You are on page 1of 14

PEDOLOGY AS A SCIENCE: THE ROLE OF DATA, MODELS AND THEORIES IN THE STUDY OF

NATURAL SOIL SYSTEMS


J.C. DIJKERMAN
Department of Soil Science and Geology, Agricultural University, Wagenbigen (The
Netherlands)
(Accepted for publication December 4, 1973)

ABSTRACT

Dijkerman, J.C., 1974. Pedology as a science: The role of data, models and theories in the study of natural soil systems.
Geoderma, 11: 73-93.

Pedology is defined as the science that studies the genesis, nature, distribution and use potentiality of soil resources. It
uses all empirical scientific methodology, which is the topic of this paper. The role of systems, models, data and theories
as key concepts of this scientific methodology is discussed within a pedological framework.
Soil is an open system because it loses and receives material and energy at its boundaries. Examples are given of
how the soil system may be subdivided into subsystems suited to various types of pedological research.
The natural soil system is very complex. Therefore pedologists build models as convenient devices for scientific
research. The complex natural system is replaced by a Simpler or more abstract model, which can be more easily
handled either manually or mentally. The nature, function and design of models are illustrated with several examples
from the field of pedology.
Data collection is a selective process. It requires selecting from the multitude of information, which is potentially
available in the soil system, those data which we assume to be relevant for the solution of our problem. The selection of
relevant data is guided by the existing body of pedological knowledge and involves procedures of definition, mea-
surement and classification.
Theories explain the interrelationship between data and make them predictable. The term theory includes hypotheses,
scientific laws and formal theory, each of which is discussed within the content of pedology. Forms of scientific
explanation in pedology are indicated.

PEDOLOGY AS A SCIENCE

Pedology is the science that studies the genesis, nature, distribution and use potentiality of soil resources. Both field
and laboratory research are involved. Pedology is an empirical science in contrast to logic and mathematics which are
formal sciences.
Like most empirical sciences pedology uses a scientific methodology which includes the following stages: (1) the
selection of a system or systems to be studied (e.g., a pedon, a soil landscape body, a soil catena); (2) the measurement
of properties of the system and its interacting environment; (3) the ordering and condensing of the multitude of these
data (e.g. by classification); (4) the explanation of the data by hypotheses; (5) the testing of these hypotheses against
new data; (6) the structuring of "confirmed" hypotheses into scientific laws which together form a body of well-established
formal theory; .and (7) the use of scientific laws in predicting new unknown phenomena.
These stages do not necessarily follow each other in this order. Several feedback mechanisms may be present
between various stages. Some stages may occur simultaneously and some may be entirely absent, depending on the
type of research. For instance in soil survey the primary purpose is not to test a hypothesis, but to determine the nature
and distribution of the soil resources of a certain area. Thus, emphasis will be mainly on stages 1, 2 and 3. In soil-genesis
research the purpose is to explain the development of a certain soil or certain soil features. Emphasis is on setting up
and testing hypotheses (stages 4 and 5). In pedology few, if any, scientific laws have yet been formulated (stage 6).
However, in some cases laws of the basic sciences (e.g., thermodynamics) can be used to explain and predict certain
soil phenomena (stage 7).

SOIL SYSTEMS, THE OBJECTS OF STUDY

Soil is an open system (Jenny, 1941; Simonson, 1959) because it loses and receives material and energy at its
boundaries (see Fig. 1). Like many other open systems it is characterized by what can be called hierarchical orr-
ganization. That means it may be subdivided into smaller and smaller subsystems of decreasing complexity which are
nested into each other to form a systems hierarchy (Whyte et al., 1969). Each system or subsystem is an assemblage of
objects united by some form of interaction or interdependence in such a manner as to form an entirety or whole (Patten,
1971). Table 1 gives examples of soil systems at various levels of organization each suited to a specific branch of
pedological research.
The choice of a system to be studied depends on the objective of the research. Traditionally pedology has been mainly
concerned with the study of various types of systems at level 3, 4 and 5. Working at level 2 are those soil scientists who
are concerned with systems that consist of entire soil landscapes. Examples of this approach that go beyond the normal
Soil Survey activities are the Russian investigations into the geochemistry of landscapes (Parfenova, 1963; Kovda et al.,
1968), the study of the geogenesis of alluvial soils in The Netherlands (e.g., Edelman, 1950), the study of the relationship
between geomorphology and soils in Australia (e.g., Butler, 1959) and in the U.S.A. (e.g., Ruhe and Walker, 1968;
Daniels et al., 1971). Working with much smaller soil systems at level 6 are the micropedologists, who are concerned
with spatial differences in soil on a microscopic scale (Brewer, 1964; Jongerius, 1964). Real progress in pedology
depends on cross fertilization of ideas of pedologists working with soil systems of various levels of organization.

Not only the level of organization is important in the choice of a system to be studied, but also the aspect we want to
concentrate on at that level, Chorley and Kennedy (1971) distinguish between eleven different types of natural systems.
In pedology we are mainly concerned with five of these types, viz. morphological systems, cascading systems,
process-response Systems, control systems and ecosystems. These different types of systems are not mutually
exclusive. Each type of system merely focuses the attention on a different aspect of the same natural system. If we are
interested mainly in morphological properties and their relationships we choose a morphological system, identified mainly
by the existence of significant correlations between its morphological properties. A catena, soil landscape body and soil
horizon are good examples. If on the other hand the interest is mainly on dynamic aspects, we choose a cascading
system, one defined by the path followed by flows of energy or mass. This is especially the realm of soil physics (heat
flow, water flow) and soil chemistry and fertility (nutrient cycle, organic-matter cycle). In pedology we are often interested
in the influence of dynamic processes on morphological properties and vice versa. This is an example of a
process-response system, which is a link between a morphological and a cascading system. Control systems occur in the
use of soils when a process-response system is managed to our advantage, e.g. by irrigation, fertilization, soil
improvement, etc. Finally, the entire ecosystem should be studied, if we are mainly interested in the relationship of soil to
its entire environment, including vegetation and animal life.
MODELS - DEVICES FOR SCIENTIFIC RESEARCH

The natural soil system is very complex. Therefore pedologists build models as convenient devices for collecting,
describing, explaining and predicting data. These models are actually also systems but of a simpler and often more
abstract nature. Thus, instead of using the complex natural system, the pedologist uses a simpler or more abstract model
with a similar structure, which can be more easily handled either manually or mentally (Bertels and Nauta, 1969).
Models can be of many forms, they can serve many purposes and they can be designed by using various kinds of
simplifications or abstractions. Table 11 characterizes models according to their nature, function and design. In the next
paragraphs each of these characteristics of models will be discussed. Several examples will be given from the field of
pedology (see Table 111).

TABLE II
Characterization of models
________________________________________________________________________________________________
Nature of models 1. concrete model
2. conceptual model a. mental model
b. verbal model
c. structural model
d. mathematical model
Function of models 1. observational model
2. experimental model
3. descriptive model
4. explanatory model
5. predictive model

Design of models 1. scale model


2. idealized model
3. analogue model
4. computer simulation model
________________________________________________________________________________________________

Nature of models

Models can be of many forms. They can be real physical objects (concrete models) or abstract concepts (conceptual
models).
(1) Concrete models consist of real physical objects that stand as a model for the system under study. An example is
an experimental soil column used to reproduce soil-leaching processes in the laboratory. Other examples are soil
samples or typical pedons used to represent soil systems that are too large to be studied in their entirety.
(2) Conceptual models consist of abstract concepts created in the human mind. They may be considered mental,
verbal structural (or diagrammatic), or mathematical models, depending on whether the concepts are present as mental
images or are explicitly expressed by verbal descriptions, diagrams or maps, or mathematical equations.
(a) Mental models are images in the human mind, used constantly to deal with the physical world around us. They
represent the ideas we have about how the real soil system is. They are not yet recorded in any way, they only exist in
the human mind. These models are used unconsciously without recognition of the assumptions that are involved.
Moreover, the purpose of a mental model is often obscure and may shift without our being aware of it. These dangers of
the mental model may be overcome by converting it to a verbal, structural or mathematical model, in which assumptions
for and functions of the model can be explicitly indicated.
(b) In verbal models the mental concepts are expressed in words. They are qualitative models lacking the precision
and accuracy necessary for quantitative treatment. Two widely used verbal models in pedology are that of the factors of
soil formation (Jenny, 1941) and that which describes the soil-forming processes as consisting of additions, losses,
transformations, and translocations (Simonson, 1959). Parts of Jenny's model have been converted to mathematical
models, but most of it is still entirely qualitative. Another example of a verbal model is the recently introduced energy
model (Runge, 1973) which states that soil development is a function of two energy vectors, leaching water and
organic-matter production.
(c) In structural models the structure or behaviour of a system is indicated in the form of a simple drawing, diagram or
map. These models require tile introduction of some simplifying concepts and a change of scale. For instance, the soil
map can be regarded as a structural model of the spatial differences among soils at the landscape level. The map
requires a change of scale and the introduction of mapping units as simplifying concepts. A soil profile diagram is a
structural model of the vertical spatial differences of soil at the pedon level. The behaviour of dynamic systems can also
be indicated by structural models. For instance a flow diagram of the organic-matter cycle in a soil and its vegetation is a
structural model of the behaviour of an ecosystem with respect to organic matter. The sequential models, used by Arnold
(1965) to illustrate his multiple working hypotheses for the evolution of a particular soil, are good examples of the use of
structural models in soil genesis.
(d) In mathematical models the system's structure and behaviour are expressed quantitatively in the form of
mathematical equations. In the last decades mathematical models have been used more and more in soil science,
especially in soil chemistry and soil physics. Examples are the ion-exchange equations and the equations of liquid, gas or
heat transport through tile soil. Boast (1973) discusses the application of mathematical modeling techniques for the study
of processes which occur in soil with examples of mathematical models describing the movement of chemicals in soil by
water. Some examples of mathematical models that represent soil-forming processes are: (i) the mathematical formula
which characterizes physical soil ripening of marine and fluviatile sediments when they dry and become soil (Pons and
Zonneveld, 1965); (ii) the humus formulas for organic-matter accumulation and decomposition (Nye and Greenland,
1960; Kortleven, 1963); and (iii) the formula for assessing the degree of decomposition of clay minerals during
podzolization (Verbitskiy and Pontyakhin, 1964).
As more and more processes that occur in soils can be described by mathematical models, a quantitative treatment of
certain soil-forming and soil-behavioural processes is now within the realm of possibility. Promising in this respect are the
techniques of systems analysis which provide the conceptual framework and mathematical means of handling large
amounts of data of complex systems by computer simulation (Kline, 1973).
Most soil-forming and soil-behavioural processes are extremely complex with many physical, chemical and biological
processes involved. Seldom is sufficient basic knowledge available to describe such a process with absolute certainty by
a mathematical model indicating all factors involved. In those cases statistical models may be very helpful (Krumbein
and Graybill, 1965). They are a special type of mathematical model, which in the mathematical formula have a built-in
random component by which the reliability of the prediction made from the model can be evaluated. Statistical models
are especially important to test hypotheses against reality (stage 5 of the scientific method) (Ferrari, 1965). Simple
statistical models have already been used in pedology for a considerable time (see e.g., Jenny, 1941). More recently ad-
vanced statistical models are being used in soil research (Morris, 1970).

Function of models

Models can serve many purposes at all stages in the scientific method. They can be used to collect observational data
(observational model), to collect experimental data (experimental model), to describe the system under study
(descriptive model), or to explain or predict the relationships and behaviour of the system under study (explanatory
model, predictive model). Thus, various types of models of the same system may be set up for various purposes, all
equally useful and each giving a different perspective of the system. The validity of a model should be judged on the
basis of its usefulness for the purpose for which it was intended. However, it is not always easy to distinguish models on
the basis of their function. Some models have more than one function and the function of a model may shift as research
develops.
Observational models are often used in pedology to collect data from natural soil systems that are too large to be
studied in their entirety. These models are usually called samples. Thus the large soil system to be studied is replaced by
a smaller soil system of similar structure called a sample. A sample consists of all sampling units drawn to represent a
single population. For instance a number of pedons (sampling units) may be selected which together form a sample that
stands as a model for a soil landscape body (a population of pedons).
Soil samples brought to the laboratory for analyses are often selected to represent soil horizons, The sampling error
indicates the difference between values obtained by means of the model and true values for the system under study. In
soil research, sampling errors are commonly much greater than laboratory measurement errors (Cline, 1944; Beckett and
Webster, 1971). Sampling errors can be estimated by using a statistically sound sampling plan, e.g., simple random
sample, stratified random sample, or systematic sample (Petersen and Calvin, 1965). The disadvantage of these
sampling plans is that they usually require many soil samples. Therefore, the pedologist often chooses a few "typical"
sites (judgement samples), which agree with his "ideal" concept of the particular soil. The danger of these judgement
samples is that the data obtained from them are often biased and tend to confirm the pedologists "ideal" concept and
enhance his existing ideas. Furthermore, there is no way to estimate the accuracy of the data (sampling error) obtained
from this model. Therefore, although judgement samples are often very useful in the early stages of research, they
cannot be used as observational models in the later stages, when hypotheses need to be tested. Then a sample is
needed that is statistically sound.
Experimental models are used to collect data by simulating soil processes under artificially controlled conditions. In
pedology they are becoming increasingly important (see e.g., Hallsworth and Crawford, 1965). The experimental soil
column used in the laboratory to simulate soil-forming processes is a good example. The danger of this kind of model is
oversimplification. Conditions in soil columns are often so different from those in the field that data obtained by these
experiments cannot always be used as evidence for hypotheses to be applied to natural soil systems.
Descriptive models are used to characterize the system under study. The soil map is one of the most important
descriptive models in pedology. Other examples are the Munsell notation 10YR 5/4, which is a descriptive model of a soil
colour, and the N value, which is a descriptive model of a physical soil-ripening stage.
Explanatory models are used to understand the relationships and behaviour of the system under study. The state
factors of soil formation formerly were widely used as an explanatory model to explain differences among soils at the
landscape level. Currently, however, many pedologists are not content with such an explanation as it does not indicate
the physical, chemical and biological processes involved. The state factors of soil formation are regarded now more as a
descriptive model useful to characterize the soil-forming environment. This example shows how the function of a model
may change when research progresses. Good examples of explanatory models are the stability diagrams of minerals for
various soil environmental conditions. They are based on a theory of thermodynamic equilibrium (see e.g., Mohr et al.,
1972).
Predictive models are used to forecast the relationships and behaviour of the system under study. All explanatory
models are also predictive models. In fact the validity of explanatory models can only be tested by using them as
predictive models. Not all predictive models, however, are also explanatory. Certain empirical relationships may be
established which do not elucidate cause and effect relationships, but can be used to predict on a limited scale. A good
example is the soil catena model used by soil surveyors during mapping to predict soil differences at the landscape level.

Design of models

A model is a simplification or abstraction of the system tinder study so that it can be more easily handled either
manually or mentally for a specific purpose. In the design of models several kinds of simplifications or abstractions may
be used. For instance it may be a change of scale (scale model), the introduction of some ideal assumptions, which
make it easier to analyze the system (idealized model), the use of an analogue medium or model (analogue model), or
the use of a computer to simulate the real system under study (computer simulation model). Often more than one kind of
simplification or abstraction is used in the design of models, and thus the above-mentioned types of models are not
mutually exclusive.
Scale models are models in which the system to be studied is scaled down (or up) to a size or number convenient for
study or comprehension. Scale models are mainly used for observational and experimental purposes. An example of a
scale model that involves a reduction in numbers is the sample (soil sample, pedon). Reduction in size is involved in
aerial photographs. The experimental soil column involves reduction in scale plus also many other simplifications. It
could therefore also be considered an idealized model. The same statements hold for a soil map.
Idealized models. The real system often is so complex that it cannot be easily analyzed. Therefore simplifications are
introduced, either in reality in a laboratory experiment under controlled environmental conditions or on paper, by making
certain assumptions. A good example is the idealized system assumed to be present when applying Stokes' law in the
measurement of particle-size distribution. In field pedology, lithosequences, biosequences, etc., are examples of
idealized models which assume that differences between soils are only caused by variations in one soil-forming factor
while other factors have remained the same. In this way it is possible to study the influence of one factor on soil
formation. Obviously, however, the assumptions often involve gross simplifications.
An idealized model of soil is also used in the quantitative assessment of soil profile development by calculating gains
and losses. It requires several assumptions, which have been discussed by Barshad (1964). Recently the theory of
thermodynamic equilibrium has been used to construct stability diagrams of minerals for various soil environmental
conditions (e.g., Mohr et al, 1972). This requires the use of idealized thermodynamic equilibrium models, which assume
a closed system that is in a state of equilibrium. This obviously is not a valid assumption for soil systems. However, as
Stumm and Morgan (1970) point out: "to be useful a model need not be realistic as long as it produces fruitful
generalizations and valuable insight into the nature of ... processes and improves our ability to describe and to measure
natural ... systems".
Analogue models make use of another medium that is similar in nature but easier to handle. For instance dangerous
medical experiments are usually performed first on animals instead of human beings. For reasons of simplicity
agronomic experiments are often carried out on a few major kinds of soil of the area. These so-called benchmark soils
are then used as analogue models by extrapolating the data to other similar kinds of soil that have not been studied
directly. Another type of analogue model is based on transplantation of a well-developed model from one field of science
to comparable Situations in another branch of science. For instance tile analogy existing between metamorphism and
weathering has been used to apply the well-known facies principle of geology to soil formation (Van der Plas and van
Schuylenborgh, 1970; Chesworth, 1973).
A computer simulation model is a special type of analogue model that makes use of a computer as medium to simulate
the behaviour of tile real system under Study. This model is built by systems analysis, which is a new technique to handle
complex systems of various sciences. Patten (1971, 1972) describes how to build such models and gives examples of
their use with respect to the ecosystem. Computer simulation models can be used for dynamic systems to study the
changes of the system with time under various circumstances. They are Suitable for rather complex systems because,
although the structure of a system needs to be described mathematically, no analytical solution of the equations is
required. The equations are used to calculate the changes of the system over very short time periods during which rate
variables are assumed to be constant.
Computer simulation models are especially useful for feedback systems. These are systems in which the output
influences the input. Forrester (1969) distinguishes between two types of feedback systems, negative and positive. In
systems dominated by negative feedback loops, the output is such that it minimizes the net input. Such systems are goal
seeking, they tend towards an equilibrium situation. Examples in pedology are the processes controlling the
organic-matter equilibrium and the formation of stable geomorphic surfaces. In systems dominated by positive feedback
loops, the output is such that it enhances the input. Such systems often show exponential growth until one of the
parameters of the system changes. Examples in the field of pedology are gully erosion and the formation of textural
subsoil bands. Once a gully or a band starts to form it enhances its own growth (Dijkerman et al., 1967).
Recently simulation models have become important tools in agricultural research (Jones, 1970), geology (Harbaugh
and Bonham-Carter, 1970) and ecology (Patten, 1971, 1972). Various kinds of processes that occur in soils have been
simulated by computer (e.g., De Wit and Van Keulen, 1972; Dutt et al., 1972). Kline (1973) gives examples of computer
simulation models for soil-plant relationships and soil genesis.

DATA COLLECTION, A SELECTIVE PROCESS

Data are obtained by measuring selected properties of the system under study, either directly or through the aid of an
observational or experimental model. Table IV is a schematic representation of sources of pedological data. It shows the
classical division into field and laboratory data, and divides both into two categories according to whether the data arise
from controlled experiments or from natural uncontrolled situations.
In pedology most data are observational. Often their interpretation is difficult because of the great many causal factors
involved. Advanced statistical models may help to sort out the importance of each factor. Experimental data are usually
easier to analyse because few causal factors are allowed to vary during the experiment. Experimental laboratory data are
especially useful in detailed studies of fundamental pedological processes. However, their interpretation for real field
situations can be difficult because of the use of simplified experimental models, Experimental field data are often
essential for predicting the behaviour of a soil for a specified use. They are especially important in soil-survey
interpretation.
Data collection is a selective process. This holds both for observational and experimental data. We receive a multitude
of information, too much to describe and comprehend. The moment that a pedologist begins to describe a soil system he
becomes selective, because it is not possible to describe everything. He describes only that which attracts his attention
and which is significant for the problem under study. Thus, although data collection is an objective process, the choice of
data to be collected is subjective, governed by some a-priori hypothesis concerning what is important and what is not.
Ob-
servation and description are strongly influenced by the existing body of pedological knowledge and understanding. For
instance clay skins, if present, will certainly be recorded in profile descriptions nowadays, whereas thirty years ago they
were seldom mentioned, not because they were not present, but because they were not recognized as significant. They
did not have a place in prevailing hypotheses of soil genesis.
The process of data collection may be subdivided into three parts: definition, measurement and classification (Harvey,
1969). These three procedures act as filters which sort out from the potentially available information in the system, that
which is relevant. In doing so non-relevant information is lost and a first ordering of data is established. This helps in
understanding the system under study. The danger is that the a-priori hypothesis which guided the selection of definition,
measurement and classification procedures turns out to be wrong so that relevant information is lost,

Definition

Properties that need to be measured or observed should be carefully defined in order to avoid confusion and to make a
comparison of data possible. Definition is a procedure for specifying the meanings of terms. It forms the bridges we build
to link perceptual experiences with terms, or terms with each other.
For collecting data operational definitions are essential (Bridgman, 1927). They are definitions that specify the meaning
of a term by indicating a set of operations that have to be carried out in order to receive information about that term.
Thus, they make clear how the measurement or observation needs to be carried out. For instance, if we need to gather
data about the size of soil particles it is necessary to specify clearly what we mean. We can, for instance, measure with a
ruler, especially if gravel-sized particles are involved, but do we measure the longest dimension, or the smallest
dimension, or both? We can also use a series of sieves, or if smaller-sized particles are involved, the pipette method or
the hydrometer, all of which are linked to different concepts of the size of soil particles.
It is a great merit of the American soil classification "Soil Taxonomy" (Soil Survey Staff, 1960, 1967, 1970) that it
attempts to provide precise and unambiguous operational definitions for many commonly used observational and
experimental concepts in pedology. At the same time, however, there is a real danger that the rigid definitions will so
strongly filter the information stream that many relevant data are lost. This is a danger inherent to the use of strictly
defined operational terms. Another example is the system of nomenclature in soil micromorphology developed by Brewer
(1964). Therefore we should look upon operational definitions as only temporary bridges built for a specific purpose and
based upon the existing body of pedological knowledge and methods, As our knowledge increases and our methods of
observation and measurement become more refined, operational definitions will have to be subject to change. For each
specific research problem in pedology we will have to decide whether the existing operational definitions are satisfactory
for our purpose or whether it is better to provide new ones especially adapted to our specific requirements. We should
also bear in mind that transfer of knowledge is greatly hindered by continuous changing of definitions. Therefore, where
possible, preference should be given to existing well-established definitions.

Measurement

Measurement is the process of assigning numerical values to properties of an object according to rules. The rules are
specified in the operational definition of the property. These operations give rise to four scales of measurement, the
nominal, ordinal, interval and ratio scales (Stevens, 1946, 1958). Each scale is a number system with a different
mathematical structure, which is used as a descriptive mathematical model for the property to be measured.
The nominal scale is used for classifying objects in terms of the equality of one of their characteristics. For example,
the shape of soil structural elements (peds) is classified on a nominal scale into blocks, plates, prisms, etc. Instead of
names, numbers may be assigned to the classes, e.g., 1 = blocks, 2 = plates, 3 = prisms, etc. There is no quantitative
relation between the successive numbers. Any class may be given any number. The nominal scale is the most primitive
form of measurement. One can argue that the use of a nominal scale is not really measurement. As Stevens (1946)
points out, it does consist of assigning numerals according to rule. The rule is "do not assign the same numeral to
different classes or different numerals to the same class". However, the numerical values 1, 2, 3, etc. have only
symbolic meaning, one could just as well use a, b, c, etc.
The ordinal scale allows more refinement. It is used when objects can be arranged in some order of rank according to
the magnitude of a property. For instance, the grade of soil structure is given on an ordinal scale 0 = structureless, 1=
weak, 2 = moderate, 3 = strong. Soil consistence is another example. The numbers assigned to ordinal classes allow a
sequence, each successive numeral represents more of a given quality than its predecessor. Successive numerals on
the ordinal scale, however, are not equally spaced. The "length of step" from 1 to 2 is not necessarily the same as from
2 to 3.
In the interval scale the "lengths of steps" are the same but there is no absolute zero point. For example, soil
temperature is measured on the interval scale in oC or oF. A temperature of 8oC is not twice as hot as one of 4oC since
the OoC is arbitrary.
In the ratio scale these relationships do hold, however, because they have an absolute zero point. Examples are the
Kelvin scale for temperature, the scale for time intervals, the measurement of length or thickness, the measurement of
hue, chroma, and value of soil colour by Munsell colour charts, and the measurement of percentages of sand, silt and
clay in soil-texture determination.
When choosing an operational definition for the property we want to measure, the consequences of the definition with
respect to measurement scale should be considered because it determines the type of data-processing techniques that
can be applied, especially the kind of statistical analyses that can be used (see for example Harvey, 1969, fig.1 7.1).
Interval and ratio scales are parametric scales. They provide data which can be used to calculate or estimate certain
parameters (mean and standard deviation) of a population. These characteristic constants are basic to all parametric
statistical tests. On the other hand, nominal and ordinal scales are non-parametric scales to which only non-parametric
statistical methods can be applied.
Formerly in pedology, most field characteristics of soil were measured on nominal or ordinal scales, whereas currently
more and more soil characteristics measurements on the ratio scale are possible. Soil colour is the outstanding example
in this respect. For some properties we have no choice, but for others we do. Shape of soil structural elements, although
usually expressed on the nominal scale, could also be measured on the ratio scale. Krumbein and Graybill (1965) give
several examples of how to express shape on a ratio scale. Soil consistence, usually expressed on an ordinal scale,
could be partly measured on a ratio scale with the aid of a penetrometer (Davidson, 1965).
All measurements are liable to error. Thus data may be wrong or inaccurate. When collecting data it is essential to
know something about the accuracy and precision of the measurements. Accuracy and precision are not the same.
Accuracy refers to the closeness of the measurements to the true value, a reference or standard that is accepted as the
truth. Precision refers to the closeness of the measurements among themselves or their reproducibility (Eisenhart, 1952).
For instance, when several experienced soil surveyors independently estimate the texture of a soil sample, the
agreement among estimates is a measure of the precision of the field-texture measurement, whereas the difference
between the field estimate and laboratory analysis indicates its accuracy. As pedology moves from a more qualitative to
a more quantitative science a better knowledge of the accuracy and precision of pedological data will be essential. An
interesting approach in this direction is that of Beckett (1968), who discusses the precision and accuracy of a soil map.

Classification

Definition, measurement and classification, the three procedures, which are used to filter the potential information
stream in order to extract relevant data, are strongly interconnected. Classification occurs both before and after
measurement. For instance, the measurement of soil texture requires first a classification of the size of soil particles (for
example into sand, silt and clay) after which operational definitions and measurement procedures can be worked out.
After the precentages of sand, silt and clay have been measured, we can condense these raw data into one term by
fitting them into the appropriate textural class, e.g., silt loam.
Classification orders and condenses data. Its purpose is so to organize our knowledge that the properties of objects
may be remembered and their relationships may be understood most easily for a specific objective (Cline, 1949). In
pedology it is used both for soil properties (e.g., soil texture, soil structure) and for entire soil bodies (e.g., pedon, soil
landscape body). In the process of classification we may lose information, presumably information that is irrelevant for
our purpose. In pedological research we cannot simply accept existing classification systems but need to ask if the
purpose and universe for which a classification system was made are sufficiently close to our purpose and universe so
that we will not risk losing relevant information or including too much irrelevant information. For instance, the
soil-structure classification as proposed in the Soil Survey Manual (Soil Survey Staff, 1951) for the universe of American
soils was found to be unsatisfactory for Dutch soils. The apedal geogenetic structures (Jongerius, 1957), which are very
common in Dutch alluvial soils, could only be described as structureless, massive or single grain, which resulted in loss
of useful information. However, if at all possible, preference should be given to existing well-accepted classification
systems, because they facilitate a transfer of knowledge. In some cases if changes seem necessary, such classification
systems can be adapted to a specific purpose by creating a few more subdivisions.

THEORIES, THE EXPLANATION OF DATA

The aim of every empirical science is not only to collect data but to explain them so that interrelationships between
them become clear and predictable. The need for explanation arises from the questions we ask. Asking the right
questions is most important, We usually ask questions when we need to solve a problem or when we react to a set of
data which surprise us because we expected them to be different. It should be realized that the explanation we seek, the
questions that we ask, are strongly influenced by the existing body of theoretical knowledge with which we are
preconditioned by our training. Usually within a period of several years the same kinds of questions are being asked by
scientists of one discipline over and over again for different situations. Major breakthroughs occur only when this tradition
is broken and scientists are no longer content with the traditional answers to the questions or with the type of questions
that are being asked (Harvey, 1969).
In pedology, questions about the genesis of soils have received much attention for many years. However, as the
human population on earth increases, proper use and management of our soil resources become more and more im-
portant. For the future we may therefore expect a shift away from soil genesis towards questions about land evaluation,
soil-survey interpretation, and the behaviour of different kinds of soil under various types of uses.
The answers to our questions are theories which explain the interrelationships between the data. According to
Bridgman (1927), "an explanation (defined in operational terms) consists of reducing a situation to elements with which
we are so familiar that we accept them as a matter of course, so that our Curiosity rests".

Hypotheses, scientific laws and formal theory

The term theory has been used here in its broadest sense for ally explanation of facts. It includes hypotheses, scientific
laws and formal theory. Hypotheses are theories that have not yet been tested extensively. They are often indicated as
working hypotheses, as they may guide the research as to the type of data that need to be collected. Scientific laws are
theories that have been very extensively tested and never rejected. This, however, does not necessarily mean that
scientific laws are always true, because theories can never be absolutely proved; they can only be falsified (Popper,
1965). Formal theory refers to a deductive theoretical framework composed of a logical system of propositions and
conclusions, which link the laws of all entire field of knowledge into one unifying noncontradictory structure.
All theories in pedology are non-formal theories, that is, the reasoning used in explanation does not follow the rules of
formal logic by using deductive reasoning from axioms (propositions). Schelling (1970), however, in a series of
propositions and conclusions has made a first attempt to treat pedological theory in a formal way. Such treatment is very
important as it reveals weaknesses, gaps and contradictions within the existing body of pedological knowledge, which
can be used for formulating future research projects.
In pedology there are very few scientific laws. Perhaps some of tile conclusions in Schelling's formal treatment of
pedological theory should be considered scientific laws. Tile theory about the distribution of soils according to broad
climatic-vegetational zones, which has long dominated the field of pedology, is often referred to as the law of soil
zonality. It is now seriously being questioned. One weakness is tile great number of exceptions. For example Podzols and
several other so-called temperate-region zonal soils have been found to occur in the tropics. To be a law in the sense
accepted in the physical sciences, there should be no exceptions (Harris, 1968). Another weakness is the great variability
within each of the major soil zones, which makes its usefulness as a basis for agronomic practices very limited
(Gorshenin, 1968). Also an important weakness of the law of soil zonality is that it overlooks the factor of time.
Soil-forming processes in tile tropics and in temperate zones are not always greatly different, but it is the age of the soil
which in many cases causes the main differences.
Hypotheses are very common in pedology. The soil surveyor when mapping soils constantly makes use of hypotheses.
Most of the lilies drawn on the soil map are predicted from the correlation between surface features (e.g., topography,
vegetation) and kind of soil. If for a 1: 10,000 detailed soil map a soil surveyor makes four augerholes per hectare, which
is more than usual, and each augerhole has a cross-section of 25 cm2 , then lie verifies his prediction only for 100 cm2 of
the 108 cm2 area (hectare); this is only a 0,0001% sample. For a 1: 50,000 reconnaissance soil map with 8 augerholes
per 100 ha, the prediction is only checked in 2 x10-6% of the total area. Thus, it is clear that the Soil Surveyor needs to
be an excellent scientist who is able to develop and use reliable hypotheses in order to make a good soil map.
Hypotheses are also frequently used in soil-survey interpretation, e.g. judging soil drainage from gley features,
workability from texture and consistence, and permeability from soil-structure characteristics. Another common use of
hypotheses in the field of pedology is the labelling of soil horizons as A1, A2, B and C when describing soil profiles.
These are not data but rather interpretations of data. They represent the hypotheses of the pedologist as to how these
horizons have been formed and how they differ from the parent material. The identification of different geogenetic layers
by Roman numerals I, II, III, etc. also depends on the personal judgement of the soil surveyor.
It is clear from the above examples that hypotheses are widely used in pedology. There is the danger that hypotheses
are used unconsciously and that statements inferred from them are reported without indication of the supporting
evidence.
Many times in pedology more than one hypothesis can be proposed to explain the observed features. A very
open-minded approach in those cases is the method of the multiple working hypotheses, in which several alternative
hypotheses are tested against the available data. Interesting examples of this method in the field of soil genesis are
given by Arnold (1965, 1968). The use of multiple working hypotheses allows one to consider and analyse simultaneously
several rational explanations of a phenomenon and reduces the risk that one hypothesis will become a ruling theory.
The best explanation is often a synthesis of the more acceptable ideas contained in several hypotheses. The use of
multiple working hypotheses could be a more common practice to the benefit of pedology. It could be especially useful in
horizon designations of soil-profile descriptions indicating several possible alternatives. Multiple working hypotheses also
should be used as a basis for a balanced choice of the place of augerhole observations in soil-survey work.

Forms of scientific explanation in pedology

Explanation is any satisfactory answer to a why or how question. In pedology we distinguish between genetic and
functional explanations. The type of explanation that is required depends on the kind of question that is being asked.
A genetic explanation gives answers to the questions: Why does it have these features? How did the system originate
and develop? Such an explanation describes the kind and sequence of events that produced the phenomenon being
explained. It is an important mode of explanation in pedology because a historical sequence of different types of
geogenetic and pedogenetic processes is often used to explain and predict spatial differences of soil. In soil survey and
soil morphology knowledge of how the various soil phenomena have originated commonly provides cities to predict
where one can expect them. However, the hypothesis that, under certain circumstances, space and time can be
considered interchangeable (ergodic hypothesis, see Chorley and Kennedy, 1971), has not been seriously tested in
pedology.
A functional explanation gives answers to the questions: How does the system function? How does it behave under
specified conditions? How can it be controlled? These explanations are essential in soil-survey interpretation, as for
example, in explaining and predicting how a certain soil functions as a medium for plant growth or as a foundation for
roads. Special measures may be needed in order to improve or control the system's function. The functional explanation
will indicate how this can be done most easily.
Explanation consists of relating a situation to statements which we believe to be true. The statements may be axioms
(propositions) as in formal theory but more often they are scientific laws. Thus, in genetic explanations it is not sufficient
to describe only the historical sequence of events. Ill functional explanations a mere description of how a system
operates is not adequate. In both genetic and functional explanations it must be made as clear as possible how the
processes that take place necessarily follow from established scientific laws: these may be either pedological laws or
general laws of the basic sciences. Our most satisfactory explanations are those in which we can relate the pedological
data to the general laws of one of the basic sciences. For example, the laws of thermodynamics have much to offer to
pedology in providing explanations of the weathering of minerals under various soil conditions and the transport of mass
and energy through soil systems. However, it would be naive to suppose that, within the near future, the origin and func-
tioning of such complex systems as natural soil bodies can be explained entirely in terms of laws of the basic sciences.
Consequently, efforts should be made to develop laws which generalize the existing body of pedological knowledge.

ACKNOWLEDGEMENTS

The author wishes to thank Dr. J. Bennerna of the Agricultural University of Wageningen, The Netherlands, Dr. J.
Schelling of the Dutch Soil Survey Institute, Wageningen, The Netherlands, and Dr. L.P. Wilding of tile Ohio State
University, Columbus, Ohio, U.S.A., for their valuable discussions, suggestions and criticisms of ail earlier draft of this
paper.

REFERENCES

Arnold, R.W., 1965. Multiple working hypotheses in soil genesis. Soil Sci. Soc. Am. Proc., 29: 717-725.
Arnold, R.W., 1968. Pedological significance of lithologic discontinuities. Trans. Int. Congr. Soil Sci,, 9th, Adelaide, 4:
595-603.
Barshad, I., 1964. Chemistry of soil development. In: F.E. Bear (Editor), Chemistry of the Soil. A.C.S. Monograph, 160:
1-70.
Beckett, P.H.T., 1968. Method and scale of land resource surveys, in relation to precision and cost. In: G.A. Stewart
(Editor), Land Evaluation. Macmillan, Melbourne, pp. 53-63.
Beckett, P.H.T. and Webster, R., 1971. Soil variability: a review. Soils Fert., 34: 1-15.
Bertels, K. and Nauta, D., 1969. Inleiding tot bet modelbegrip. (Introduction to the concept of model.) W. de Haan,
Bussum, 183 pp.
Boast, C.W., 1973. Modeling the movement of chemicals in soils by water. Soil Sci., 115: 224-230.
Brewer, R., 1964. Fabric and Mineral Analysis of Soils. Wiley, New York, N.Y., 470 pp.
Bridgman, P.W., 1927. The Logic of Modern Physics. Macmillan, New York, N.Y.
Butler, B.E., 1959. Periodic phenomena in landscapes as a basis for soil studies. C.S.I.R.O. Australia, Soil Publ., 14: 20
pp.
Chesworth, W., 1973. The residua system of chemical weathering: a model for the chemical breakdown of silicate rocks
at the surface of the earth. J. Soil Sci., 24: 69-81.
Chorley, R.J. and Kennedy, B.A., 1971. Physical Geography, a Systems Approach. Prentice-Hall, London, 370 pp.
Cline, M.G., 1944. Principles of soil sampling. Soil Sci., 58: 275-288.
Cline, M.G., 1949. Basic principles of soil classification. Soil Sci., 67: 81-91.
Daniels, R.B., Gable, E.E. and Cady, I.G., 1971. The relation between geomorphology and soil morphology and genesis.
Adv. Agron., 23: 51-88. (Edited by N.C. Brady).
Davidson, D.T., 1965. Penetrometer measurements. In: C.E. Black (Editor), Methods of Soil Analysis, 1. Agron. Monogr.,
9: 472-485.
De Wit, C.T. and Van Keulen, H., 197 2. Simulation of Transport Processes in Soils. Pudoc, Wageningen, 100 pp.
Dijkerman, J.C., Cline, M.G. and Olson, G.W., 1967. Properties and genesis of textural subsoil lamellae. Soil Sci., 104:
7-16.
Dutt, G.R., Shaffer, M.J. and Moore, W.J., 1972. Computer simulation model of dynamic biophysicochemical processes
in soils. Tucson, Univ. of Arizona, Agr. Exp. Sta., Techn. Bull., 196: 101 pp.
Edelman, C.H., 1950. Soils of The Netherlands. North-Holland, Amsterdam, 177 pp.
Eisenhart, C., 1952. The reliability of measured values-Part 1, Fundamental concepts. Photogramm. Eng., 18: 542-554.
Ferrari, Th.l., 1965. Models and their testing: consideration of the methodology of agricultural research. Neth. J. Agric.
Sci., 13: 366-377.
Forrester, J.W., 1969. Principles of systems. Wright Allen Press, Cambridge, Mass., 2nd ed., 380 pp. (Text and
workbook).
Gorshenin, K., 1968. Zonality. Sov. Soil Sci., 11: 1559-1561.
Hallsworth, E.G. and Crawford, D.V. (Editors), 1965. Experimental Pedology. Proceedings Univ. of Nottingham Eleventh
Easter School in Agricultural Science, Butterworths, London, 414 pp.
Harbaugh, J.W. and Bonham-Carter, C., 1970. Computer Simulation in Geology. John Wiley, New York, N.Y., 575 pp.
Harris, S.A., 1968. Comments on the validity of the law of zonality. Trans. Int. Congr. - Soil Sci., 9th, Adelaide, 4:
585-593.
Harvey, D., 1969. Explanation in Geography. Edward Arnold, London, 521 pp.
Jenny, H., 1941. Factors of Soil Formation. McGraw-Hill, New York, N.Y., 281 pp.
Jones, J.G.W. (Editor), 1970. The Use of Models in Agricultural and Biological Research. Proceedings of a Symposium.
The Grassland Research Institute, Hurley, Maidenhead Berkshire, 134 pp.
Jongerius, A., 1957. Morphologic investigations of soil structure. Meded. Sticht. Bodemkartering, Bodemkundige Stud., 2
(in Dutch with English summary).
Jongerius, A. (Editor), 1964. Soil Micromorphology. Elsevier, Amsterdam, 590 pp.
Kline, J.R., 1973. Mathematical simulation of soil-plant relationships and soil genesis. Soil Sci., 115: 240-249.
Kortleven, 1., 1963. Quantitative Aspects of Humus Accumulation and Humus Decomposition. Thesis, Agricultural
University, Wageningen. Versl. Landbouwkd. Onderz., No. 69.1 (in Dutch with English summary).
Kovda, V.A., Samoilova, E.M., Vasilevskaya, V.D. and Yakushevskaya, I.V., 1968. Geochemical differentiation of
products of weathering and soil formation on the Russian Great Plain. Trans. Int. Congr. Soil Sci., 9th, Adelaide, 4:
293-301.
Krumbein, W.C. and Graybill, F.A., 1965. An Introduction to Statistical Models in Geology. McGraw-Hill, New York, 475
pp.
Mohr, E.C.J., Van Baren, F.A. and Van Schuylenborgh, J., 1972. Tropical Soils. Van Hoeve, The Hague, 3rd ed., 481 pp.
Morris, J.M., 1970. Multivariate methods in the study of soils. Soils Fert., 33 (4): 313-318,
Nye, P.H. and Greenland, D.I., 1960. The soil under shifting cultivation. Commonw. Bur. Soils Sci., Tech. Commun., 61:
156 pp.
Parfenova, Ye.I., 1963. B.B. Polynov - The creator of the study of the geochemistry of landscapes. Sov. Soil Sci., 2:
111-118.
Patten, B.C., 1971. Systems Analysis and Simulation in Ecology, 1. Academic Press, New York, N.Y., 607 pp.
Patten, B.C., 1972. Systems Analysis and Simulation in Ecology, 2. Academic Press, New York, N.Y., 592 pp.

Petersen, R.G. and Calvin, L.D., 1965. Sampling. In: C.E. Black (Editor), Methods of Soil Analysis. Agron. Monogr., 9:
54-72.
Pons, L.J. and Zonneveld, I.S., 1965. Soil ripening and soil classification. International Institute for Land Reclamation and
Improvement, Publ. 13. Veenman en Zonen, Wageningen.
Popper, K., 1965. The Logic of Scientific Discovery. Harper, New York, N.Y., 479 pp.
Ruhe, R.V. and Walker, P.H., 1968. Hillslope models and soil formation. I.Open systems. Trans. Int. Congr. Soil Sci.,
9th, Adelaide, 4: 551-560.
Runge, E.C.A., 1973. Soil development sequences and energy models. Soil Sci., 115: 183-193.
Schelling, J., 1970. Soil genesis, soil classification and soil survey. Geoderma, 4: 165-193.
Simonson, R.W., 1959. Outline of a generalized theory of soil genesis. Soil Sci. Soc. Am. Proc., 23: 152-156.
Soil Survey Staff, 1951, 1962. Soil Survey Manual, Handbook No.18. U.S. Dept. Agric., Washington, D.C., 503 pp. +
Supplement.
Soil Survey Staff, 1960, 1967, 1970. Soil Taxonomy. U.S. Dept. Agric., Washington, D.C. (in press) (revised 7th
Approximation + Supplements).
Stevens, S.S., 1946. On the theory of scales of measurement. Science, 103: 677-680.

Stevens, S.S., 1958. Measurement and man. Science, 127: 383-390.


Stumm, W. and Morgan, J.J., 1970. Aquatic Chemistry. Wiley Interscience, New York, N.Y., 583 pp.
Van der Plas, L. and Van Schuylenborgh, J., 1970. Petrochemical calculations applied to soils - with special reference to
soil formation. Geoderma, 4: 357-385.
Verbitskiy, P.G. and Pantyakhin, A.I., 1964. Determination of the degree of disintegration of clay minerals during podzol
formation. Sov. Soil Sci., 13: 1396-1399.
Whyte, L.L., Wilson, A.G., Wilson, D. (Editors), 1969. Hierarchical Structures. Proceedings of a Symposium. American
Elsevier, New York, N.Y.

You might also like