable side effect. The working environ-
ment of a user is in danger of being restricted to the seemingly
"one right way", entailing lack of flexibility in task handling. It
is quite certain that the day-to-day work problems will change with
respect to goals and execution conditions. A user may be doomed to
a breakdown unless he is not able to adapt his tools quickly. If he
repeatedly has to call for an expert in order to adapt his procedures,
TOWARDS AN ERGONOMIC DESIGN OF SOFTWARE TOOLS 271

a rationalization effect will vanish rapidly. Furthermore, from an


ergonomic point of view it is undesirable that a user is strictly
expert dependent. One should develop the user and enable him to
build and to alter procedures himself, for example by means of a
language to provide command procedures.

If a user can design his own procedures, he must know exactly


what are the intermediate results of a complex sequence, and which
procedural variants are possible. This kind of intellectual control
does not take place if the user is always confronted with already
prepared procedures. In order to develop user's disposition for fur-
ther learning, the user should be always in control of the procedural
relationships of his work (if he wants to do so). Thus, the danger
of deskilling the user's work vanishes.

Empirical investigations into this workstation design are


necessary. However, the designer may be advised to consider the re-
commended design conception until he has evidence to the contrary.

Generalized instrumental relationship

There is another principle for the design of combination tools.


The most striking feature of this sort of tool is that it can be ap-
plied almost generally. For instance, a user wants to insert a word
in a text represented at the terminal interface. He submits the com-
mand "insert" and the word in question. Furthermore, the user may
insert other objects, with "insert" being the appropriate command:
he may insert digits, or sentences, or messages, or files. In man/
machine systems one may call this kind of tool "general command" or
"total function". The user abstracts from the· instrumental relation-
ship which may be realized given a specific object, for instance,
a word. He considers the characteristic effect of a tool and attempts
to achieve this effect, although he has just been faced with another
object. A user who acts prudently may attempt find out for which
kinds of objects a tool may serve as an instrument at all. See fig.B.

Instrumental generalization is a design prinoiple which leads


to software tools being standardized at the terminal interface, but
not necessarily at the functional interface. A virtual "insert" com-
mand may refer to different access operations with representations
of these tools at the functional interface. Therefore, the standard-
ization of tools at the terminal interface must not be confused with
so-called standard software.

Standardization at the terminal interface may facilitate the


learning process. A user can transfer his experience with "general
tools" to other workstations, since he is already familiar with the
characteristic effect of some of these tools.
272 W. DZIDA

Fig. 8. Instrumental relationship generalized over different kinds


of objects

Creative users are likely to find out to which productive re-


lationships a general tool can still be applied although the system
designer has not yet taken this case into consideration. So, deskill-
ing of work need not be a phenomenon which necessarily results from
the introduction of man/machine systems.

Heterogeneous productive relationships

It is also possible to build combination tools following the


design principle of the 'Swiss army knife'. This tool serves for cut-
ting an apple into pieces, or drawing a cork from a bottle, or clean-
ing ones finger nails, or tamping down the tobacco in a pipe, etc.
Different productive relationships may be realized by this combina-
tion tool, and it is highly suitable for being carried in the trouser
pocket.

The purpose and design principle of this tool should not, how-
ever, be transferred without further ado to the design of software
tools. The objects to be manipulated by this combination tool would
not constitute a uniform productive relationship, nor would the user
recognize any hierarchical or sequential relationship between the
activities performed. The idea behind this tool is probably related
to a need of the user for easy transport. A portable computer would
have been designed according to this idea. However, for workstation
design, such a tool would cause some undesirable ergonomic side ef-
fects. The user would have to accomplish merely subsidiary activities
with results being totally disconnected. The user would never re-
cognize for what purposes the partial results are suitable. Therefore,
heterogenous productive relationships are unlikely to provide a
TOWARDS AN ERGONOMIC DESIGN OF SOFTWARE TOOLS 273

meaningful working environment. For equipment design in a military


environment, however, this design principle may be adequate.

SUMMARY

Man/machine systems have been developed for carrying out dia-


logues with the user. It is necessary for both two partners in a
dialogue to have a model of each other. In man/machine interaction
the user has a model of the system and the machine has a valid re-
presentation of the working activities which may be performed by a
user. From the user's point of view it is helpful to model the man/
machine interface in terms of three different modes: terminal, session
and functional interface.

An ergonomic approach to the design of software tools can make


use of the three mode interface model. Evaluation criteria have been
developed to assist the human factors specialist in judging the
session interface. These criteria, however, serve also a design of
goals for a system designer. He has to consider that the three kinds
of interfaces are closely connected, i.e., the design of the func-
tional interface determines features of the two other interfaces.

On the other hand, the human factors specialist is also con-


cerned with the functional interface when he is going to measure or
evaluate interface characteristics of a terminal or of a session.
In order to develop an ergonomic conception for interface design he
should discuss with the system designer the potential advantages or
disadvantages of a certain design of software tools. Thus, the system
designer may anticipate the implications of a design decision for
self-descriptiveness, ease of learning, flexibility, etc. If com-
puter scientists and psychologists would cooperate better than in
the past, some of the described design principles for combination
tools could be tested empirically.

Cognitive Ergonomics, or sometimes called Software Ergonomics,


will be established as a subdiscipline in its own right. Two major
objectives should be approached within the next decade:
1. Develop a formal and transparent method for the analysis of
human work, useful for system analysis and user partici-
pation; the structural aspect of work should be more strongly
emphasized than the procedural.
2. Resume the research on tool-based problem domains within the
diSCipline of cognitive psychology, rather than investigate
rule-based problems only.
Both research problems may be attacked by interdisciplinary co-
operation. Researchers in the field of ergonomics should respond to
this new challenge.
274 w. DZIDA
REFERENCES

Alderfer, C. P., 1976, Change processes in organizations, in: "Hand-


book of Industrial ana Organizational Psychology,"
M. D. Dunette, ed., Rand McNally, Chicago.
Cakir, A., Hart, D. J., and Stewart, T. F. M., 1979, "The VDT
Manual," IFRA, Darmstadt.
cakir, A., Hart, D. J., and Stewart, T. F. M., 1980, "Bildschirmar-
beitsplclitze," Springer, Berlin, GOttingen, New York.
Card, S. K., Moran, T. P., and Newell, A., 1980, The Keystroke-Level
Model for user performance time with interactive systems,
Commun. ACM, 23:396.
Darlington, J., Dzida, W., and Herda, S., to be published, The role
of excursions in interactive systems, Int. J. of Man-Machine-
Studies.
Dzida, W., Herda, S., and Itzfeldt, W. D., 1978, User-perceived
quality of interactive systems, IEEE Trans. Softw. Eng.,
SE-4: 270.
Hacker, W., 1978, "Allgemeine Arbeits- und Ingenieurpsychologie,"
Huber, Bern.
IBM, 1979, "Human Factors of Workstations with Display Terminals,"
IBM document number G 320-6102-1.
Oberquelle, H., 1981, Communication by graphic net representations,
Bericht Nr. 75, Fachbereich Informatik, Universitat Hamburg.
Petri, C. A., 1980, Introduction to General Net Theory, in: "Net
Theory and Applications," W. Brauer, ed., Springer, Berlin,
Heidelberg, New York.
Verwaltungsberufsgenossenschaft, FachausschuB Verwaltung, 1980,
Safety Regulations for Display Work Places in the Office
Sector, Hamburg.
Volpert, W., in print, The model of the hierarchical-sequential or-
ganization of action, in: "Cognitive and Motivational Aspects
of Action," W. Hacker,w. Volpert, and M. v. Cranach, eds.,
Deutscher Verlag der Wissenschaften, Berlin.
Williamson, H., 1981, User Environment Model, Report of the 1st Meet-
ing of the European User Environment Subgroup of IFIP, Work-
ing Group 6.5, GMD-IST, St. Augustin.
MUNICH THESES OF ERGONOMICS

Karl H. E. Kroemer

Virginia Polytechnic Institute and State University

Blacksburg, Virginia/USA

The dichotomy "Ergonomics vs. Engineering", often mentioned in


the 60s, is no longer valid. Many ergonomists are now educated (at
least in the USA) as engineers, mostly in Industrial Engineering
Departments. Hence, many engineers have general or even indepth know-
ledge of ergonomics. Or, conversely, many human factors/ergonomics
people are in fact engineers (Fig. 1).

This indicates that ergonomics has become one of the accepted


inputs into engineering planning, design, and implementation. While,
admittedly, this inclusion of ergonomics into the engineering curri-
culum and thinking is not yet complete, it is certainly far from the
adversary role often quoted in the 60s (see Fig. 2, page 18). In fact,
there is an obvious trend to consider ergonomics as a necessary and
basic component of any planning and design projects in which people
and machines are involved.

The following statements are meant, partially, to be "provocat-


ive", that is to elicit response and interest. However, they are
mainly meant to indicate current status, status nearly achieved, or
obvious trends in Ergonomics.

Thesis 1: Ergonomics (Human Factors) is established as a discipline.

The ergonomist is no longer a "trouble shooter" only, called in


after the design is nearly completed and deficiencies in the human-
machine interface are apparent. The ergonomist can expect to parti-
cipate in the early phases of the design, that is in the concept,
planning, and initial design steps of a manned system.

275
276 K. H. E. KROEMER

Fig. 1. Ergonomics and engineering science

Thesis 2: Ergonomics needs to be "proactive" instead of "reactive".

In the past, much ergonomic research has been based upon ob-
servations of how person-machine interaction developed after the
system was established. Thus, ergonomic data usually consider the
reactions of people within established systems. Ergonomics is moving
away from its descriptive phase, in which much work was done in retro-
fitting and curing. Ergonomics is moving into the prescriptive phase,
which provides data and recommendations for the design of future
effective, safe and satisfying work systems.

Thesis 3: Ergonomics develops from detail data to systems solutions.

Ergonomic state-of-the-art includes a wealth of detail knowledge.


However, interactions between many, and complex systems variables
(and the resulting ergonomic recommendations for the design of the
system, and its components) were usually not formally considered.
Thus, the major task of ergonomics lies in the organization of er-
gonomic research, and of its findings, so that complex systems are
considered. Future research activities in ergonomics are therefore
likely to emphasize models of complex person-machine interactions.
NAME INDEX

Adams, J.M. 240,254 Brown, F.R. 151,157


Alderfer, C.P. 267, 274 Bubb, H. 31,43,55
Aradudin, A. 92 Bubb, P. 40,55
Arora, S. 151, 156 Blichi, E.C. 100
Ayoub, M.M. 147,148,153,155,156, Bunt, H.C. 255
157,199,214,216
Cakir, A. 274
Barlow, H.B. 250,255 Carbonell, J.R. 37,55
Barnes, R.E. 214,216 Card, S.K. 262,274
Barnes, R.M. 146,147,156 Casali, J.G. 194
Baron, S. 74 Cavonius, C.R. 228,233
Bates, Jr., C. 153,156 Chaffee, J.W. 147,156
Beatty, W.F. 187 Chaffin, D.B. 109,110,113,201,216
Beer, J. 201,217 Chaikin, G. 59,60,67,74
Bernotat, R. 43,55,57,73,74,252, Chambers, R.P. 250,254
254 Chapanis, A. 19,29,87,89,92,112,
Biberman, L.M. 231, 233 150,156
Bilodeau, E.A. 56 Christ, R.E. 224,233
Bishop, R.B. 210,215,216 Christensen, J.M. 19,30
Bobbert, A.C. 201,216 Churchill, E. 103,104,105,109,110,
Boller, H. 74 113,114
Booker, R.L. 224,233 Clark, C. 188,194
Booth, J.M. 249,254 Clement, W.F. 56
Borg, G. 210,215,216 Coblentz, A. 112
Bouwman, A.G. 249,254,255 Coburn, R. 19,28,30
Brooks, B.M. 201,216 Connover, D.W. 200, 217

277
278 NAME INDEX

Cook, J.S. 150,156 Gazis, D. 86,92


Corlett, E.N. 210,215,216 Geddes, L.A. 92
Costanza, E.B. 224,233 Gnedenkow, B.W. 37,55
Courtney-Pratt, J.S. 250,254 Gomer, F. 194
Coury, B.G. 209,214,216 Grady, R.M. 156
Craik, K.J.W. 253,254 Graham, D. 56
Crossman, E.R.F.W. 52,55 Grandjean, E. 112,149,150,156,196,
CUshman, W.H. 145 216
Czaja, S. 201,217 Greenberg, L. 201,216
Gruneberg, M. 113
Daenzer, W.F. 19,30 Guttman, H. 73,75,80,81,92
Damon, A. 125,143,200,216
Daniels, E.G. 214,216 Haberfellner, R. 19,20,21,22,30
Darlington, J. 264,265,274 Hacker, W. 266,274
Das, B. 156 Halcomb, C.G. 148,153,155,156,157
Dathe, M. 23,20 Haley, K.B. 112
Deen, R.C. 87,92 Hart, D.J. 274
DeGreene, K.B. 15,23,30,87,89,92 Haslegrave, C.M. 82,85,92
Dempster, W.T. 148,149,156 Herda, S. 264,265,274
DOring, B. 15,23,30,71,75 Herman, R. 92
Donges, E. 71,74 Hertzberg, H.T.E. 100,201,216
Drury, C.G. 112,195,200,204,209, Hilf, H. 32,55
214,216 Hillmann, K. 74
Duket, S. 75 Hoffman, E.R. 52,56
Dunnette, M.D. 92 Hornseth, J.P. 150,157
Dzida, W. 235,254,257,264,265, Hoyos, C. Graf 15,30
274 Hunt, D. 57,73,74
Hutchingson, R.D. 180,194
Easterby, R. 109,110,113 Hutchinson, T.P. 82,92
Eddington, A. 5
Edwards, E. 45,49,55 Ironman, R. 238,254
Edwards, M. 87,92 Itzfeldt, W.D. 274
Elkind, J.I. 39,55
Ely, J.H. 149,156 Jacobsen, K.D. 229,233
Emanuel, A. 147,156 Jahns, D. 64,74
Engel, F.L. 237,254 Jastrzebowski, W. 1
Jensen, R.S. 252,254
Farley, R.R. 146,147,156 Jex, H.R. 39,56
Farrell, R.J. 249,254 Johannsen, G. 40,49,55,71,74
Fechner, C. 74 Jurgens, W. 93,100,110
Felts, W.J.L. 148,156
Ferrell, W.R. 51,71,75 Keesee, R.L. 228,233
Fiala, E. 52,55 Kelly, C.R. 37,43,55,183,194
Fitts, P.N. 53 Kennedy, K.W. 107,110,147,148,
Flaherty, R.T. 236,254 149,153
Frost, G. 183,194 Khalil, T.M. 200,201,216
Kikta, P. 109,113
Gabel, W.C. 148,156 Kinkade, R.G. 112,155,157,180,
Gartner, K.P. 185,194,254 194,196,217,219,234
Garret, J.W. 198,200,216 Kirchner, J.H. 23,20
NAME INDEX 279

Klerk, J.A. 255 Noton, P. 237,254


Knight, J.L. 88,92
Kobrick, J.L. 150,156 Oberquelle, H. 263,274
Kohler, W. 258 Oborne, D.J. 56,112,113,254
Kovalenko, I.N. 37,55 Olson, P.L. 92
Kraiss, K.F. 30,180,194 O'Neill, D. 199,201,202,217
Krampe, H. 37,55 Orlandsky, J. 149,156
Krendel, E.S. 39,56
Kretschmer, E. 101 Pawenski, C.S. 204,216
Kroemer, K.H.E. 75,101,102,108, Petri, C.A. 263,274
109,110,113,114,213,216 Pew, R.W. 71,74,186,194
Kubat, J. 55 Pheasant, S. 199,201,202,217
Poulton, E.C. 37,49,52,56,253,254
Law, C..H. 204,216
Lehmann, G. 150,156 Reynolds, 105,113
Lehmann, K. 60,74 Rigby, L. 200,202,217
Lehnert, G. 75 Robinette, K.M. 104,105,109,110,
Leopold, F.F. 255 113,114
Levis, J.A. 56,112,254 Roebuck, J. 63,75,102,108
Lewin, T. 100 Roethlisberger, F.J. 9
Longford, R.C. 55 Rohmert, W. 30,75
LoPresti, P. 199,216 Roth, J.T. 148,149,157
Luczak, H. 75 Rothery, R. W. 92
Lund, M.W. 150,156 Rouse, W. 71,74,75
Rubarth, B. 201,217
Maddox, M.E. 223,228,231,234 Rfihmann, H. 40,56,115,143,144,146,
Marras, W.S. 108,113 162,178
Mayo, E. 9 Ruffell-Smith, H.P. 201,216
McConville, J.T. 103,104,105, Runge, W. 55
107,109,110,113,114 Rutenfranz, J. 57,64,75
McCormick, E.J. 16,30,180,194
McCrimmon, K.R. 23,30 Saelzer, H. 60,74
McDaniel, J.W. 214,216 Salvendy, G. 92
McFarland, R.A. 112,143,216 Sanders, A.F. 251,255
McLean, J.F. 52,56 Saran, C. 199,217
McRuer, D.T. 39,52,56 Schiflett, S. 194
Meister, D. 16,19,23,30 Schmidtke, H. 1,23,30,42,56,143,
Miller, D. 74 144,159,178,180,194,254
Moraal, J. 30,77,180,194 Schober, H. 144
Moran, T.P. 274 Schulze, B.G. 47,56
Moray, N. 64,74 Schumacher, W. 37,56
Morgan, C.T. 150,151,156 Seifert, D. 37,71,75
Muller, H.F. 255 Shackel, B. 19,23,30,60,75
Murrell, K. 74,112 Sharp, E.D. 150,157
Sheldon, W.H. 101
Naish, J.M. 53,56 Sheppard, D. 240,254
Napier, J.R. 196,217 Sheridan, T.B. 51,55,71,74,75
Nesland, K. 62,74 Shneiderman, B. 246,255
Newell, A. 274 Siegel, A.I. 151,157
North, D.M. 190,194 Singleton, W.T. 17,23,30
280 NAME INDEX

Smallwood, R.D. 37
Snook, S.H. 214,217
Snyder, H.L. 219,223,225,228,231,
234
Spillers, W. 71,75
Squires, P.c. 147,157
Stark, L. 237,254
Stein, W. 74
Steinfeld, E. 201,217
Stewart, T.F.M. 274
Stier, G. 150,156
Stoudt, H.W. 103,113,114,143,149,
157,216
Swain, A. 73,75,80,81,92
Szostak, H. 52,55

Task, H.L. 225,231,234


Thompson, R.R. 168,178
Thomson, R.M. 149,156
Thomson, W. 75,102,108
Tromp, J.K. 247,255
Truijens, C.L. 253,255

Ulate, C. 215,217

Van Cott, H.P. 112,155,157,180,


194,196,217,219,234
Van Katwijk, A.F.V. 246,255
Van Meeteren, A. 235,236,240,250,
251,254,255
Van Nes, F.L. 247,255
Varterasian, J.H. 168,178
Volpert, W. 267,274
Vossel, G. 15,30

Wagenaar, W.A. 238,242,255


Ward, J.S. 201,216
Wechsler, D. 85,92
Wierwille, W.W. 179
Williamson, H. 258,259,274
williges, R.C. 188,194
Woodson, W.E. 112,155,157,180,
189,190,194,196,200,217,219,
234
Wortman, D. 71,75

Zangenmeister, C. 23,30
Zegeer, C.V. 87,92
SUBJECT INDEX

Acceleration, 100 Braking distance,43


Acceptability, 2
Accidents,9 Cathode-ray tube (CRT),220
Anthropometric data,82,83,84,93, Character design,223
111 Clearance, 133, 134
Anthropometric reference systems, Clothing, 88,90,91
93 Coding, 182
Anthropometry , Cognitive ergonomics,258,273
engineering, 101 Compatibility,52,53
functional, 146 Compensatory task,49,50
measurement techniques, 105 Computer language, 10
Arm reach, 145 Consoles,115,126,136,145
Arm rest,164,166 dimensions, 153
Assembly line work,39 segmented, 151
Automatization, 2,269 Contrast,222,223,228,257
sensitivity function,231
Back rest, 164 Control arrangement,182
Back-slap, 166 design, 179,198
Bang-bang controls, 186 display compatibility,180
Bearableness,2,3,4,6 display ratio,180
Behaviour,85,86 feel, 184
Betriebsverfassungsgesetz,4 functions, 12
Body dimensions,106,107 identification, 182
measurement, 7,94 Controls, 179, 195,202
strength, 7 isometric, 183
support, 159 isotonic, 183
templates, 121, 133, 160 cutoff-frequency, 40

281
282 SUBJECT INDEX

Design criteria,3,4,88 Handle surface,202


of workstations,26 Handrails, 195,262
responsibilities, 15 Hard copies,242
theory, 58 Head-up-display,53
Deskilling,258,271,272 Heel-point, 11,160
Dimensionality, 45 Hip-point (H-point) ,111,160
Display, 49 Holographic images,230
amplification, 51 Human error probability, 81
design,228 Humanization, 8
synthetic,40,43
Displays, color,223,224 Illustrations,239
electroluminescent, 221 Information, 77
gas discharge,220 boards,115,128,131,139
liquid crystal,222 density,222
multicolor,221 flow,65
solid-state,223 Interactive systems,260,265
three-dimensional, 229 Interface, functional,259
transient,242 session,259
visual,219 terminal, 259
ISO,8
Environment, physical,64 Isolation,2
social,64
Eye ellipse,111,160 Job content, 1
move men t, 65 design,26
point, 134, 160 enlargemen t, 267
enrichment, 267
Fatigue, 12,70
visual,225 Keystroke model,262
Field experiments,61
of view,116-124 Laboratory experiment,62
Filters,228,229 Learning time,70
Flicker,228 Line of sight,116,150
Foot rest,142,172-175 Luminance,223
Forms,241,247
Function allocation,23 Man-machine interaction,237
-machine-loop, 47
Glare,228 -machine-system,3,15,17,32,58,
Goal elaboration,21 63,80
Gripping postures,196 Manner of control,39
Manning requirements,25
Hand control reach,120-127 Manual control,12,183
forces,212 material handling,202,205
reach capability,119,147 Maps,240
Handle, 195 Measuring machines, 160
angles,214 Mechanization,1,32
design, 198 Mental load,32,65,145
diameter, 199 Models,61,109,111
positions,202 Multi-function controls,188
shape,201 Muscular strength,108
size,198
SUBJECT INDEX 283

Natural frequency,4,164 Software Ergonomics,10,258,267,273


Neck-rest, 175-177 system,257,265
Non-linear system,12 tools,258,260
Solution generation,22
Operation, sequential,36 selection,~3
simultaneous, 37, 39 Standardization,8,223,228,257,271
Standards,8,58,227
Panels, 115,129 Standing support,170-172
Pattern recognition,236 Stereotypes, 181
Perch, 170 Strain,2,9,31
Performance,85,86 Stress,2,31,70
analysis, 32 Subjects,67,103
DDtor,150 Support stool, 170
physical, 12 Survey sampling,103
shaping factors,78,79 Symbol design,223
variability, 82 System dynamic,35
Preview display,51 monitive,49,50
Problem solving,253,258 performance,21,31
Process control,252 Systems, active,48
Pulse rate,68 design, 17,32
engineering, 15
Qualification, 11 ergonomics, 15,17,31

Radar equipment, 9 Tables, 238


Reaction time,40,42 Target detection,251
Reasonableness, 3,5 Task analysis,31
Reference-Point (R-point),160 compensatory, 52
Relationship, causal,266 dynamic, 39
hierarchical-sequential, 266 guiding, 45
instrumental, 265, 271 manner of,47
procedural, 269 performance,89
productive, 266, 272 pursuit, 49, 50, 52
Resolution, 223 stabilizing, 45
static,42
Safety,10,88,95,145 Touch panels,226
belts, 160 Traffic lights,86
Search,237 Transfer function,36,166
Seat,4,162
adjustment, 162, 166 User guidance,13
damping, 168 Utilogs,246
index point (SIP) ,160
reference point (SRP),111,148, VDU window,250
159,160 Verbal communication,235
Seating reference point (SgRP), Vibration,4,9,159,166
160 Viewing angles,150,151
Selection,27 Vigilance, 13
criteria, 103 Visual communication,235
Signal noise discrimination,250 considerations, 149
Software design,258,261 display units (VDUs),219
engineering, 10 field, 150
284 SUBJECT INDEX

Word-processors,245 Work surface height,149


Work area, 146 Workload,7,58,64
chairs, 168-170

You might also like