VOL. 83, NO.

Bll JOURNALOF GEOPHYSICAL

RESEARCH NOVEMBER
10, 1978

PRESENT-DAY PLATE MOTIONS

J. Bernard Minster

Seismological Laboratory, California Institute of Technology, Pasadena, California 91125

Thomas H. Jordan

Geological Research Division, Scripps Institution of Oceanography

La Jolla, California 92093

Abstract. A data set comprising 110 spread- data set were begun in 1975, and an interim
ing rates, 78 transform fault azimuths, and 142 model was derived [Jordan et al., 1976].
earthquake slip vectors has been inverted to We present in this paper a new relative mo-
yield a new instantaneous plate motion model, tion model, RM2, based on a much improved data
designated Relative Motion 2 (RM2). The model set. Consistent with our previous work, we have
represents a considerable improvement over our attempted to obtain a simple model compatible
previous estimate, RM1 [Minster et al., 1974]. with the available high-quality observations of
The mean averaging interval for the spreading relative motions. Only relative motion data
rate data has been reduced to less than 3 m.y. which involve at least one oceanic plate have
A detailed comparison of RM2with angular veloc- been used, since the data from intracontinental
ity vectors which best fit the data along in- environments exhibit complexities not easily
dividual plate boundaries indicates that RM2 described in terms of rigid plate kinematics
performs close to optimally in most regions, [e.g., Molnar and Tapponnier, 1975]. We have
with several notable exceptions. The model sys- not attempted to model the complex tectonics of
tematically misfits data along the India-Antarc- the western Pacific (e.g., the Philippine
tica and Pacific-India plate boundaries. We plate), becauselittle kinematical information
hypothesize that these discrepancies are mani- is available concerning behind-the-arc spread-
festations of internal deformation within the ing, and the assumptions fundamental to a simple
Indian plate; the data are compatible with plate model (e.g., triple-junction closure) may
northwest-southeast compression across the Nine- not apply.
tyeast Ridge at a rate of about 1 cm/yr. RM2 The value of any model can be judged by its
also fails to satisfy the east-west trending predictive capability and by its ability to
transform fault azimuths observed in the French- withstand the test of new observations.. in this
American Mid-Ocean Undersea Study area, which is respect the success of our original model, RM1,
shown to be a consequence of closure constraints has been mixed. For example, the relative mo-
about the Azores triple junction. Slow movement tion between the North American and the South
between North and South America is required by American plates was predicted by RM1 entirely on
the data set, although the angular velocity vec- the basis of data from other plate boundaries.
tor describing this motion remains poorly con- Although no data yet exist which confirm direct-
strained. The existence of a Bering plate, pos- ly the existence of such relative motion, the
tulated in our previous study, is not necessary model implies that a component of north-south
if we accept the proposal of Engdahland others convergence exists between the South American
that the Aleutian slip vector data are biased by and the Caribbean plates [Jordan, 1975]. It ap-
slab effects. Absolute motion models are de- pears that some convergence is indeed required
rived from several kinematical hypotheses and by recent studies [Talwani et al., 1976; Rial,
compared with the data from hot spot traces 1978],
younger than 10 m.y. Although some of the On the other hand, RM1 failed to satisfy an
models are inconsistent with the Wilson-Morgan extensive set of new data collected in the
hypothesis, the overall resolving power of the South Atlantic Ocean [Forsyth, 1975; Sclater
hot spot data is poor, and the directions of ab- et al., 1976a]. •ne investigation of this
solute motion for the several slower-moving •failure is an important aspect of this study.
plates are not usefully constrained. We show that RM_Iincorrectly predicts the plate
kinematics in the South Atlantic because the
Introduction presently available data are inconsistent with
the plate geometry assumed in deriving RM1. We
PreSent-day plate motions can be modeled demonstrate that this inconsistency can be reme-
using systematic inversion methods. In our ini- died by postulating the existence of internal
tial study [Minster et al., 1974] (referred to deformation within the Indian plate, although
as paper 1), a linearized least squares algo- alternate explanations are possible.
rithm was formulated and applied to an exten- Other problems with the Rbil model have been
sive, globally distributed data set. Angular noted [Jordan et al., 1976]. The well-mapped
velocity vectors for 11 major plates were esti- fracture zones in the French-American Mid-Ocean
mated from these data, and this model was desig- Undersea Study (Famous) area yield an apparent
nated Relative Motion 1 (•M1). The Caribbean azimuth for Africa-North America motion that is
plate was subsequently added to this model by due east [Macdonald and Luyendyk, 1977], whereas
Jordan [1975]. Revisions and additions to the RM1 predicts an azimuth of S79øE, parallel to
the general trends of the nearby major transform
Copyright 1978 by the American Geophysical Union. faults (e.g., the Oceanographer transform fault).

Paper number 8B0653. 5331

0148-0227/78/118B-06S35 01. O0
5332 Minster and Jordan' Present-Day Plate Motions

In RM1 the slip vector data from the North data quality. They are used to weight the data
Pacific were modeled using a Bering plate whose in the inversion algorithm and to derive esti-
motion differs from that of North America. mates of the uncertainties in the model para-
Engdahl et al. [1977] have demonstrated that the meters. Although we have attempted to use a
focal mechanisms from this region can be affec- consistent set of criteria in assigning these
ted by slab structure, perhaps biasing the ob- errors, the estimates are nevertheless crude
servations. They have suggested that correc- indicators of data quality. With this in mind,
tions for this bias may eliminate the need for we have adopted a conservative stand and have
a Bering plate. These and other problems are deliberately overestimated these uncertainties.
examined in this paper. This bias is apparent in Figure 3, where it is
seen that the sample standard deviation of the
The Revised Data Set normalized residual distribution is significant-
ly less than its expected value of 1.
The 330 data used in this study are listed in
Table 1. The data locations are shown in Fig- Model RM2: General Description
ure 1, delineating the major plate boundaries.
These relative motion data comprise 110 rates of Inversion of the data was performed using the
sea floor spreading derived from magnetic anom- linearized, iterative, weighted least squares
aly profiles, 78 transform fault azimuths, and algorithm described in paper 1. Our extensive
142 earthquake slip vectors. In compiling and experience in applying this algorithm to the
editing this data set we have generally followed plate motion problem has demonstrated to us its
the guidelines in paper 1. In particular, we effectiveness. Although the algorithm involves
have excluded data from diffuse plate bounda- the linearization of a nonlinear problem, con-
ries, specifically continent-continent bounda- vergence has always been rapid, and no diffi-
ries. Therefore the details of Asian and Indo- culties associated with local minima have been
nesian tectonics are not represented by our evident. The uncertainties in the model para-
model. meters derived from the linear theory have
Rate data have been determined directly from proven to be effective measures of the errors
published magnetic anomaly profiles using the induced on the model by errors in the data.
time scale of Talwani et al. [1971]. In paper The inversion algorithm has been applied to
1, anomalies 3 and 5 were generally used to es- the data set listed in Table 1 to obtain an
timate rates; we thus averaged the plate 11-plate model, designated RM2. The plate geom-
speeds over the last 5-10 m.y. In this study etry is identical to that of RM1, except that
we have redetermined the spreading rates using the Bering plate has been recombined with the
anomalies 2 and 2' in every instance, except North American plate and a Caribbean plate has
for a few slowly spreading profiles where the been added. RM1, supplemented with the CARB-
anomalies out to 3 were employed. Hence the NOAMangular velocity vector derived by Jordan
mean averaging interval for the rate data is [1975], was used as a starting model in the
less than 3 m.y. In most cases the rates were inversion algorithm. Convergence was attained
determined by comparing the corrected profiles in five iterations.
with synthetics, generally those published by Model RM2 is specified in Table 2 by its geo-
the authors of the original observational Study. h•dron [McKenzie and Parker, 1974]. Although a
However, for the anomaly profiles along the more compact specifica,tion is possible, this
Pacific-Antarctic Ridge [Molnar et al., 1975] we format conveniently provides an explicit rela-
generated our own synthetics. For the several tive rotation vector f6r each plate boundary.
studies where a direct inversion for magnetiza- The RM2 geohedron is illustrated in Figure 2.
tion was made [Macdonald, 1977; Macdonald and In the notation of paper 1 the quantity
Holcombe, 1978; McGregor et al., 1977], the minimized by the fitting procedure is the
original authors' results were used directly. variable

In paper 1 the directions of plate motion im- N [aø

i - ai (m)]
~
plied by earthquake focal mechanisms were esti-
mated by projecting the slip vectors onto a i=l 0. 2
horizontal plane. Although this procedure is
almost universally adopted, it is only approx- where N = 330 is the total number of data. The
imately correct for shallow thrust events in 11-plate model is specified by 30 parameters.
subduction zones with oblique convergence, and If the data were normally distributed and the
it can introduce a slight bias. In this study variances were exactly known, X2 would be chi-
•he more exact procedure of rotating the slip square distributed with 300 degrees of freedom,
vectors into the horizontal plane was employed and a sample value would lie in the interval
for earthquakes along inclined seismic zones. (300 + 49) 95%of the time. The value of X2
This problem is discussed in the appendix. for RM2 is 109, almost a factor of 3 less than
The most precise estimates of relative motion its expected value. Thus the data are fit sig-
direction are the azimuths of well-mapped trans- nificantly better than they would be if their
form faults. In determining these azimuths we assigned uncertainties were correct.
have used detailed bathymetric surveys where This fact is also evident from the histo-
available, relying on contours which cross grams of normalized residuals plotted in Fig-
charted ship tracks. Interpretive diagrams ure 3. The sample variances of these distribu-
have been avoided to minimize the feedback tions are about 1/3 their expected value of uni-
between data and plate tectonic models. ty. This discrepancy could be corrected by uni-
The uncertainties listed in Table 1 are based formly reducing the standard errors assigned to
on a case-by-case subjective evaluation of the the databy a factor of (3)-«. Sucha reduction
 Minster and Jordan- Present-Day Plate Motions 5333

TABLE 1, THE RH2 DATA SET.

LAT,N LON.E DATkY S.O. HODEL RESID, IHP, REFERE&CE

LAT.K LCW.E D*TUP S.D, HODEL RESID, IMP, REFERENCE

*•* KAZC SOAK emac

23.40 -108.50 5.80 0,40 5.73 0.07 0.405 LARSON K CHASE (19701

24,10 -1•g.00 N50b 10,0 N51k -0.9 0,040 RUSNAK ET SV-10,70

-6.90 -78,60
-SO.•O N738
N60E 15.0 N82E
10.0 N81E 9.0 0.045
Z1,3 0.021ABE
STAUDER
(1972 ( •9751
25.00 -109.60 N50k 5,0 N50N 0,2 0,161
26,80 -111,20 N53# 10,0 N47M 5,6 0,040 -14.•0 -75.80
-21.80 -70.DO N•SE[0.0
Ne6F 15.0 N778
NeOE-8.5
-8.9 O.O',7
O.O2O STAUDER
S•AUDER •1•75
1973
29.30 -113.60 N45k 5.0 N44W 005 00157 -23,10 -70.10 N748 10,0 N788 3,6 0,046
-24.10 -70,30 N81E 15.0 N78E -3,3 0,020
$V 23.10 -107,40 k55k 10.0 N52W 2.7 0.040 MCLNAR (1973I -25.50 -70.70 N89E 15.0 N78E -11.1 0.018
25.30 -104.20 N45b IO,O N50W -4,8 0,041 -•7,40 -70.90 N738 10,0 N788 5.0 0,044
26,20 -110,40 N45k 15.0 N48M -3,4 0.018 -30.00 -71.50 S80F 20.0 N78E -21.7 0.011
26,30 -110,20 N4qk 10,0 N48W 0,6 0,041 SYKES (1970 -30,60 -71.40 N87E 20.0 N78E -8.8 0.011
28,50 -112,10 N42k 15,0 N45W -3,4 0,018 KCLNAR (1973) -30.60 -71.50 N79E 20.0 N78E -0.7 0.011
2q.70 -ll].70 N45W 10.0 N44# 1.3 0.040 SYKES (1970
31,70 -114.40 kSlk IO,O NiIW 9,7 0,040 -30,70-71.20
-31,50-71.00 N73F
N82E 15.0 N788
15,0 N78E -4,5,• 0.019
0,019
50,80 -130.00 •'lik 15.0 N19# -4.2 0,015 TOBIN & SYKES (1968) -33.50 -71.40 N88E 20.0 N78E -q.5 0.011
54.10 -132.60 N26W 10.0 NI6W 9.8 0.034 HODGSGN & HILNE (1951) -34.00 -72.20 N748 20.0 N79E 4,6 0.011 #
58.30-136.90 N22k 10.0 NI3W 8.5 0.034 STAUDER (19601
61,05 -147.50 N24W 10,0 NIB# 6.3 0,029 3TAUDER 6 BetLINGER (19661 -37,80
-38,10 -73,40
-73.40 N81E
S79E 15,0 N79E
20.0 N79E -1,9 •,018
-21.9 .010
-38.10 -73,00 H80E 10.0 N79E -1.1 0.042
56.68
56.60 :•52.2o
52.90 N•7• 10.0
FI23W lO.O N26W
•25• [.9 0.024
-2.5 0.024 -" -38.20 -73.20 N81E 15.0 N79E -2.0 0.018
56.50 -154.40 N25W lO.O N2•W -1.5 0.02• -38.50 -73.50 N70E 20,0 N7qE 9.[ 0.0[0
55.70 -155.80 N24W 10.0 N28• -4.0 0.023 -46.30 -74.80 N878 20.0 N79E -7.5 0.011 #
53.10-167.60 N)IW 20.0 N36k -5.4 •.005 STAUDER (1•68
51,40 179,10 FI]BW 20,0 N45# -6,5 0,004 STAUOER I 1968 e•*• •AZC ANTA
52.10 175,70 •l•7N 20.0 N46N -9,2 0.004
R& -40.50 -42.00 7,60 0,80 6.64 0.91 0.246 KLITGCRD ET AL.I19•31
52.30
51.90[34,60
74.30 F'28W
N2qH20.0
20,0 N476
N47H-18,8
-17.9 0.004
0,004 #
53,03 171.11 N34W 20,0 N48M -14,5 9.004 CORHIER (1975l TF -41,30 -88,50 K80E 10,0 N85E 5,2 0,058 KLITGCRD ET &L.(19733
54,28 168.23 N45W 20,0 N50k -4,8 0,004 #
57,80 163,50 FieOk 10,0 N52W 8,4 0.017 STAUDER & MUALCHIN 119761 •V -36,20 -100.90 Ne2E 10.0 SE6E 12.4 0.061 ANDERSON ET AL,(19741
56,00 163,30 N59k 10,0 N52• 6,7 0,017 # -36,30 -97,20 F7qE 20.0 S88E 12.6 0.015
44.20 156.10 N50h 10,0 N58M -8,1 0,015 -36,40 -47.20 N89F 15.0 S888 2,6 0.026
47,20 153,80 •'54k 10,0 N60W -5,8 0.015 -36.50 -47.50 S878 10.0 S888 -1.2 0.060
47.10 153.60 N58k lO.O N60W -1.9 0.015 " -41,70 -84.00 N86E lO.O N82E -4.0 0.060
46.50 151.20 •61k 10.0 N60M 0.8
45.60 150.90 •60k 10.0 N62M -1.6 .).015 e*ee PCFC ANTA eeee
44.q0 150.10 N59k 10.0 N62M -3.2 0.014
44.90 149.50 N60k 10.0 N63W -2.5 0.014
44.70 150.70 N49k 10.0 N62k -12.8 0.014 # •A-35.80
-35,60•10.90
10.90 10.600.40
[0.30 0.40 10,44
10.43 0.16 0,t11
-0.13 0,11o HERRON (197•
HCLNAR ET AL.(19751
44.60 150.t0 •60• 10.0 N62M -2.1 0.014 -36,00 -lll,00 10.20 0,50 10,42 -0,22 0,070
44,30 149.50 NSBE 10,0 N63W -4.6 0,01• # -4•.00 -111.20 q.90 0.40 10.11 -0.21 0,095
44.00 liB.tO N59k 10.0 NbIW -4.4 0.014 -44.40 -112.20 10.00 0.40 9.94 0,06 O.08q
43.90 148.70 Nsqw 10.0 N63W -4.1 O.0ll -47.60 -112.qO q.80 0.40 q.70 0.10 0.084
4t.60 148.00 F159k 10.0 N64W -4.5 0.014 -51.40 -11e.10 •.50 0.50 q,31 0,19 O,U50
-53,70 -118.00 9.80 0.50 q.lO 0.70 0.049
-54.50 -lie.70 8.80 0.50 q.01 -0.21 0.049
-55,•0 -12[,20 q,10 0,50 8,40 0.20 0,0•
17.•0 -105.40 8.90 0.40 8.99 -O.OS 0.139 SCLATER ET AL.I1971! -54.50 -13•.20 8.80 0.50 8.54 0.2• O.ObO
16.70 -105.20 q.20 0.40 9.24 -0.04 0.133 " -56.00 -145.00 8,40 0.50 8.17 0,23 0.052
16.00 -105,20 q.40 0,40 q,51 -0,11 0,126 -58.50 -149.00 8.00 0.50 7.•6 0.24 0,056
14.qo -104.30 10.10 0.40 9.98 0.12 0.120 -54.60 -151.30 7.70 0.60 7.56 0.14 0.040 "
-60,50 -151,00 7,80 0.50 7.47 0.33 0,058 PITHAK ET ALo(19681
4.10-104,0010,400,40 10,30 0,10 0,11• 63.20 -163.10 6.00 0.70 6.67 -0.67 0.035 MOLNAR ET AL,([9751
2,80-104,00
12.50 -104.0010.80
11,00 0,40 10,80
0.40 10,91 O,OO
0,0• O, 13 HERROk (1•72! -64.10 -164.00 5,60 0,50 6.30 -0,70 0.073
11,00 -101,80 11,50 0,40 11,44 0,01 0,117 SCLATER ET AL,(1971! -65.00 -174.00 5.80 0.50 5.96 -0.[• 0.078 PITYAK ET AL.(19681

TF 16.00 -105.00 N80E 10.0 N85E 5.2 0.037 LARSOK g CHASE (19701 TF -49.80 -115.00 N72k 5.0 N74W -2.0 0.036 NCLNAR ET AL.(19753
q.40 -103.50 •848 3,0 N8•E 0.0 0.235 ROSENCAHL -53.00 -118.50 N70W 5.0 N72N -1.9 0.038
-54.•0 -114,00 H68k 5,0 N71W 3.4 0.040
-55.?0 -125,00 N68k 3.0 N69W -0,5 0,110 "
-55.50 -130.00 N67W 3.0 N66# 0.8 0.110
56,20 -143,00 N60k 3,0 N60W -0,2 0,113
-54.80 -[50.50 N63k 5.0 N55N 7,8 0.047
SV 17.30 -100.10 N448 15.0 N378 -7.3 0.032 qOLNAR g SYKES (1969{ -64.50 -170.50 N48k 5.0 N42W 5.9 0.069
16,30 -95,40 N36E 10.0 N35E -1.4 0.051 -64.50 175,20 H34k 3.0 N35N -1,2 0,247
16.00 -qT,qO N40 c lO.O N37E -3.1 0.059
lS,qO -o6.•0 N396 15,0 N36E -3,5 0,023 •V -56.00 -[Z3.40 N67M 10.0 N6qw -2.0 0.010 MOLNAR ET AL.(1975)
-54.R0 -136.00 N61W 10.0 N64W -2.8 0.010
COCO CAPB ee** -56,60 -142.50 k•lb lO.O N60M 0.7 0.010
-65.70 -179.30 NS1# 15.0 N36W 14.7 0.009
12.50 -87,40 N348 15,0 N278 -7,4 4,030 MOLNAR & SYKES (19691
11.40 - 86.20 •'328 10.0 N29E' -4.4 O.ObO " *•** PCFC INOI
8.90 -83.40 N306 15.0 N2qE -0.5 0.021
SV -24.70 -176.70 N88# 10.0 N83N 5.5 0.022 ISACKS ET AL.([9691
-24.•0 -177.20 N70¬ lO.O N82W -ll.q 0.02S
-29.q0 -177.80 N76W 20.0 N82W -6.5 0.007
-24.90 -177.70 N86W 20.0 N82W 3.6 0.007 #
-30,10 -177.20 N73W 10.0 N82W -8.9 0,024
TF 16.40 -87.00 N7gE 3.0 N738 0.7 0.412 UCI•UPl ( 19731 -30,20 -177.80 S88W 20.0 N82W %6 0.007 #
16,90 -85,00 N75E 3,0 N75E -0.2 0.272 #
-30.50 -178.10 N88# 20,0 N83k 5,4 0,008 JCHNSCN K •LhAR (19723
17.50 -•3.00 N7qE 3.0 N778 -2.1 .).188 HOLCOPBE ET AL.119•31
-30.70 -178.30 N76W 15.0 N83# -6.8 0.014
19.20 -80.00 N7c;E 3.0 N80E 1.2 J.172 "
-30,70 -178,30 N78W 15,0 N83W -4,8 0,014
lq,70 -77,00 N83F 3,0 N83E 0,3 0,•81 •
-30.40 -177.60 N90W 20.0 N82W 7.9 0.008
19.70 - 75,00 NBtE 3.0 N85E -0,7 0,416 a
-32,20 -178.10 N•OW 20.0 N82W 7.8 0.008
-33.•0 -179.00 S72k 20,0 N83W 25,2 0,009 #
SV 15.27 -84,25 h666 10,0 N70E 4.2 0.0o0 KANAHORi & STEWAPT (19781
-49.00 163.70 S56• 25.0 S67N 10.9 0.024 8•NGHAR & SYKES (1969)
16.80 -85.90 N70E 20.0 N746 3.9 0.007 MOLNAR & SYKES
-56,20 158,20 S OE 20.0 S35W 35,0 0,059
18.qO -81.20 NelE 15.0 N7OE -2.1 0.006
lq,10 -64,80 5826 15,0 584E -2,1 0,074 'L
17,70 -61,60 N!OE 15,0 9816 9.3 0,106

Ra 72.10 I.O0 t.60 0.40 1,59 0,01 0.171 VOGT ET AL.(19703

70,00 -IS.00 1.80 0,30 1.73 0.07 0.259
2.30 -qq,60 4.40 0.30 4.56 -O. ie 0.280 HEY (1974I 61,60 -27.00 1,qO 0.30 2.01 -0.11 0.[68 TALWANI ET AL,I1971J
2.40 -•8,70 4.90 0,30 4,71 0.19 0,245 61.00 -ZB.00 Z,00 0.30 Z.03 -0.03 0./63
2.t0 -•6.00 5.00 0.30 5.16 -0.!6 0.160 60,00 -zq.40 Z.00 0.30 2,06 -0,06 0.[5•
2.50 -94.00 5.50 0.30 5.48 0.02 0.120 46,00 -2T.SO 2.TO 0.50 Z,3• 0,34 0,0•3 PITNAN
2.40 -93.00 5.60 0'30 5.64 -0.04 O.IU7 44,00 -28.50 2.80 0.50 2,40 0,40 0.045
0.90 -87.00 6,40 0.40 6.60 -0,20 0.017 40,00 -30.00 2.80 0.50 2.46 0.34 0.0•0 KRAUSE K ¬ATKIWS 11970J
0,80 -86,10 6,60 0,30 6,•3 -0,13 0,157 "
O.qO -eE.0q 7,00 0.20 6,73 0ø27 0.355 KLITGCRD• MUCIE 119741 TF 79.00 2.50 S528 10o0 S548 -2.0 0.069 JOkNSCN ET AL.(197Z{
O.qO -86,00 6,20 0.40 6.?5 -0.55 0.090 HEY (19741 71.00 -8.00 S658 5.0 S668 -1,4 0.204
3,30 -83,20 7,20 0,30 7,11 0,09 0,237 "
66,50
52.50 -20,00
-33.50 S828
S858 lO,O
3.0 S75• 7,2
S83 1.5 0,037AVERY
0.316 OLIVET ET
ET AL,(
19683
TF 1.40 -85.30 F' 58 5.0 N 78 2.2 0.200 LCNSOALE & KLITGORD (19781
2.50 -84.50 N 68 3.0 N 68 -0.2 0,532 # SV 79.80 2,60 S438 10.0 S538 -10.2 0,071 CONANT( 19721
T9,80 2.40 S478 10.0 S538 -6,3 0,070
SV •.00 -40.50 N IW 10.0 N 78 8.1 0.064 FORSYTH (19721 80,20 -1.00 S528 !0,0 S568 -3,5 0.070
1,50 -85,30 NOE 20,0 N 78 ?,1 0,012 MCLNAR & SYKES (19691 70.90 -/,60 $658 10,0 S668 -1,Z 0,051
66.70 -18.20 S658 10.0 S748 -8.8 0.038
eeee NAZC PCFC •"•'• 66.30 -14.80 S738 10.0 S758 -1.7 0.037 SYKES
52.80 -34.30 S858 10.0 S848 1.2 0,029 SOLOMON 119731

-5,80
-9.40 -106,80
-110.00 15,30
16.30 0,80
0.80 15,43
15.93 -0,13
0.37 0,071REA
0.073 REA •1976
1976
-q.qo -110.10 15.50 0.80 15.98 -0.48 0.074
-10,80 -llO,30 16,60 1,00 16.08 0,52 0,048
-11,40 -110,50 16.10 1.00 16,14 -0.04 0.049 ß
TF 37,00 -ZZ.00 S84W 5.0 S86W 2.4 0,783 LAUGHTON ET AL,(19751
-12.00 -110.80 15.90 0.60 16.21 -0.31 0.138
-19,00 -ill.00 16.50 1,00 16,78 -O.Z8 0,064 HERRON (1972I SV 37.50 -24.70 S82k 15.0 S75M -6.7 0.145 MCKENZIE (19721
-20.00 -111.80 16.10 0.60 16,84 -0,74 0,187 REA • 8LAKELY (1975) 37,60 -Z4.70 S89W 20.0 S75M -13.6 0,079 UOI&S ET AL.(19761
-28,00 -112,00 17,50 0,80 17,10 0,40 0,145 HERRON 11972{ 37,40 -19,00 Se3h 20.0 N81W 16,3 0,065
35.90 17.60 N63W 20.0 N73M -9.9 0.140 HAOLEY K KANAYORI (1575)
TF -3,70 -102.40 S818 3.0 S818 0.3 0.138 LOWSDALE11978 36,10 -10,60 N42W 15.0 N47# -5.2 0,545 UOIAS ET AL.I1976)
-4,50 -105,50 9788 3,0 STqE -1,2 0,141 LONSDALE 11978 A) 36,20 -7.60 N3q• 25,0 N40N -0,5 0,193 HCKENZIE (19721
-6,00 -107.00 S738 5,0 S788 -5.5 0.050 HANHERICKXET AL,11975)
-13.50 -11Z.00 S70E 10.0 S768 -6.4 0.012

SV -4.40 -105,90 S758 20,0 S798 -4,0 0,003 ANDERSON ET AL,11974{

-4.50 106.00 S768 15.0 S798 -2,9 0.006 ANDERSON & SCLATER 119721
-4.60 -105.80 S778 15.0 S7qE -2.0 0,006 RA 36,80 -33,30 2,20 0,30 2,31 -O,ll 0,166 MACDONALD (19771
-13.30 -111.50 S75• 20.0 36,00 -34,00 Z,40 0,30 2,34 0,06 0.165 61RO K PHILLIPS (19751
-26.60 -114.10 S75E 15o0 778 -0.9
768 -1.7 0.005
0.003ANDERSON ET AL,(1974{ 27.50
26.20
-44.30
-44.80
2.50 0.30 2.59 -0.09 0.190 PHILLIPS (19671
-26,40
-28,70
-113,60
-112.70
S71E
S626
10,0
20,0
S76E
S778
-5,2
-14,6
0,011
0,003 # 22.50 -45,00 ,40
ß80 0,30
0.40 2,62-0,2•
2,6S 0,1[ 0,200
0.130 MCGREGOR
PHILLIPS ET AL,(19773
119671
5334 Minster and Jordan' Present-Day Plate Motions

TABLE 1, ICONT tNUED ! TAELE 1. ICONTINUEOl

LAT.k L•N.E DATUM S.D. MODEL RESID. IIM•. REFER El•,E LAT. k LCk. E DATUM S,O. MODEL RESID, ll•. REFERENCE

TF 36.90 -33.10 S88E 10.0 S78E 9.9 0.053 MACDONALD & LUYENDYK 11977l
36.60 -33.,0 S89E 10.0 S78E 10.9 0.053 '
19.30 39.00 1.60 0.40 1.51 0.09 0.162 ALLAN & MCRELLI (1970)
23.90
18050 -46.00 S84E
-46.80 S85E 11.0.0S80E
I0.0 S80E 404.7•00067
0.073FOX ETAL.(196•J)
HEEZEN & THARP 18.00 40.00 1.60 0.40 1.58 0.02 0.128
14.80 54.80 •.12 0.20 2.12 0.00 0.087 LAUGHTON ET AL.(/9701
SV 35030 -36010 N90E 10.0 $78E 11.6 0.059 UDIAS ET AL.(1S/6) 14.70 55.60 2.20 0.20 2.14 0.06 0.108
23.90 -46.00 $76E 15.0 S80E -309 0.032 SYKES 41967) 14.60 5•.00 2.2• 0.20 2./5 0.11 0,121 "
14.40 •3.60 2.0• 0.20 2.0• -0.03 0.067
14,30 5•.50 2,20 0.20 2,17 0.03
13.70 57.30 2.22 0.20 2.21 O.O1 0.182
RA -6.00 o11.70 3.70 0.40 3.82 -0.12 0.115 VAN ANDEL ET 13.40 50.7• 1.96 0.10 2.02 -0.06 0.18/
-7.60 13.40 3.80 0.40 3.84 -0.04 0./13 1•.20 50,90 2.00 0.10 2.03 -0.03 0./86
3.80 0.40 3.86 -0.06 12,10 4•.80 /.90 0,10 1.89 0.0/ 0.309
-8.40
-9.20 -[3.30
3.20 3.90 0.40 0.1083..87 0.03 12.00 45.60 1.92 0.10 /.89 0.03 0,3/8

-24.90-13.00
-28.30 -13.00 ;.50
.90 0.40
0.40 3,95 -0.03
3.93 0.55 0.072
0.076DICKSCN
ETAL.1/9681
-30.50 -14.00 4.00 0.20 3,91 0.09 0.278 TC [2.00
/3.50 46.00
51.50 N30E
N35r 5.g N30E
10. N34E 0.4 0,236
-1,4 0.698LAUGHlON
" ETAL.(19701
-38.20 -15.00 4.00 0.50 3.80 0.20 0.042
-54.20 -1.20 3.10 0.30 3.29 -0./9 0./42 SCLATER ET AL.11976 SV 14,00 51.70 N•OE 15.0 N30E -0.1 0.077 SYKES (1970
-54.70 -1.00 3.20 0.30 3.21 -0.07 0./44

TF 10.80 -4Z. 30 S88E 2.0 S88E 0.4 0.506 EITTREIM & EWI!•G 1/9751
lO.;O -40.90 Se6F 5.0 S88E -2.3 0.075 HEEZEN & THARP 11965)
9,40 -40,00 S88E 5,0 S89E -0,7 0,070 Fq• •ATE• (RAI. UNITS ARE CM/Y•.
8.80 -38.70 S88E tO.O S89E -1.3 0.016 FC• TR•kSFCRM FAUL ? (TF! AN• SLIP VECTOR ($VJ AZIMUTHS/ UNITS ARE DEGREES.
7.60 -36.60 S89E 10.0 N90E -1.3 0.015 "
7.20 -34.30 S89E IO.O N89E -2.4
4.00 -31.90 Ne8F 10.0 N88E -0.4 U.O/2 "
1.90 -30.60 N86F 10.0 N87E 1.1 0.012
1.10 -26.00 N86E IO.O N85E -0.9 0.0/2
-0,10 -18,00 NTTE 5,0 N82E 4,S 0,062 would not change the model but would decrease
:•./o
ß 30 -24.oo NTSE
-14.50 NS/E/0.0
3.0 N84E
N81E 3.4 0.194
5.5 0.0•2 the derived model uncertainties by the same
-1.90 -/2.90 N82E 10.0 N80E -2.0 0.019
factor. However, to be conservative, we have
-2.90 -/2.50
-7.00 -12,50 N79E
NI3E10.0
5.0 N80E :.9
N80E .1 0.018
0.069 ¾AN ANDEL ET AL.11973)
retained the larger estimates of uncertainty.
SV 10.80 -43.30 NqOE 10.0 S87E 2.9 0.02/ SYKES I 19671
-0.20 - 18.70 N88E lO.O NSZE -5.S 0.015 It can be seen from Figure 3 that the distri-
-0.50 -19.90
-54.30 -2.40 N85E
N66E 10.0 N83E
10.0 N74E -i.•.1 0.022
0.014FORSYTH 11975I bution of normalized residuals for the slip vec-
see. ANTA SOAM
tor data departs from the assumed Gaussian be-
havior in another manner: the distribution is
RA -55.30 -1.70 2.00 0.40 1.94 0.06 0.082 SCLATER ET AL.I/976
56.10 -4.70 1.90 0.40 1.90 0.00 0,085 skewed toward negative values. Much of this
TI: -55.70 -3.00 S84F 3.0 S86E -2.0 O.bO8 SCLATER ET AL.(1976 skewhess is attributable to the predominantly
SV -55.50
-58.30
-Z.60
-15,30
588E
{vqoE
10.0
15.0
S86E
S86E
1.9
4.•
0.055
0.026
FORSYTH I 1975)
"
negative residuals exhibited by the slip vectors
-60.80
-60.60
-24.60
-25.70
S74E
S73E
10,0
10.0
9858
S85E
-11,3
-12.3
0,084
0.086
'
'
from the Aleutians and the Kurils, a feature
-60.80 - 1•;. 70 S82E 15.0 S85E -3.3 0.032 discussed in more detail below.

**** AFPC ANTA

B•.cause the data set is large and because the
geometry of the problem is complex, the perfor-
RA
-54,70
-53,90 0,0
3050 •,60
,40 0.40 1,59
0,40 1,58-0o19
0.020,063
0,062.SCLATER ETALeII.976 mance of RM2 cannot be fully described by these
-52.20 38.80
-44.20 14,50 •.TO
.60 0.30 1.62
0.20 1.65 0.080.1/2
-0.05 0.268 NORTON
8ERGH & NORTON 1/9761 simple. statistics. A complete assessment of
-44.30 39.00 /.60 0.40 1.65 -O.OS 0.067 SCHLICH
-38,80 47030 1.60 0.30 1o65 -0005 0012b RM2's success in explaining the observations re-
T;-52.2
-54,3• 1•.00
1,80 N36E
NI•E 5,0
5.0 N45E
N3•E 0.9 0,181
-2.5 0.12, SCLATER
NORTON ( ET AL,11976
1976J quires that each data subset pertaining to an
-46.00 35,20 NITE 3,0 N16E -L,O 0,234 8ERGH & NGRTDN li9161
individual plate boundary be considered sepa-
-43.70 3,.30 N/TE 5.0 N13E-:.• O.Oa7 ' rately. For a large number of plate pairs a
relative rotation vector, or at least a best
SV -54,60 1.70 N47E 10.0 N45E -1.9 0.046 FORSYTH 11975)
54.40 5.90 N40E 10.0 NILE /.1 O.O*l " fitting pole (BFP), can be determined from that
-53090 8.70 N47E 1000 N39E -8.5 0.037 NORTON (1976)
-50,90 2g.10 NZ6E 15.0 N2/E -5,1 0,010 data subset alone. These vectors and poles have
-48,80 31,50 N/7E 20.0 N/SE 1,9 0,005 SYKES I i970)
-45,60 34.10 NIIE 15.0 N/IE -0.2 0,009 NORTON ( 1976I been obtained by inversion and are listed in
-45,40 35,00 N16E 15,0 N/6E 0,/ 0,009
-45,50
-44,90
35.10
35,70
NI. SE
N25E
15,0
1.5.Q
N/6E
NI6E
/,0
-9,4
0,009
0,009
'
"
Table 3. The corresponding BFP's are shown with
-38,90 46,20 NI. IE 15,0 N 9E -5.,3 0,0/2 ' the RM1 and RM2 poles on Figures 4-6. The dif-
-'36.20 52.50 N 5E 20.0 N 5E 0,2 0,008
ferences between these poles and those for RM2
measure .the constraints imposed on RM2 by the
simultaneous inversion scheme. These differ-
?,qO60.00
6.8059,10 1.40
.70 0,40 2.12
0.40 1.95 0.58
0.450,107
3.105 MCKENZIE
g SCLATER
1197/I
5,10 61,60 2.70 0.40 2,3• 0.]/ 0.100 ences are not large, which is evidence that RM2
4,10 62,60 2,80 0.30 2,56 0.24 0./73
-20.30 66.50 4.S0 0.40 4.85 O.Ob 0.07/ performs close to optimally in most regions.
-•l.5O 68.50 4,90 0,40 5,03 -0o13 0,071
-21.80 e8.70 5.00 0.40 5.06 -0,06 0.072 Notable exceptions involve the INDI-ANTA, INDI-
-22.00 68.90 4,90 0.40 5.08 -0.18 0.072
-24.30 70.00 5.20 0.40 5.30 -0.10 0.07• PCFC, and AFRC-NOAMpoles, discussed below.
TF 11.00
0.50
57.50
67.00
N•OE
N43E
5.0
5.0
N3IE
N40E
0.7
-2.8
0.420
0.090
LAUGHTON 4L970)
FISHER ET AL,(197/)
The estimated model uncertainties o0,
-5.50 68.50 N45E 5.0 NIbE 1.1 0.064 and o• are much smaller in Table 2 than in Ta-
-9.00 67.30 N52E 3.0 NSIE -0.7 0,165 ENGEL & FISHER
-13.50 66.50 NS7E 3,0 NS6E -0,9 0.153 ble 3. This is, of course, a direct consequence
/6,00 66,50 N60E 5,0 NS8E -2,1 O,Ob2 FISHER ET AL,(1971I
-17.40 66.20 Nb2E 3.0 NS9E -2.9 0.142 ENGEL & FISHER 11975) of tl•e self-consistency constraints inherent to
-20.00 67.00 k60E /0.0 N60E -0.4 0.0[2 FISHER ET AL.ILg7/)
the rigid plate model, as discussed in paper 1.
An impressive example of this behavior is pro-
RA -28.00 74.00 6.10 0,50 6.58 -0.48 0.052 MCKENZIE & SCLATER (/9711 vided by the COCO-PCFCrotation vector, which
-2•.50
-41.00
75ø00
79.00
6.10
6.80
0.40
0.30
6.68
7.18
-0.58
-0.3•
0.080
0./43
SCLATER
SCPLICH
ET AL.11976 83
is heavily constrained by two triple-junction
-41.00
-42.00
el,00
90.00
7,30
7.40
0.60
0.60
7.22
7.38
0.08
0.02
J,036
0.038
MCKENZIE & SCLATER 119711
closure conditions; these constraints reduce
-44.00
-50.00
93.00
114.00
7.20
7.50
0.40
0.40
7.44
7.40
-0,24
O./O
0.088
0.114 WEISSEL • HAYES 119721
th• •. nominal uncertainty of the rotation rate by
-50.50 134,00 7,40 0.40 6.97 0,43 0,142 a factor of 4.
-62.50 157.80 6.50 0.40 6.32 0,18 O./b9 FALCOKER 119721
-62.40 158.10 6.80 0.40 6.31 0.49 0./59
It should be emphasized that the uncertain-
T; -39.50
36.50
78.50
79.00
N42E
N48•
L5.0
15.0
N45E
N43E
2.5
-4.6
0.007
0.007
MCKENZIE & SCLATER 119711
ties in the model parameters given in Table 2
-39,50
-43.00
80,50
e4.50
N39E
N34E
15,0
15.0
N43E
N40E
3,9
6.4
0,007
0.006
correspond to marginal distributions. A com-
-46.00
-49.00
96,00
124.00
N29E
N/3F
/5.0
5,0
N3ZE
N/ZE
3.0
-1.4
0.006
0.055 WEISSEL & HAYES 1/9741
plete description of the model uncertainties,
-58.00 149.00 N/qk lO.O N/2• 6.7 0.018 HAYES & CCNNOLLY
including the various error cross-correlations,
-60.00 155.00
-61.80 153.00 NI9•
N26k 10,0"•:N/7•
lS,b N20W 1.8
.8 0.0/9 FALCONER
0.009
ß
119721
requires the specification of a 30 x 30 (sym-
SV -45.80 q6.iO N/TE 10.0 N32'E 14.9 0.013 8ANGHAR
& SYKESI19691 metric) variance matrix. A more complete dis-
-55.20 146.10 NI3• 15.0 N 8• 4.8 0.008 N
-55.30 146,20 N22• 25.0 N 8# 13,• 0.003 N
cussion of this point is given in paper 1.
 Minster and Jordan: Present-Day Plate Motions 5335

O* 3 O* 60* 90* 120•' 150' IAO*

6(•*

40*

20*

20*

-40*

-60*

Fig. 1. Plate geometryand geographicaldistribution of the data used in producing

model RM2. Circles are sea floor spreading rates, squares are transform faults, and
triangles are slip vectors. Seven EURA-NOAM data at high latitudes are not shown
on the figure.

Listed in Table 1 are quantities which we tions in the data functionals due to geometri-
have termed 'data importances.' As defined in cal complexities and allow the models to be
paper 1, they are the diagonal elements of an plotted as smooth lines on the diagrams. More
orthogonal projection operator in the data space important, the deviations from the locally best
and are indicative of the distribution of infor- fitting parameters required by closure conditions
mation among the data (paper 1, and Minster et are readily apparent.
al. [1977]). Importances are additive and sum The Pacific-North America boundary. It was
to the number of inverted parameters, 30 in the concluded in paper 1 that the slip vector data
case of RM2. They depend on the geometry of the along the Aleutian-Kuril trench system are not
data set, and on the data uncertainties, but not consistent wi•h the NOAM-PCFC relative motion
on the actual values of the data. The final inferred from data in the Gulf of California and
model depends heavily on the most important data in the northwest Pacific. We suggested that this
and is robust with respect to the least impor- inconsistency was diagnostic of deformation of
tant data.
the North American plate and attempted to model
Cumulative importances for individual plate it by including a hypothetical Bering plate in
boundaries are listed by data types in Table 2 RM1. However, the BERI-PCFC pole was determined
for RM2 and in Table 3 for the best fitting by only 10 slip vectors. Engdahl et al. [1977]
vectors. The cumulative importance for all slip pointed out that our data were a poor representa-
vector data is only 4.6, compared with 11.1 for tation of the earthquake population along the
the transform fault azimuths, despite the fact trench and that the slip vector orientations for
that the former outnumber the latter by nearly individual events in the vicinity of 175'E could
2:1. This reflects the lower uncertainties-- be significantly biased by the laterally hetero-
by a factor of 2 to 3--generally assigned to geneous seismic velocity structure of the down-
transform fault data. The most important datum going slab. In the present study the number of
(0.95) is the rate across the Mid-Cayman Rise data along this trench system has been increased
[Macdonald and Holcombe, 1978]; alone, it es- to 27, including 15 high-quality slip vectors
sentially determines the relative speed of NOAM- from the Kuril-Kamchatka Arc recently published
CARB. When the entire data set is considered, by Stauder and Mualchin [1976]. Because of the
50% of the cumulative importance is associated evidence for bias due to slab structure pre-
with the 49 most important data, and only 10% sented by Engdahl et al. [1977], we assigned
with the 151 least important data. Importances large uncertainties (+20*) to the data lying
are very useful for a detailed comparison of between 165'E and 165'W longitude. It can be
data and models, as is illustrated in the next seen from Figure 7 that these data are in fact
section. systematically misfit by RM2 and the BFP in the
direction observed in paper 1 and predicted by
Model RM2: Detailed Assessment the model of Engdahl et al. [1977]. On the other
hand, data from the Kuril-Kamchatka Arc are fit-
This discussion is devoted to a detailed ted by the model without difficulty, in agree-
evaluation of RM2 on a region-by-region basis. ment with the conclusion of Engdahl et al. [1977]
The fit of RM1 and RM2 to the data for indi- that slip vectors in this region are not likely
vidual plate boundaries is illustrated in to be significantly biased by slab structure.
Figures 7-20. The data and model values are Since the fit of the data elsewhere along the
depicted as residuals with respect to the best boundary is satisfactory (Figure 7), we conclude
fitting angular velocity vectors and poles that there is little evidence for deformation
listed in Table 3. Base lines provided by the within the North American plate of the sort
best fitting vectors remove the large varia- hypothesized in paper 1.
5336 Minster and Jordan' Present Day Plate Motions

•• •0•0• • •0•0• •• 0

•oo •o•o•• o •-o• o• •o •

ß ß ß ,, ß ß ß ß ,,

q-•
Q o

00000 0000000000000 000000000

o la

I ,--I I I I ,--I ,--I I ,--I I ,--I I I I I I I

I I I I

r..D
r..D
r..D
r..D
r..D
r..D
000 •
 Minster and Jordan: Present-Day Plate Motions 5337

N• X N;• X

'rN

Fig. 2. RM2 geohedron (stereo pair). The geohedron depicts relative motions in
angular velocity space [McKenzie and Parker, 1974]. Individual plates correspond
to vertices. The z axis coincides with the rotation axis of the earth, and the x
axis is along the Greenwich meridian. Vectors representing the three reference axes
have magnitudes of 0.5ø/m.y. Open circle is coordinate origin for AM0-2. Closed circle
is coordinate origin for AM1-2.

The East Pacific Rise. The data set for the and Mudie [1974]. As seen in Figure 10 and in
COCO-PCFCboundary includes a redetermination Table l, the data along this bodndary are in-
of the Siqueiros transform fault azimuth from ternally consistent. A particularly satisfying
revised bathymetry [Rosendahl, 1976]. RM2 per- feature is that the recent bathymetry of
forms very well along this boundary and consti- Lonsdale and Klitgord [1978] clearly requires
tutes a slight improvement over RM1 (Figure 8). the COCO-NAZCpole to lie north of the equator;
The data set for the NAZC-PCFC boundary has the transforms at 84.5 ø and 85.3øW trend east
been significantly revised and augmented, es- of north. The implied shift from the RM1 pole
pecially the rate data set. Between 6 ø and
12øS the magnetics are poor, and the data
relatively scattered (Figure 9), as might be ex- Rotes
pected for east-west profiles in the vicinity of
themagnetic equator. Nevertheless, Rea's
[1976a, b] data indicate a lower rate than was
used in paper 1. Herron's [1972] profile at 19øS
is easily readable, despite the small size of the
published figure, but the bathymetry indicates
that a fracture zone may be crossed to the west -i.o i.o

of the ridge. Thus the western part of the pro-

Tronsform Fpults
file is suspect beyond anomaly 2, and we assigned n =78
m =0.066
a large uncertainty to the measurement. A se- = .

quence of high-quality profiles at 20øS has been

discussed by Rea and Blakely [1975]. Since
their published profiles are rate adjusted and
could not be remeasured, we adopted their esti-
-i.o i.o
mated spreading rate (16.1 cm/yr) and assigned
it an uncertainty of 0.6 cm/yr, a conservative Slip Vectors
n=142
value in view of the datum's quality. However, m=-0.065 i•l
this rate is less than that obtained at 19øS and - ,

is not fit well by the model. It is also dif-

ficult to reconcile this rate with the comparable
rates. much further north and a higher rate to
the south: the profile at 28øS [Herron, 1972]
yields a rate which exceeds 17 cm/yr.
The azimuths along the NAZC-PCFCboundary
have been much improved by the recent 'bathymet-
ric studies of Mammerickx et al. [1975] and
Lonsdale [1978a, b]. However, the position of
the NAZC-PCFC pole has not been significantly Fig. 3. Histogramsof normalizedresidualS"?:'.:f:br
altered by these revisions; the •M1 and •M2 each data type, with samplesize, samplemean,?•-
poles, and the BFP, lie very Close together, and sample variance. The theoretically ideal
well within the RM2 error ellipse. Gaussian distributions with zero mean and unit
The Galapagosspreading center. The rate variance are shown for comparison. Shaded area
data along the COCO-NAZC boundary are taken in lower histogram represents residuals for
from the study by Hey [1974]. We also included Aleutian and Kuril slip vectors, which show
a good deep-tow profile published by Klitgord negative bias.
5338 Minster and Jordan' Present-Day Plate Motions

position is in complete concordance with the

shift dictated by the lower spreading rates
along the NAZC-PCFC boundary. It should be
noted that the strike of the Panama transform
fault is very consistent with the new pole
00000000000000(:3 position, a point we shall return to in the
next section.
The Chile Rise. The slower opening rate
along the NAZC-PCFC boundary also affects the
motion along the Chile Rise. in particular, the
RM2 rate is considerably less than the 7.6-cm/yr
estimate derived from the profile of Klitgord
et al. [1973], which we consider to be the best
rate observation along this boundary and is the
only value included in the data set. However,
the RM2 rate is between this value and the lower
estimates of Morgan et al. [1968] and Herron and
Hayes [1969].
Eastern Pacific subduction zones. Strongly
coupled to the opening of the East Pacific Rise
are the convergence rates and directions along
the middle American and South American trenches.
We have adopted a set of slip vectors estimated
by Stauder [1973, 1975] and Abe [1972] to repre-
sent the direction of subduction in South
America. The residuals for these data show a
slightly negative trend, although Abe's [1972]
well-determined solution has a large positive
residual. The negative trend could be eliminated
by increasing the rate along the NAZC-PCFC
o o o o o o o o o o o o o o o o boundary. However, the COCO-NOAMand COCO-CARB
slip. vectors also exhibit this negative residual
trend, and the possibility that these data are
o
biased, like the Aleutian slip vectors, cannot
be discounted. In any case, the scatter in the
r'-,-• •D o C-4 •.[•-3' •"• •-. ,• O")...T . •-• ,-• data is large, the average misfit is small, and
Oc•O . 00000000 - O0
the data importances are low; hence any bias
will not significantly affect the model.
o

AFRC
SOAM

CARB
SOAM

ß RMI
ß RM2
+ BFP

Fig. 4. Poles for model RM2, with their 95%

(20) confidence ellipses. P•M1poles and best
fitting poles (Table 3), where available, are
also shown.
 Minster and Jordan' Present-Day Plate Motions 5559

-150 ø -120 ø -90 ø _60 ø -30 ø NOAM-PCFO

30
I I I I I I i i i i i
_ -

2O
. .

zc '"'-"-/ 60 ø

-•o
, •. _
cocoL_, -20

-30
-

I I :, I I I I I\ I I I
.

140 160 180 -160 -140 -120 -I00

30 ø Longitude, deg E.

Fig. 7. Data and models for NOAM-PCFC shown as

residuals with respect to predicted values cal-
-GARB 'SOAM• culated from best fitting angular velocity vec-
tors. Azimuths are measured in degrees counter-
o clockwise; rates are in centimeters per year.
Circles are sea floor spreading rates, squares

• COCO
NAZC are transform faults,
vectors.
listed
and triangles
Error bars are subjective
in Table 1.
are slip
and are
The solid line is RM2 (this
study) and the dashed line is RM1 (paper 1).
' NOAM Here the dashed-dotted line corresponds to the
Bering-Pacific pole determined in paper 1.
CARB
* RMI ponent of apparently asymmetric spreading is ob-
- ß RM2 served on many profiles [Molnar et al., 1975;
Stein et al., 1977], so the rates were estimated
+ BFP only from pairs of corresponding anomalies on
both sides of the axis. All measurements were
60 ø
based on anomaly 2' or younger anomalies.
Fig. 5. Same as Figure 4 for 150ø-30øW, 60øS- Transform fault azimuths were derived from the
70øN. bathymetry of Molnar et al. [1975], but esti-
mates were obtained from ship track crossings
The Pacific-Antarctic Ridge. Because it is a rather than their interpretive map. It is clear
keystone of the global model, particular atten- from Figure !1 that P•2 is very close to the
tion was devoted to the PCFC-ANTA boundary. The best fitting vector and represents an improvement
data along this boundary are of sufficient num- over RM1 in this region. The difference in the
ber and quality to provide significant coupling, RM1 and RM2 poles is mainly attributable to the
via the Antarctic plate, among the plates in southwesternmost transform fault, an important
the Pacific and the plates with boundaries in datum (• = 0.25) not included in paper !. Some
the South Atlantic and Indian oceans. The con- internal inconsistency of unknown origin is evi-
figuration of the PCFC-ANTA boundary has been dent in the rate data (Figure !1): the rates
investigated by Molnar et al. [1975], and our are greater in the middle of the boundary than
data set is based primarily on this study. those required by the rates at the ends of the
Since these authors did not use synthetic mag- boundary. Nevertheless, most of the data are
netic profiles, we computed synthetics and re- fitted within their uncertainties, and the rela-
interpreted the magnetics. A significant cam- tive rotation vector is one of the best deter-
mined in the RM2 geohedron.
The India-Pacific boundary. The data used
_$0 o 0o 30 ø 60 ø

40 ø
GOGO-PCFC
i • i i i' i i •- I
.

ii
20 ø i .

i I

0o .

I i I i i I I I
-0.8 -0.4 0 0.4 0.8 -20 -I0 0 I0 20
Rate, cm/yr. Azimuth, deg.
Fig. 6. Same as Figure 4 for 50øW-70øE, 30øS-
50øN. Fig. 8. Same as Figure 7 for COCO-PCFC.
5540 Mñnster and Jordan' Present-Day Plate Motñons

NAZC-PCFC

30 40 50 60 70
30

2O

•0
-I.6 -0.8 0 0.8 1.6 -20 -I 0 0 I0 20
Rote, cm/yr. Azimuth, deg. o
Fig. 9. Same as Figure 7 for NAZC-PCFC. -I0

-20
along this boundary, consisting entirely of
earthquake slip vectors, are the same as those -30
in paper 1, but the data north of 25øS were
Distance from 8FP, deg.
eliminated because of documented behind-the-arc
spreading in the Lau Basin [e.g., Lawyer et al., Fig. 11. Same as Figure 7 for PCFC-ANTA.
1976]. Nevertheless, the geometry is such that
a BFP could be determined from the 14 remaining Ridge system, in disagreement with the hypothe-
slip vectors (Table 3). We observe that this sis of Falconer [1973] that this segment is a
best fitting pole is almost identical to the strike slip fault. We strongly suspect that
pole determined by Falconer [1973] exclusively these inconsistencies result from internal de-
from seismicity data along the Macquarie Ridge, formation within the Indian plate (see below).
a completely independent data set. However, as Motions about the Azores triple junction.
seen in Figures 4 and 12, both RM1 and RM2 dif- The plate boundaries which form the Azores
fer significantly from this pole, a direct re- triple junction are individually well con-
sult of requiring closure around the INDI-PCFC- strained. Figure 13 is a residual plot for the
ANTA triple junction. Consequently, the global northern Mid-Atlantic Ridge data. The longi-
models are a poor fit to the southernmost slip tude of the EURA-NOAMpole is reasonably well
vectors, determined by Banghar and Sykes [1969]. fixed by the precise azimuth data along the
Furthermore, these models predict a significant Charlie-Gibbs transform fault and a number of
component of compression across the Macquarie fault plane solutions in the Arctic, but its
latitude is more uncertain. Both the RM1 pole
and the BFP lie near the mouth of the Lena
COCO-NAZC

0.8
PCFC- NDI
-20 I I I I I I ' I I • I i I
-•0.4

0
I
-0.4 -3O -- -. ¾- ß --

-0.8 i' •. ..

- I -
-I00 -95 -90 -85 I
I I I :1 /
2O
- / -
!
& I0 - / -
//
- t+ -
-2O
I I I I'
-30 -20 -I0 0 I0 20 30
Longitude, deg E. Azimuth. deg.

Fig. 10. Same as Figure 7 for COCO-NAZC. Fig. 12. Same as Figure 7 for PCFC-INDI.
 Minster and Jordan' Present-Day Plate Motions 5341

EURA-NOAM AFRC-NOAM
0.8-i I ' I ' I ' I ' IJ ! I I I , I I[ I I 40
-

J \ - 30
i-
2- :1 = II -_20
•,- i • i • • • i i
- I0

-
-

-0.8 -0.4 0 0.4 0.8 -20 -I0 0 I0 20

-0.8 -

Rate, cm/yr. Azimuth, deg.

I , I • I , I • I
30 40 50 60 70 Fig. 15. Same as Figure 7 for AFRC-NOAM.
2O
The data set south of the Azores on the Mid-
Atlantic Ridge has also been improved. Several
special studies have yielded muchbetter mag-
netics, and these imply a significantly lower
rate during the last 3 m.y. than was used in
paper 1. The azimuth data along the AFRC-NOAM
boundaryhave also been revised. In paper 1
-2O the general trends of the Oceanographertrans-
form fault (S77øE) and the Atlantis transform
Distance from BFP, deg. fault (S81øE)were used and were well fit by
RM1. In the present data set these azimuths
Fig. 13. Sameas Figure 7 for EURA-NOAM. have been deleted and replaced by the azimuths
of transforms A (S88øE) and B (S89øE) in the
River, the position most compatible with the Famousarea [Macdonaldand Luyendyk, 1977]. The
rate data. The RM2pole is several degrees difference betweenthe azimuthsof the major
further south (65.8øN, 132.4øE), and its fit to transform faults and transforms A and B has been
the rate data south of 60øNis not as good. attributed to a change in the direction of plate
However, this pole is more consistent with the motion within the last 5 m.y. [Macdonald,1977;
conclusions reached by Chapmanand Solomon Fox et al., 1978; Atwater and Macdonald, 1977].
[1976] in their study of northeast Asian tec- A slip vector showing east-west motion on the
tonics.
Oceanographertransform fault [Ud•as et al.,
The data set along the Azores-Gibraltar line 1976], supporting this conclusion, has also
is considerably improved over our previous been included.
study. We deleted the datum east of Gibraltar
The revised data along the AFRC-NOAM
boundary
because of its probable involvement with the are internally consistent, as indicated by the
Alboran plate [Andrieux et al., 1971] but added performanceof the best fitting angular velocity
three new slip vectors west of Gibraltar. The vector, but the AFRC-NOAM azimuth data are poor-
most important addition, however, is the azimuth ly fitted by RM2 (Figure 15). It is clear that
of the Gloria transform fault (•--0.783), well the misfit is forced by the closure condition
defined by Laughton et al. [1972, 1975]. This about the Azores triple junction. To satisfy
datumplaces a strong constraint on the longi- the triple junction condition, the AFRC-NOAM
tude of the AFP.
C-EURApole. Although the indi- pole mustbe on the great circle connectingthe
vidual slip vectors are not particularly well EURA-NOAM
and AFRC-EURA
poles (Figures 4 and 6).
determined, their variation from northwest com- The BFP is not; it lies to the west near the
pression on the east to southwest extension on
northeastern tip of Greenland, as is required by
the west requires that the pole be not far south the revised azimuth data. The triple-junction
of the boundary, a conclusion established by great circle cannot be shifted to include the
McKenzie[1972]. As a result, the pole is very
tightly constrained, and the RM2 solution is AFRC-SOAM
very close to the BFP (Figures 6 and 14). .ll

I I I I I I I I
I I I 'i I I I I
AFRC-EURA .
-

3O I J I 1

- _J
2O _ _

- r - - I -
- I - I -
- [ - - I -
I • I
- I -
' _1 - I -
- I 1
-2O

-3O
-
-
-
-
I ! I
==/
!
/-t - I
I
I
[
! ! I
-
-
' -
-
-
I I I •
l
I
i

I
i I I
-

-25 -20 -15 -I0 -0.8 -0.4 0 0.4 0.8 -20 -I0 0 I0 20
Longitude, deg E. Rate, cm/yr. Azimuth, deg.

Fig. 14. Same

as Figure7 for AFRC-EURA. Fig. 16. Same
as Figure7 for AFRC-SOAM.
5342 Minster and Jordan' Present-Day Plate Motions

CARB-NOAM INDI-AFRC
I I I I I I I I
2O _

-2O

-90 -80 -70 -60

Longitude, deg E.
--

Fig. 17. Same as Figure 7 for CARB-NOAM.

-
AFRC-NOAM BFP without seriously misfitting the
data along one or both of the other boundaries.
For example, any good fit to both the AFRC-NOAM - i i it:. i -i I I i i .
and the EURA-NOAMdata set yields an AFRC-EURA -0.8 -0.4 0 0.4 0.8 -20 -I0 0 I0 20
pole that is much to the west of the RM2 pole and Rote, cm/yr. Azimuth, deg.
implies compressive motion along the entire
Azores-Gibraltar line, a prediction in flagrant Fig. 19. Same as Figure 7 for INDI-AFRC. The
disagreement with the observed earthquake mech- dotted line represents the instantaneous motion
model of McKenzie and Sclater [1971].
anisms. Hence the RM2 solution is significantly
different from the AFRC-NOAM BFP. The RM1 and
P•M2 poles are each included within the other's connecting the EURA-NOAMand AFRC-EURA poles is
95% confidence ellipses. Both models predict fixed by truly 'instantaneous' data, i.e., the
directions of AFRC-NOAM motion which match the slip vectors in the North Atlantic and along the
observed general trend of the Oceanographer Azores-Gibraltar Line. Therefore the conflict is
transform fault but which misfit the azimuths among data which involve little or no time
of transforms A and B by about 10 ø. averaging.
A possible explanation for this discrepancy Perhaps the east-west transforms observed in
concerns the way the RM2 data set averages over the Famous area are not unbiased indicators of
time. It is conceivable that the east-west AFRC-NOAMmotion. This would be the case, for
trends observed in the Famous region are so example, if these short fault segments were leaky
recent that the pole shifts required by this in the sense of Menard and Atwater [1969], i.e.,
reorientation are not represented in the data
from the other plate boundaries. INDI-ANTA
However, we believe that this explanation can
0.8
be rejected. The location of the great circle

AFRC- ANTA
I I I I I I I
0.8- -
ß
--

• 0 •':
-0.4 -
:....... ,t.
..........
T -
-0.8
- _

-0ß8 - - 80 I00 120 140 160

I I I I I I I i i i i i i i i i

0 •0 40 BO 2O
i I --

ß[
.
" i ii
- ! t
-- _.

-20
-2O
ß I i I I I I i i i i i i i i i

Longitude, deg E. Longitude, deg E.

Fig. 18. Same as Figure 7 for AFRC-ANTA. The Fig. 20. Same as Figure 7 for INDI-AFRC. The
dotted line represents the instantaneous motion dotted line represents the instantaneous motion
model of McKenzie and Sclater [1971]. model of McKenzie and Sclater [1971].
 Minster and Jordan: Present-Day Plate Motions 5545

if a component of extension existed across these Caribbean plate motion. Although a Caribbean
faults. For this explanation to be correct, the plate was not included in the RM1 model derived
rate of opening normal to the faults would have in paper 1, the topic of Caribbean plate motion
to be about 0.4 cm/yr. Although the field data was treated in detail by Jordan [1975]. He de-
do not appear to support this hypothesis rived a NOAM-CARBangular velocity vector using
[Detrick et al., 1973; Arcyana, 1975; Chouk- a spreading rate of 2.2 cm/yr across the Mid-
roune et al., 1977], the ability of these Cayman Rise estimated from topographic decay
studies (as well as ours) to resolve such a [Holcombe et al., 1973]. For the present study
component is an open question. we were fortunate to have available a much more
The incompatibility of the Famous trends with reliable rate (2.0 + 0.4 cm/yr since 2.3 m.y.
the R_M2model remains problematic. It is inter- B.P.) determined from a magnetic profile across
esting to note, however, that the RM2-predicted the Mid-Cayman Rise by Macdonald and Holcombe
azimuths are essentially perpendicular to the [1978]. This rate is essentially identical to
rise-crest segments in the Famous area. the previous estimate. Four slip vectors from
The Americas: One plate or two? A major the Molnar and Sykes [1969] set used by Jordan
conclusion of paper 1 was that significant [1975] were deleted, one from the West Indies
relative motion exists between North and South Arc, because it may lie south of the CARB-NOAM-
America. The present study supports this SOAM triple junction, and three from Hispaniola
conclusion, although direct observational evi- and the Puerto Rico Trench, where the data show
dence for NOAM-SOAMmotion is still lacking. An internal scatter and the stress and strain fields
inversion of the global data set was performed are complex [Jordan, 1975]. A slip vector for
with the Americas grouped into a single plate. the 1976 Guatemala earthquake [Kanamori and
This model was rejected because it does not sat- Stewart, 1978] was added. The changes to the
isfy the relative motion data in the Atlantic. direction data shifted the NOAM-CARBpole north-
In particular, the following are true: westward from the position computed by Jordan
1. The rates along the AFRC-NOAMboundary [1975]. It can be seen from Figure 5, however,
are misfit, model values being 0.4 cm/yr too low. that this shift is in the direction least con-
2. The azimuths along the AFRC-SOAMboundary strained by the data, as indicated by the orien-
yield systematically positive residuals of about tation of the RM2 confidence ellipse. Jordan's
5ø. pole lies within this confidence ellipse, and
3. The EURA-NOAMpole is shifted northward to the difference between these poles is not re-
81øN, 118OE, well outside the RM2 95% confidence solvable by the present data set (Figure 17).
ellipse. Consequently, the variation in rates The CARB-SOAMpole (Figure 4) is also shifted
along this boundary does not match the observa- with respect to Jordan's solution, but, again,
tions. the shift is along the major axis of the error
4. The AFRC-EURA pole is shifted westward to ellipse. This pole is unconstrained by data
12•S, 38•W. Such a pole implies compressive mo- along the CARB-SOAMboundary, so its 95% confi-
tion along the entire Azores-Gibraltar line. As dence ellipse is quite large. The change in
noted above, this consequence is in direct con- its location reflects the shifts in both NOAM-
flict with the extension observed on the western SOAM and NOAM-CARBpoles. Nevertheless,
portion of this boundary. Jordan's conclusion that a component of north-
We conclude that a nonzero NOAM-SOAMangular south motion exists along this boundary is un-
velocity is required by the revised data set. affected (Table 5).
To derive RM2, we adopted the convention of The Bouvet triple junction. RM1 did not pre-
paper 1 and partitioned the AFRC-NOAMand AFRC- dict correctly the relative motions of SOAM-ANTA
SOkM data sets at 15ON, where the distance and AFRC-ANTA [Forsyth, 1975; Sclater et al.,
between the Mid-Atlantic Ridge and the West 1976a]. In paper 1, these boundaries were very
Indies Arc is least. poorly constrained by data, but this deficiency
This grouping affords an excellent fit to the has been remedied by a number of recent special
data along the AFRC-SOAMboundary (Figure 16). studies (Table 1). P=M2provides an excellent
One datum on this boundary deserves particular fit to the data around Bouvet triple junction
mention. Eittreim and Ewing [1975] have mapped (Table 1, Figures 16 and 18), whereas RM1 per-
a recent, apparently continuous fault within the forms miserably. Three explanations for this
Vema fault zone. Their data yield a remarkably discrepancy were investigated:
well determined azimuth of relative motion; we 1. RM1 is located in a local minimum of the
assigned this datum an uncertainty of +2 ø, the fitting function manifold. This possibility can
lowest given to any direction datum. Its
be dismissed; inverting the RM1 data set with
residual computed from RM2 is only 0.4 •. In
P•M2 as a starting model yields the published
contrast, the residual computed from the model RM1 solution.
with a single Americas plate is nearly 5 ø.
Although some motion is required, the NOAM- 2. The SOAM-ANTA and AFRC-ANTA vectors are
SOAMangular velocity vector is not precisely very sensitive to small errors in the P•41 data
constrained. This is indicated by the large set. This possibility can also be excluded;
confidence ellipse associated with the pole the error ellipsoids for these vectors are ac-
(Figure 5). It is also evidenced by the fact tually quite small (paper 1, Table 2, and Figures
that the RM2 pole is nearly 30 • north of the 5 and 7). The prediction error computed from the
RM1 pole, completely reversing the sense of P•M1 variance matrix is much smaller than the
motion predicted along the boundary postulated RM1 misfit to the new data. If the new data
to lie somewhere between 10 • and 20oN. Discus- along the SOAM-ANTA and AFRC-ANTA boundaries are
sion of the inferred relative motion may be excluded from the revised data set, a solution
found in a later section. similar to RM1 is obtained.
5344 Minster and Jordan: Present-Day Plate Motions

3. The global data set is inconsistent with two rate estimates in the Red Sea [Allan and
the plate geometry assumed by KM1. Morelli, 1970], and these are also well fitted.
Hypothesis 3 is our preferred explanation and Because of the mediocre quality of the azimuth
was in fact advocated by Forsyth [1975] in his data and the variety of possible interpretations
original study of this problem. For reasons de- of Red Sea tectonics [e.g., Le Pichon et al.,
tailed below, we believe that the data set for 1973], we did not attempt to model the northern
plate motions about the Indian triple junction Red Sea in this work. Since the Arabian plate
are inconsistent with our model, and we ascribe is unconstrained along its other boundaries, the
this inconsistency to internal deformation with- RM2 and best fitting ARAB-AFRC vectors are
in the Indian plate. identical.
Plate motions in the Indian Ocean. This
brings us to the major difficulty that we en-
countered in constructing RM2: as pointed out The Indian Plate Problem
by Jordan et al. [1976] and Minster and Jordan
[1977], each of the three legs of the Indian Although RM2 is a very good fit to the data
triple junction are populated by internally con- set as a whole, we have not been able to fit the
sistent data, but the three best fitting vec- Indian Ocean data satisfactorily by an P•2 type
tors sum to a vector (the closure vector) sig- model. These discrepancies may simply result
nificantly different from zero (Table 3 and from bad data, contaminated by systematic
Figure 6). observational errors that we do not understand.
The AFRC-ANTA boundary is densely populated We are aware that data bias is the probable ex-
by good observations. The 6 rates, 6 transform planation for the misfit to the Aleutian slip
faults, and 11 slip vectors along this boundary vectors; in paper 1, we attributed this misfit,
constrain the angular velocity vector very well. evidently incorrectly, to internal deformation
The most important of these data is the well within the North American plate. The existence
mappedMelville transform fanit (• = 0.53) near of systematic errors in the Indian Ocean data ob-
the northeastern end of the boundary [Engel and viously cannot be ruled out at this time. How-
Fisher, 1975], which controls the latitude of ever, because its implications are important, an
the pole. RM2 performs close to optimally along alternate hypothesis--internal deformation with-
this boundary (Figure 18). in the major plates--deserves investigation.
As noted by McKenzie and Sclater [1971], the In RM2, Indian Ocean tectonics are modeled by
transform faults along the central Indian and three plates, ANTA, AFRC, and INDI. There is no
Carlsberg ridges tightly constrain the INDI- geological or seismic evidence for deformation
AFRC pole, and these constraints have been within Antarctica; in fact, the intraplate
strengthened by improved bathymetry [Engel and seismicity of Antarctica appears to be the
Fisher, 1975]. As shown on Figure 19, there is lowest of any major plate [e.g., Tarr, 1974].
a minor discrepancy between the rate data and In contrast, both the African and the Indian
the transform fault azimuths: the northernmost plate are characterized by high intraplate seis-
rates are too large by a few tenths of a centi- micity, and observations of significant post-
meter per year. In an effort to fit these rates Miocene intraplate deformation have been re-
the best fitting vector skews slightly with ported [e.g., McKenzie et al., 1970; Sykes,
respect to the transform fault data, and P•2 is 1970b; Eittreim and Ewing, 1972].
actually a better fit to the azimuths than the To investigate hypothetical intraplate defor-
BFP. However, the Carlsberg Ridge is opening mation, we have chopped these plates into two
slowly and lies close to the magnetic equator; pieces and modeled each as a rigid entity, as we
the magnetics along this boundary are not of ex- did for NOAM and SOB_M.. This procedure is ob-
ceptional quality [McKenzie and Sclater, 1971], viously unsatisfactory for representing widely
and we are not disturbed by this slight misfit. distributed strain, and we are implicitly as-
The problem of data inconsistency is evident suming that most of the deformation is localized
along the southeast Indian Ridge. The data are within a relatively narrow zone.
not quite as good along this boundary, but they Deformation of the African plate. Active ex-
determine a BFP and angular rate which constitute tension across the African rift valleys is well
an acceptable fit (Figure 20). RM1 fits these documented [e.g., McKenzie et al., 1970; Maasha
data very well, but RM2 fits poorly; the RM2 and Molnar, 1972; Le Pichon et al., 1973]. To
pole is significantly different from the BFP test the hypothesis that the RM2 misfit along
(Figure 6) and does not match the gradient in the INDI-ANTA boundary stems from ignoring this
the spreading rates. The situation is now deformation, another global inversion was per-
clear: RM1 satisfies the INDI-AFRC and INDI- formed. The data along the African plate
ANTA data but misses badly along the AFRC-ANTA boundaries in the Red Sea and west of 20øE were
boundary; R•.i2 corrects the misfit but then does assigned to a Nubian plate (NUBI), and the data
not satisfy the INDI-ANTA data. The most com- east of 40øE were assigned to a Somalian plate
prehensive local study of this triple junction (SOMA). We arbitrarily assumed that the posi-
was published by McKenzie and Sclater [1971]. tion of the NUBI-SOMA-ANTA triple junction is
Their instantaneous motion model is also shown somewhere between 20 • and 40•E. Since we did
on Figures 18-20. It is different from either not feel justified in specifying its position
RM1 or RM2 but does not constitute a better more accurately, the 10 data along the southwest
solution. Indian Ridge in this interval were deleted. As
The motion of Arabia. In the Gulf of Aden expected, the resulting model is a better fit
the rates obtained by Laughton et al. [1970, Ta- to the data set than RM2. In particular, the
ble 1] are used directly. These data show very INDI-ANTA angular velocity vector is very close
little scatter and are fitted by RM2 very well. to the best fitting solution in Table 3, and the
The only other data used in the inversions are fit to data along this boundary is much
 Minster and Jordan: Present-Day Plate Motions 5545

improved. However, the resulting SOMA-NUBI

pole TABLE 4. Hypothetical AUST-WIND Rotation
is at 43øS, 48øE, and the angular rate is 0.17•/ Vectors Described in Text
m.y., which implies east-west compressive motion
across the African rift valleys at a rate ex- e,•N •,•E •,deg/m.y.
ceeding 1 cm/yr'. This prediction clearly contra-
dicts the geophysical evidence. If a nonzero A -45.90 36.95 0.134
componentof extension is imposedon this bound- B -38.04 15.26 0.130
ary, the fit to the INDI-ANTA data set is de-
graded with respect to RM2.
Therefore problems with RM2 in the Indian
Ocean cannot be remedied by simply postulating between WIND and AUST in a northwest-southeast
internal deformation in Africa, because the re- direction.
sulting model violates other constraints. Al- Our modeling procedures do not require the
though the evidence for extension across the existence of a specific boundary separating the
African rift zone is compelling, we have not Indian plate into two portions. However, we
been able to resolve this motion successfully in
speculate that any deformation within the Indian
our global modeling studies, a conclusion also plate may in fact be localized in the vicinity
stated in paper 1. of the Ninetyeast Ridge. This linear feature
In a recent parallel study, Chase [1978] has behaved as an active transform fault in the
produced a global plate model which predicts Cretaceous [e.g., McKenzie and Sclater, 1971;
opening of the rift valleys. The differences Schlich, 1975; Sclater et al., 1976b], and al-
between his model and the model described above
though it has been commonly considered to be
are evidently due to differences in the inverted quiescent during recent times, Stein and Okal
data sets. We note that Chase's poles do not [1978] have suggested that it is now the site
provide a satisfactory fit to our data set along of significant seismic and tectonic activity.
the RM2 AFRC-ANTAboundary. Also, the misfit The nature of this tectonic activity is un-
to the INDI-ANTA data set described for RM2 is a
doubtedly complex, but Stein and Okal argue that
feature of his solution as well.
the bottom morphology and seismic source mech-
Deformation of the Indian plate. The hypoth- anisms are consistent with northwest-southeast
esis that the Indian plate is deforming is sug- compression in the region, in agreement with the
gested by two aspects of the RM2 fit discussed angular velocity vectors in Table 4. Vector A
in the previous pages: RM2's performance is un- predicts a rate of deformation of about 1 cm/yr,
satisfactory along both INDI-ANTA and INDI-PCFC computed at 15ON, 90øE. This rate is equivalent
boundaries. To test the hypothesis that INDI to a strain rate of 10-8/yr, if the deformation
deformation is responsible for these discrepan- were distributed over a zone 1000 km wide, and
cies, the western portion of the Indian plate is grossly compatible with the level of regional
(WIND) was separated from the eastern portion seismicity [Stein and Okal, 1978].
(AUST). Six INDI-ANTA data within a transition In summary, the hypothesis that deformation
zone between 90 • and 130•E were deleted. Data
is occurring within the Indian plate suffices
on the Indian plate boundaries west of 90•E were to resolve the difficulties encountered in fit-
assigned to WIND, and data east of 130ø were as- ting the instantaneous relative motion data.
signed to AUST. With this configuration the Although the nature of this deformation remains
global data set was inverted. The resulting speculative, at least a partial localization of
AUST-WIND angular velocity vector is labeled A the deformation in the vicinity of the Ninety-
in Table 4. Again, introduction of more model east Ridge is suggested by other observations.
parameters permits a better fit to the observa- We note that if extension across the African rift
tions: The remaining data along the southeast zone is incorporated into the plate tectonic
Indian Ridge are satisfied, and the AUST-PCFC model, deformation within the Indian plate pre-
pole lies within 2• of the INDI-PCFC BFP of dicted by the model will be greater.
Tab le 3.
From Table 3 we can estimate the hypothetical Predictions and Implications
AUST-WIND vector independently of the data along
the southeast Indian Ridge. Deformation of the Along plate boundaries where data are not
Indian plate can be approximately described by available or where interpretation is hindered by
the closure vector of the circuit WIND-AFRC-ANTA- geological complications, RM2provides a useful
PCFC-AUST. This vector may be calculated using basis for predictions and comparisons of global
the best fitting angular velocity vector for motions with local field evidence. We discuss
each boundary traversed by the circuit. The re- here a few selected examples. In this discus-
sult is not unique, since the PCFC-AUSTrate is sion, prediction errors were calculated using
not constrained, and a one-parameter family of the quadratic form described by Jordan [1975].
closure vectors is therefore generated. To Central California. Because of possible bias
specify a memberof this family, we arbitrarily associated with extension in the Basin and Range
chose to minimize the relative velocity of AUST Province, data along the San Andreas fault sys-
with respect to WIND at a point along the Nine- tem were not used in the inversion (Figure 1).
tyeast Ridge. Numerical experiments show that In central California, P•2 predicts a rate of
the result is quite insensitive to this point's relative motion between the Pacific and North
location. The derived angular velocity vector American plates of 5.6 ñ 0.3 cm/yr, in a direc-
is labeled B in Table 4. tion N35•W ñ 2 • (Table 5). On the basis of geo-
In view of the uncertainties involved (and logical evidence, Hall and Sieh [1977] estimate
the ad hoc criterion used to construct vector B), a slip rate of 3.7 ñ 0.3 cm/yr along the San
the two solutions in Table 4 are remarkably Andreas in central California, averaged over
similar. Both imply slow compressive motion three millenia, which agrees with Thatcher's
5346 Minster and Jordan: Present-Day Plate Motions

TABLE 5. RM2-Predicted Plate Motions at Selected Points

Rate
Plate Pair Position cm/yr Azimuth

PCFC-NOAM 36.4øN, 121.0øW 5.6 + 0.3 N35øW ñ 2.

SOAM-NOAM 20.0øN, 56.0øW 0.2 + 0.3 N71øW ñ 58.
15.0øN, 60.0øW 0.4 + 0.3 N62øW ñ 31.
SOAM-CARB 10.0øN, 66.0øW 2.3 -+ 0.5 N77øW ñ 10.
10.0øN, 74.0øW 2.2 + 0.5 N78øW ñ 10.
NAZC-CARB 7.5øN, 79.0øW 5.4 + 0.5 N71øE ñ 5.
ANTA-SOAM 50.0øS, 75.0øW 2.1 + 0.2 S88øE ñ 5.
ARAB-INDI 22.0øN, 62.0øE 1.4 + 0.2 N83øE ñ 9.
14.0øN, 59.0øE 0.8 + 0.2 N55øE ñ 14.

[1977] geodetical result and is similar to largely aseismic. However, some seismicity does
Huffman's [1972] estimated rate over the past exist. For example, a magnitude 6.2 earthquake
10 m.y. Between the Carrizo plain and Hollister, occurred October 23, 1964, at 19.8øN, 56.1øW.
the San Andreas fault exhibits a well defined The mechanism for this event is consistent with
azimuth of N41øW ñ 2 ø. This comparison suggests right lateral strike slip motion in a direction
that either RM2 does not model adequately the N55øW [Molnar and Sykes, 1969; J. Dorel, per-
relative motion of PCFC-NOAM, or that a signifi- sonal communication, 1975], which does not dis-
cant fraction of this motion is taken up else- agree with the RM2 prediction of N71øW ñ 58 ø
where. Based on our earlier discussion of this (Table 5). It is, however, inconsistent with
boundary and of Figure 7, we favor the second the RM1 model, which predicts left lateral
hypothesis. mot ion.
If one attributes the bulk of the discrepancy Southern boundary of the Caribbean plate. RM2
to deformation distributed within the Basin and predicts a component of north-south convergence
Range province, then one must postulate a global across the CARB-SOAMboundary. Although the
extension of this province of about 2 cm/yr, in rates are somewhat higher, the azimuths for
a direction N25øW. In contrast, Thompson and CARB-SOAM motion are almost identical to those
Burke [1973] estimate that the Basin and Range deduced by Jordan [1975] using the RM1 model.
underwent 100 km of extension in N55øW direction Consequently, Jordan's conclusions concerning
during the last 15 m.y., equivalent to an average motions along this boundary are substantiated
rate of 0.7 cm/yr. Their results are generally by this study. They are also supported by
compatible with Davis and Burchfiel' s [1973] sug- Ladd's [1976] model of Tertiary plate motions.
gestion that the Garlock fault is a major intra- Direct evidence for north-south compressive mo-
continental transform, with a horizontal dis- tion has been obtained by Talwani et al. [1976]
placement rate subsequently evaluated by Clark from an analysis of multichannel seismic reflec-
and Lajoie [1974] at 0.7 cm/yr during Holocenetime. tion records from the south margin of the Vene-
Therefore, geological evidence indicates that zuelan Basin and by Rial [1978] from a study of
Basin and Range extensional tectonics do not focal mechanisms in Columbia and Venezuela. No
constitute a sufficient explanation of the dis- such compression is predicted by a model which
crepancy between RM2 and observations in central assumes a single American plate. We take this
California. In particular, RM2 remains about to be an additional argument in favor of modeling
1.2 cm/yr faster than the estimated rate of slip NOAMand SOAMas two separate plates with a zone
on the San Andreas fault corrected for the con- of decoupling between 10 ø and 20øN.
tribution from the Basin and Range. An attrac- Jordan's [1975] portrayal of the tectonic re-
tive hypothesis is that some of the PCFC-NOAM lationships in the Panama Basin is also compat-
relative motion is accomodated on fault systems ible with RM2. The RM2 COCO-NAZCpole lies
west of the San Andreas. For example, Weber north of the equator, and the Panama transform
and Lajoie [1977] conclude that right lateral fault, as mapped by Lonsdale and Klitgord [1978],
slip has occurred along the San Gregorio fault closely approximates a small circle about this
zone during the last 200,000 years, with a rate pole, even though it was not used in the inver-
ranging from 0.6 to 1.3 cm/yr. This observation sion. Thus RM2 is consistent with the hypothe-
appears to reconcile observed and calculated sis that the Panama Basin east of this transform
rates, but there does remain a slight azimuth is not acting as a separate plate, as suggested
discrepancy, and such agreement might well be by Molnar and Sykes [1969] and Lonsdale and
fortuitous. Klitgord [1978], but in fact is part of the
Relative motion of North and South America. Nazca plate. Although RM2 predicts a slightly
As argued above, relative motion between North lower NAZC-CARBrate than RM1, the azimuths of
and South America is required by our data set. relative motion are nearly identical (Table 5)
Figure 5 and Table 2 indicate that the NOAM-SOAM and are consistent with the hypothesis that the
vector is poorly constrained, and a wide range motion is accommodated by a left lateral trans-
of possible relative velocities are allowed by form fault along the southern continental mar-
the data. Very little direct evidence for this gin of Panama [Jordan, 1975].
relative motion exists, and the movement could Subduction of southern Chile. Seismic activ-
be distributed across a broad zone between, say, ity along the Chile trench decreases sharply
10 ø and 20øN. Since the relative velocities are south of the NAZC-ANTA-SOAMtriple junction
predicted to be small, the deformation may be [Tarr, 1974]. Few earthquakes (only one with
ß Minster and Jordan' Present-Day Plate Motions 5347

mb >_ 6) have been reported in this region be-

tween 1963 and 1975. The RM2 predicted conver-
gence rate between ANTA and SOAM is only 2.1 +
0.2 cm/yr (Table 5), 6.7 cm/yr less than the
subduction velocity north of the triple junction
and 30% lower than the RM1 prediction. Yet
other convergence zones •with comparable rates
such as the West Indies Arc or the South Sand-
wich Trench are significantly more seismically
active. If our model is correct, then subduc-
tion in southern Chile takes place largely a-
seismically, or this boundary constitutes an
extensive seismic gap.
The Owen fracture zone. The Owen fracture
zone represents the INDI-ARAB boundary [e.g.,
McKenzieand Sclater, 1971] and exhibits only
weak seismicity. As shown in Table 5, RM2 does
predict a low rate of relative motion between
o o o o o o o o o
these two plates, but the predicted azimuths do
ZZZZZZZZ• not agree well with the observations. At 14øN,
Laughton' s [1970] bathymetric map indicates an
azimuth of N30øE for the Owen fracture zone,
compared with the model value of N55øE + 14ø;
and at 22øN a fault plane solution by Sykes
[1967] has a slip vector orientation of N50øE,
versus a model value of N83øE + 9 ø . Taken at
face value, these data suggest that the INDI-
ARAB pole should be translated to the northeast.
Interestingly, the inversion with INDI separated
into WIND and AUST, described above, yields an
WIND-ARAB pole positioned 3 ø north of the RM2
pole. The azimuth calculated at 14øN, 59øE is
N44øN, in better agreement with the observations,
although the azimuth calculated at 22øN, 62øE is
nearly identical to that for RM2.

ZZZZZZZZo• Absolute Motions

The P•M2geohedron (Table 2 and Figure 2) com-

pletely describes the relative motion model.
+t +t +1 +t +1 ß ß ß ß
To specify an 'absolute' reference frame, we
c• d,4 c•.4 need only to choose an origin in angular veloc-
ity space. A particular frame of interest in
discussions of plate dynamics is one fixed with
respect to the average position of the deep
mantle, assumed to be rigid or at least to have
typical internal motions much slower than the
motions of the plates; we refer to this frame
as the mean mesospheric frame.
In paper 1 we constructed an absolute motion
model (;241) based on the Wilson-Morgan fixed
hot spot hypothesis and concluded that this
hypothesis was consistent with the available
instantaneous motion data. However, we noted
the difficulties in estimating rates and direc-
tions of hot spot migration that are compatible
with the short time intervals appropriate to
the relative motion model, especially for hot
spot traces on the slower plates. Because of
these difficuities we are intrinsically limited
in our ability to construct more refined tests
of the Wilson-Morgan hypothesis and to discrim-
inate among various instantaneous absolute mo-
tion models using hot spot data.
To investigate this limitation, we have de-
rived an absolute motion model by again invert-
ing hot spot data but restricting the data set
to include only those constraints on hot spot
migration pertinent to the last 10 m.y. This
time span is really the minimum interval for
which good hot spot data can be obtained, al-
though it exceeds by over a factor of 3 the
mean averaging interval for the spreading rate
S548 Minster and Jordan' Present-Day Plate Motions

TABLE 7. Model AM1-2

Absolute Rotation Vector

Error Ellipse
%max • max • mln'
Plate
øN deg øE deg deg/m.y. deg/m.y. deg deg deg
AFRC 18.76 33.93 -21.76 42.20 0.139 0.055 S73øE 40.40 33.24
ANTA 21.85 91.81 75.55 63.20 0.054 0.091 N120E 93.01 56.12
ARAB 27.29 12.40 -3.94 18.22 0.388 0.067 S76øE 16.38 12.11
CARB -42.80 39.20 66.75 40.98 0.129 0.104 N300E 43.21 23.90
COCO 21.89 3.08 -115.71 2.81 1.422 .0.119 S32øE 3.35 2.25
EURA 0.70 124.35 -23.19 146.67 0.038 0.057 S67øE 151.10 118.90
INDI 19.23 6.96 35.64 6.57 0.716 0.076 S25øE 7.16 5.97
NAZC 47.99 9.36 -93.81 8.14 0.585 0.097 S02øE 9.37 5.43
NOAM -58.31 16.21 -40.67 39.62 0.247 0.080 S57øE 23.12 12.14
PCFC -61.66 5.11 97.19 7.71 0.967 0.085 S16øE 5.23 3.50
SOAM -82.28 19.27 75.67 85.88 0.285 0.084 NO30E 19.28 11.38

Symbols and conventions are the same as in Table 2

data. The azimuths of nine hot spot traces However, even supposing that the Wilson-Morgan
and the rates for five were chosen on the basis hypothesis is valid, which we have not proved,
of this criterion (Table 6). The data set is with what precision can the motions of the
dominated by the information from Pacific island plates in the mean mesospheric frame be predic-
chains; no Atlantic or Indian Ocean hot spots ted by the hot spot data? The answer to this
were employed. The rate at Hawaii represents question is indicated by the standard errors of
our interpretation of the K-Ar ages between estimation listed in Table 7. Although the ab-
Hawaii and French frigate shoals summarized by solute velocities of the fast moving oceanic
Da.lrymple et al. [1974]. For four other Pacific plates (e.g., PCFC) have relative errors which
archipelagos the K-Ar ages of Duncan et al. are small, the relative errors for the slowly
[1974] and Duncan and McDougall [1974, 1976] have moving continental plates (e.g., EURA) are quite
been used. Azimuth estimates for the traces were large and in some cases exceed 100%. Hence the
obtained from bathymetric charts, and the rate absolute motion directions of several plates,
estimates were projected along these directions. particularly ANTA and EURA, are not usefully
The mean rate estimates for individual island constrained by the hot spot data used in this
chains have formal standard errors of about +1 experiment. For example, at the position of Ice-
cm/yr [Duncan and McDouga!l, 1976], but these land the motion of EURAwith respect to the mean
have been increased to allow for possible errors mesospheric frame is predicted by AM1-2 to be
due to biased sampling. (We note that since N83øWat 0.4 cm/yr, nearly diametrically opposed
vulcanism may persist at a given site for mil- to the direction of the Wyville-Thompson Ridge,
lions of years, a systematic failure to sample the presumedhot spot trace. But no signifi-
the oldest rocks generally results in rates cance should be assigned to this discrepancy,
biased to high values.) The other data in since the formal prediction errors (1•) are
Table 6, hot spot azimuths from the COCO, NAZC, +162ø and +0.8 cm/yr, respectively, and since
and NOAMplates, have been taken from paper 1. the actual azimuth of the Iceland hot spot trace
The data set in Table 6 was inverted to obtain over the last 10 m.y. is not really known
an absolute motion model designated AM1-2 (Ta- (paper 1, p.566).
ble 7 and Figure 2). In the inversion the rela- With these large uncertainties in mind it is
tive plate velocity vectors were fixed at their interesting to compare the hot spot model with
RM2 values, but the uncertainties in the RM2 absolute motion models based on other criteria.
model, represented by its variance matrix, were Three such alternate models are listed in Table
incorporated into the calculation of the vari- 8 (see also Figure 2). AMO-2 is the unique ab-
ance matrix for AM1-2. The model is a very good solute motion model constructed by requiring
fit to the selected data set: only one datum that the lithosphere as a whole possess no net
has a residual exceeding its assigned error rotation, a •riterion discussed and applied in
(the azimuth of the _Marquesas), and the rate paper 1 and by Lliboutry [1974]
and Solomon and
data are all fit to Within 1 cm/yr. Thus the Sleep [1974]. AM2-2 corresponds to Burke and
results of this experiment give us no cause to Wilson's [1972] hypothesis that the African
challenge the Wilson-Morgan hypothesis. plate is stationary with respect to the mantle,
TABLE 8. Absolute Motion Models

Pacific Rotation Vector

Model Kinematical Condition Rate
Position deg/m.y.
AM0-2 No net rotation 62.93øS, 111.50øE 0.736
AM1-2 Best fit to hot spot data 61.66øS, 97.19øE 0.967
AM2-2 African plate fixed 59.15øS, 109.60øE 1.043
AM3-2 Caribbean plate fixed 63.52øS, 104.45øE 0.853
 Minster and Jordan' Present-Day Plate Motions 5349

a criterion endorsed by Duncan and McDougall lowed. Several authors have concluded that
[1976] on the basis of Pacific hot spot data. averaged over geologically long periods of time
AM3-2 conforms to Jordan's [1975] suggestion (> 40 m.y.), hot spots have relative velocities
that the Caribbean plate is fixed in the mean with magnitudes of the order of 1 cm/yr [Morgan,
mesospheric frame, pinned in position by its 1972; Burke et al., 1973; Molnar and Atwater,
two bounding subduction zones. 1973; Molnar and Francheteau, 1975]. In some
The predictions of these absolute motion mod- sense our conservative assignment of large er-
els are compared with the selected hot spot rors to the hot spot data in Table 6 may account
data in Table 7. The Pacific poles for all of for the uncertainties generated by small random
the absolute motion models are similar (Table motions among the hot spots, but appropriate
8), and the azimuths of the Pacific island caution in interpreting any hot spot model must
chains are essentially equally well fit by be exercised until better data and more rigorous
each. However, the Pacific rate data and the tests are available.
azimuth data from the other plates do provide Nevertheless, several previously published
some discriminants. AMO-2 appears to be incon- conclusions regarding present-day absolute mo-
sistent with the rate data; its values are sig- tions appear to be warranted; these are common
nificantly less than those observed. AM2-2 is to all of the models in Table 8: 1) Plate
a good fit to the Pacific data, but it is a poor speeds correlate negatively with total continen-
fit to the azimuth data for the other three
tal area (paper 1). 2) Plate speeds correlate
plates. AM3-2 provides a good fit to the azi- positively with the fraction of plate boundary
muth data, but its Pacific rate is slightly low. being subducted [Jordan and Minster, 1974; For-
The alternative absolute motion models can be
syth and Uyeda, 1975]. 3) Plate speeds cor-
compared directly with AM1-2 in model space relate positively with geographic colatitude
using the computed estimation errors. Let m be [Solomon et al., 1975]. Simple mechanical mod-
the model vector representing AM1-2 and m' be els have been formulated to explain the first
any alternative absolute motion model. Define two of these correlations [Forsyth and Uyeda,
the quadratic form 1975; Solomon et al., 1975; Kaula, 1975], but
F = (m- m')T-v-l-(m- m') their true dynamical significance is still quite
speculative. For example, Solomon et al. [1977]
where V is the complete variance matrix for m. have suggested that these aspects may have very
Then if F > (1.96) 2, m' lies outside the AM1-2
-- little to do with dynamics; they argue that the
95% confidence hypere!lipsoid, and one can ac- absolute plate motions characteristic of Ter-
cept the conclusion that the expected value of tiary time exhibit none of the correlations
m (of which m is only an estimate) is different stated above. Although we eye their reconstruc-
from m' at the 5% risk level. (Of course, this tions and modeling assumptions with some skep-
statement assumes that normal statistics and our ticism [cf., Jurdy, 1978], we agree that more
refined tests of the mechanical models must be
linear approximations are applicable and that V
formulated.
is known exactly, which is not strictly true;
it nevertheless provides a workable basis for
making statistical decisions.) For models AMO-2 Perspective
and AM2-2, F equals 12.4 and 10.9, respectively; RM2 is a significantly better representation
we conclude that these frames are significantly of present-day plate motions than RM1. In a
different from the hot spot frame. For AM3-2, recent parallel study, Chase [1978] has present-
F equals only 3.1, so the hypothesis that the ed a global plate motion model generally quite
Caribbean plate is fixed in the hot spot frame similar to RM2. Some significant differences
cannot be rejected. We note that the frames between these two models do exist, most being
corresponding to ANTA fixed (F = 0.5) and EURA ascribable to differences in data selection and
fixed (F = 0.8) are indistinguishable from the interpretation, but the overall agreement is
hot spot frame as well. encouraging. These studies should be viewed as
It is also interesting to compare AM1, the ever more rigorous tests of the plate tectonic
absolute motion model derived in paper 1, with hypothesis. We continue to be impressed by how
AM1-2. Both models were obtained by the inver- well the large data sets (330 members in Table
sion of hot spot data, but in the case of AM1, 1) are described by simple models with very few
no rate data were used, and a much larger, more parameters (30 for RM2).
globally distributed set of hot spot azimuths We have noted, however, several problem areas
were fit. Consequently, the averaging inter- where the plate model does not adequately fit
vals for the AM1 data are generally greater than the observations. These discrepancies deserve
10 m.y. and more variable. Although the AM1 and special scrutiny: they may be the manifesta-
AM1-2 Pacific poles are similar, the AM1 rota- tions of tectonic processes or other physical
tion rate (0.83ø/m.y.) is less than that of phenomena not now understood. For example, if
AM1-2 (0.97ø/m.y.). For AM1, F = 339. This our hypothesis that the Indian plate is not be-
very large value is indicative of the fact that having rigidly is confirmed by better data in
RM1 and RM2 are significantly different relative the Indian Ocean, then several questions must
motion models in that RM1 lies well outside be addressed. How is the deformation distributed
RM2's 95% confidence hyperellipsoid. A model within the plate? What is the nature of the
derived by adding to RM2 the AM1 PCFC absolute forces driving the deformation? Consider the
rotation vector yields F = 10.0 and is inconsis- hypothesis that the deformation is localized in
tent with the data set in Table 6. the vicinity of the Ninetyeast Ridge: then a
The resolution of absolute motions by the hot situation exists where on two opposing plates at
spot data is obviously degraded if the/possi- approximately equal distances from their common
bility of a nonrigid hot spot geometry is al- boundary (a spreading center), there are two
5350 Minster and Jordan' Present-Day Plate Motions

north-south trending zones of deformation, one

extensional (the African rift) and one compres-
sional (the Ninetyeast Ridge). This unusual 2ø
NOAM-PCFC 0.6 ø 0.3 ø
configuration should provide a strong discrimi- 2ø
NAZC-SOAM 0.9 ø 0ø
nant for force-balance models of the sort pro-
PCFC-INDI 1.1 ø -0.9 ø 4ø
posed by Forsyth and Uyeda [1975], Solomon et
al. [1975], and Richardson et al. [1976]. Of
course, more data are required before this hy- This correction is clearly minor. Thus as was
pothetical situation can be accepted as reality. pointed out by Chase [1978], omitting this cor-
Throughout the bulk of this paper the problems rection does not give rise to a significant
of continental tectonics have been carefully systematic bias in the data.
avoided. It is clear that in most regions of
intracontinental deformation, the plate model Acknowledgments. We are grateful to P. Molnar
has only limited utility. However, global plate for his contributions during the early stages of
this work and to C. G. Chase and K. C. Macdonald,
motions do provide the displacement boundary
and H. W. Menard for their comments on this
conditions required to understand the kinematics
and dynamics of tectonics in complex regions
typescript. We thank our many colleagues who
allowed us to use their data in advance of pub-
[e.g., Molnar and Tapponier, 1975]. These com-
lication. This research was sponsored by the
plex regions include not only the continental
interiors but also zones of deformation along National Aeronautics and Space Administration
the continental margins [e.g., Jordan, 1975] and under grant NSG 5089 and by the Division of
even boundaries between the oceanic plates them- Geological and Planetary Sciences, California
Institute of Technology.
selves. It is possibly complexities of this
latter type which are responsible for the dif-
ficulties that we experienced in obtaining clo-
sure about the Azores triple junction. References
Unlike the relative motions the absolute mo-
tions of plates in the mean mesospheric frame Abe, K., Mechanisms and tectonic implications of
cannot be precisely constrained. Absolute mo- the 1966 and 1970 Peru earthquakes, Phys.
tion models have been derived from a number of Earth Planet. Interiors, 5, 367-379, 1972.
kinematical hypotheses, and although they are Allan, T. D., and C. Morelli, The Red Sea, in
grossly similar, significant differences among The Sea, edited by A. E. Maxwell, vol. IV.
them do exist. In our opinion, model AM1-2 part 2, pp. 493-542, Interscience, New York.
with its attendant uncertainties (Table 7) rep- Anderson, R. N., D. W. Forsyth, P. Molnar, and
resents the most satisfactory description 3. Mammerickx, Fault-plane solutions of earth-
available from the present observations. On the quakes on the Nazca plate boundaries and the
basis of these absolute motions a number of Easter plate, Earth Planet. Sci. Lett., 24,
empirical correlations appear to be warranted, 188-202, 1974.
but how these correlations relate to the funda- Anderson, R. N., and 3. G. Sclater, Topography
mental forces driving the plates is only specu- and evolution of the East Pacific Rise between
lative. 5øS and 20øS, Earth Planet. Sci. Lett., 1__4,
433-441, 19 72.
Andrieux, 3., 3. M. Fontboth, and M. Mattauer,
Sur un module explicatif de l'arc de Gibral-
Appendix tar, Earth Planet. Sci. Lett., 12, 191-198,
1971.
In the interpretation of earthquake mechanisms Arcyana, Transform fault and rift valley geology
along subduction boundaries, most authors assume by bathyscaph and diving saucer, Science, 190,
that the direction of relative plate motion is 108-116, 1975.
given by the horizontal projection of the slip Atwater, T., and K. C. Macdonald, Are spreading
vector (e.g., paper 1). If the convergence is centers perpendicular to their transform
oblique to the trench axis, this procedure faults?, Nature, 270, 715-719, 1977.
yields a biased estimate of the direction of Avery, O. E., G. D. Burton, and 3. R. Heirtzler,
relative motion. Instead, the slip vector should An aeromagnetic survey of the Norwegian Sea,
be rotated into the horizontal plane, which re- J. Geophys. Res., 73, 4583-4600, 1968.
quires correcting the slip vector azimuth by an Banghar, A. R., and L. R. Sykes, Focal mecha-
amount • given by nisms in the Indian Ocean and adjacent regions,
J. Geophys. Res., 74, 632-649, 1969.
Bergh, H. W., and I. O. Norton, Prince Edward
fracture zone and the evolution of the
arccot

cot(T
A-PFTF)
sin ] +TF- TA Mozambiquebasin, J. Geophys. Res., 81, 5221-
5239,
Bird, P.,
19 76.
and 3. D. Phillips,
near the Oceanographer fracture
Oblique spreading
zone,
where TF, PF and TA, PA are the azimuth and J. Geophys. Res., 80, 4021-4027, 19 75.
plunge of the poles of the fault plane and aux- Burke, K., and 3. T. Wilson, Is the African
iliary plane, respectively. plate stationary?, Nature, 239, 387-390, 1972.
This correction was applied to the data from Burke, K., W. S. F. Kidd, and 3. T. Wilson,
the Aleutian-Kuril, South American, and Tonga- Relative and latitudinal motion of Atlantic
Kermadec trenches. The statistical information hotspots, Nature, 245, 133-137, 1973.
is summarized below: Chapman, M. E., and S.C. Solomon, North Ameri-
 Minster and Jordan' Present-Day Plate Motions 5351
,•,

can-Eurasian plate boundary in northeast Asia, Forsyth, D. W., Mechanisms of earthquakes and
J. Geophys. Res., 81, 921-930, 1976. plate motions in the East Pacific, Earth
Chase, C. G., Plate kinematics: The Americas, Planet. Sci. Lett., 17, 189-193, 1972.
East Africa, and the rest of the world, Earth Forsyth, D. W., Fault plane solutions and tecto -•
Planet. Sci. Lett., 3__7,353-368, 1978. nics of the South Atlantic and Scotia Sea,
Choukroune, P., J. Francheteau, and X. Le Pichon, J. Geophys. Res., 80, 1429-1443, 1975.
In situ structural observations along transform Forsyth, D. W., andS. Uyeda,Onthe relative
fault 'A' in the Famous area, Mid-Atlantic importance of driving forces of plate motion;
Ridge, Geol. Soc. Amer. Bull., 89, 1013-1029, Geophys. J. Roy. APtton. Soc., 43, 163-200,
1978.
1975.
Clark, M. M., and K. R. Lajoie, Holocene behavior Fox, P. J., W. C. Pitman, and F. Shepard,
of the Garlock fault, Geol. Soc. Amer. Abstr. Crustal plates in the central Atlantic: Evi-
Programs, 6, 156, 1974. dence for at least two poles of rotation,
Conant, D. A., Six new focal mechanism solutions Science, 165, 487-489, 1969.
for the Arctic and center of rotation for Fox, P. J., F. W. Schroeder, R.M. Moody, W. C.
plate movements, M.A. thesis, Columbia Univ. Pitman, and P. J. Hoose, The bathymetry of
New York, N.Y., 1972. the Oceanographer fracture zone and the Mid-
Cormier, V. F., Tectonics near the junction of Atlantic Ridge at 35øN with implications for
the Aleutian and Kuril-Kamchatka arcs and a central north Atlantic plate motion, submitted
mechanism for middle Tertiary magmatism in the to Deep Sea Res., 1978.
Kamchatka basin, Geol. Soc. Amer. Bull., 86, Hadley, D., and H. Kanamori, Seismotectonics of
443-453, 1975. the eastern Azores-Gibraltar ridge (abstract),
Dalrymple, G. B., M. A. Lanphere, and E. D. Eos Trans. AGU, 56, 1028, 1975.
Jackson, Contributions to the petrography and Hall, N. T., and K. E. Sieh, Late Holocene rate
geochronology of volcanic rocks from the of slip on the SanAndreasfault in the north-
leeward Hawaiian Islands, Geol. Soc. Amer. ern Carrizo Plain, San Luis Obispo County,
Bull., 8__5,727-738, 1974. California, (abstract), Geol. Soc. Amer. Abstr, i
Davis, G. A., and B.C. Burchfiel, Garlock Programs, 9, 428, 1977.
fault: An intracontinental transform structure, Hayes, D. E., and J.R. Cormoily, Morphologyof
southern California, Geol. Soc. Amer. Bull., the southeastIndian Ocean,in Antarctic
84, 1407-1422, 1973. Oceanology II: The Australian-New Zealand
Derrick, R., J. D. Mudie, B. P. Luyendyk, and Sector, Antattic Res. Set., vol. 19, edited
K. C. Macdonald, Near-bottom observations of by D. E. Hayes, pp. 125-146, AGU Washington,
an active transform fault, Nature, 246, 59-61, D. C., 1972.
1973. Heezen, B.C., and M. Tharp, Tectonic fabric of
Dickson, G. O., W. C. Pitman, and J.R. Heirtz- the Atlantic and Indian Oceans and continental
ler, Magnetic anomalies in the South Atlantic drift, Phil. Trans. Roy. Soc. London, Set. A,
Ocean and ocean floor spreading, J. Geophys. 258, 90-106, 1965.
Res., 73, 2087-2100, 1968. Herron, E. M., Sea-floor spreading and the Cenø-
Duncan, R. A., and I. McDougall, Migration of zoic history of the east-central Pacific,
volcanism with time in the Marquesas Islands, Geol. Soc. Amer. Bull., 83, 1671-1691, 1972.
French Polynesia, Earth Planet. Sci. Lett., Herron, E. M., and D. E. Hayes, A geophysical
2__1,414-420, 1974. study of the Chile ridge, Earth Planet. Sci.
Duncan, R. A., and I. McDougall, Linear volca- Lett., 6, 77-83, 1969.
nism in French Polynesia, J. Volcan. Geotherm. Hey, R. N., Tectonic evolution of the Cocos-
N•s., 1, 197-227, 1976. Nazca Rise, Ph.D. thesis, Princeton Univ.,
Duncan, R. A., I. McDougall, R.M. Carter, and Princeton, N.J., 1974.
D. S. Coorobs, Pitcairn Island- •a%other Hodgson, J. H., and W. G. Milne, Directio n of
Pacific hotspot?, Nature, 251, 679-682, 1974. faulting in certain earthquakes of the North
Eittreim, S. L., and J. Ewing, Midplate tecton- Pacific, Bull. Seismol. Soc. Amer., 41,
ics in the Indian Ocean, J. Geophys. Res., 7__7, 221-242, 1951.
6413-6421, 1972. Holcombe, T. L., P. R. Vogt, J. E. Matthews,
Eittreim, S. L., and J. Ewing, Vema fracture zone and R. R. Murchison, Evidence for sea-floor
transform fault, Geology, 3, 555-558, 1975. ' spreading in the Cayman Trough, Earth Planet.
Engdahl, E. R., N.H. Sleep, and M.-T. Lin, Plate Sci. Lett., 20, 357-371, 1973.
effects in north Pacific subduction zones, Huffman, O. F., Lateral displacement of upper
Tectonophysics, 3__7,95-116, 1977. Miocene rocks and the Neogene history of
Engel, C. G., and R. L. Fisher, Granitic to offset along the San Andreas Fault in central
ultramafic rock complexes of the Indian Ocean California, Geol. Soc. Amer. Bull., 83, --

ridge system, western Indian Ocean, Geol. Soc. 2913-2946, 19 72.

Amer. Bull., 86, 1553-1578, 1975. Isacks, •B., L. R. Sykes, and J. Oliver, Focal
Falconer, R. K. H., The Indian-Antarctic-Pacific mechanisms of deep and shallow earthquakes in
triple junction, Earth Planet. Sci. Lett., 17, the Tonga-Kermadec region and the tectonics
151-158, 1972. of islandarcs, Geol.Soc.Amer'Bull., 80,
Falconer, R. K. H., Indian-Pacific rotation pole 1143-1470, 1969.
determined from earthquake epicenters, Nature Johnson, G. L., J. R. Southall, P. W. Young, and'
Phys. Sci., 243, 97-99, 1973. P. R. Vogt, Origin and structure of the Ice-
Fisher, R. L., J. G. Sclater, and D. P. McKenzie, land plateau and Kolbeinsey ridge, J. Geophys.
The evolution of the Central Indian Ridge, Res ', 77, 5688-5696, 1972.
western Indian Ocean, Geol. Soc. Amer. Bull., Johnson, T., and P. Molnar, Focal mechanisms
82, 553-562, 1971. and plate tectonics of the southwest Pacific,
5352 Minster and Jordan' Present-Day Plate Motions

J. Geophys. Res., 7__7,5000-5032, 1972ß tectonic history of the eastern Panama basin,
Jordan, T. H., The preSent-day motions of the Geol. Soc. Amer. Bull., 89, 981-999, 1978.
Caribbean plate, J. Geophys. Res., 80, Maasha, N., and P. Molnar, Earthquake fault para-
4433-4439, 1975. meters and tectonics in Africa, J. Geophys.
Jordan, T. H., and J. B. Minster, Plate motions Res., 77, 5731-5743, 1972.
with respect to the mantle (abstract), Eos Macdonald, K. C., Near-bottom magnetic anomalies,
Trans. AGU, 55, 557, 1974. asymmetric spreading, oblique spreading, and
Jordan, T. H., J. B. Minster, and P. Molnar, tectonics of the Mid-Atlantic Ridge near lat.
Present-day plate motions (abstract), Eos 37øN Geol
, ß Soc. Amer Bull ß . , 88, 541-555, 1977.
Trans. AGU, 57, 329, 1976. Macdonald, K. C., and T. L. Holcombe, Inversion
Jurdy, D. M., An alternate model for early of magnetic anomalies and sea-floor spreading
Tertiary absolute plate motions, Geology, 6, in the Cayman Trough, submitted to Earth Planet.
469-472, 1978. Sci. Lett., 1978.
Kanamori, H., and G. S. Stewart, Seismological Macdonald, K., and B. P. Luyendyk, Deep-tow
aspects of the Guatemala earthquake of studies of the structure of the Mid-Atlantic
February 4, 1976, J. Geophys. Res., 8__3, Ridge crest near 37øN (Famous), Geol. Soc. Amer.
3427-3434. Bull., 88, 621-636, 1977.
Kaula, W. M., Absolute plate motions by boundary Mammerickx, 3., R. N. Anderson, H. W. Menard, and
velocity minimizations, J. Geophys. Res., 80, S. M. Smith, Morphology and tectonic evolution
244-248, 1975. of the East-Central Pacific, Geol. Soc. Amer.
Klitgord, K. D., and 3. D. Mudie, The Galapagos Bull., 86, 111-118, 1975.
spreading centre: A near-bottom geophysical McGregor, B. A., C. G. A. Harrison, 3. W.
survey, Geophys. J. Roy. Astron. Soc., 38, Lavelle, and P. A. Rona, 14=agnetic anomaly
563-586, 1974. patterns on Mid-Atlantic Ridge crest at 26øN,
Klitgord, K. D., J. D. Mudie, J. Grow and J. Geophys. Res., 82, 231-238, 1977.
P. A. Larson, Fast sea-floor spreading on the McKenzie, D. P., Active tectonics of the Mediter-
Chile ridge, Earth Planet. Sci. Lett., 20, ranean region, Geophys. J. Roy. Astron. Soc.,
93-99, 1973. 3__0,109-185, 1972.
Krause, C. C., and N. D. Watkins, North-Atlantic McKenzie, D. P., and R. L. Parker, Plate tecton-
crustal genesis in the vicinity of the Azores, ics in • space, Earth Planet. Sci. Lett., 22,
Geophys. J. Roy. Astron. Soc., 19, 261-284. 285-293, 1974.
1970. McKenzie, D. P., and 3. G. Sclater, The evolution
Ladd, J. W., Relative motion of South America of the Indian Ocean since the late Cretaceous,
with respect to North America and Caribbean Geophys. J. Roy. Astron. Soc., 25, 437-528,
tectonics, Geol. Soc. Amer. Bull., 87, 1971.
969-976, 1976. McKenzie, D. P., D. Davies, and P. Molnar, Plate
Larson, R. L., and C. G. Chase, Relative veloci- tectonics of the Red Sea and East Africa,
ties of the Pacific, North American and Cocos Nature, 226, 1-6, 1970.
plates in the middle America region, Earth Menard, H. W., and T. Atwater, Origin of frac-
Planet. Sci. Lett., 7, 425-428, 1970. ture-zone topography, Nature, 22, 1037-1040,
Laughton, A. S., A new bathymetric chart of the 1969.
Red Sea, Phil. Trans. Roy. Soc. London, Set. A, Minster, J. B., and T. H. 3ordan, Modelling
267, 21-22, 1970. present-day motions (abstract), Eos Trans. AGU,
Laughton, A. S., D. G. Roberts, and R. Graves,
Bathymetry of the northeast Atlantic: Mid- Minster, J. B., T. H. Jordan, P. Molnar, and
Atlantic Ridge to southwest Europe, Deep Sea E. Haines, Numerical modelling of instantaneous
Res., 22, 791-810, 1975. plate tectonics, Geophys. J. Roy. Astron. Soc.,
Laughton, A. S., R. B. Whirmarsh, and M. T. Jones, 3__6,541-576, 1974.
The evolution of the Gulf of Aden, Phil. Trans. Minster, J. F., J. B. Minster, M. Treuil, and
Roy. Soc. London, Set. A, 267, 227-266, 1970. C. J. All•gre, Systematic use of trace elements
Laughton, A. S., R. B. Whirmarsh, J. S. M. in igneous processes, II, Inverse problem of
Rushy, M. L. Somers, J. Revie, B. S. McCartney, the fractional crystallization process, Contrib.•
and J. E. Nafe, A continuous east-west fault Mineral. Petrol., 6__1,49-77, 1977.
of the Azores-Gibraltar Ridge, Nature, 327, Molnar, P., Fault plane solutions of earthquakes
217-220, 1972. and direction of motion in the Gulf of
Lawver, L. A., J. W. Hawkins, and 3. G. Sclater, California and on the Rivera fracture zone,
Magnetic anomalies and crustal dilation in the Geol. Soc. Amer. Bull., 84, 1651-1658, 1973.
Lau Basin, Earth Planet. Sci. Lett., 33, 27-35, Molnar, P., and T. Atwater, Relative motion of
1976. hot spots in the mantle, Nature, 246, 288-291,
Le Pichon, X., J. Francheteau, and 3. Bonnin, 1973.
Plate Tectonics, 300 pp., Elsevier, New York, Molnar, P., and J. Francheteau, The relative
1973. motion of 'hot spots' in the Atlantic and
Lliboutry, L., Plate movement relative to rigid Indian oceans during the Cenozoic, Geophys. J.
lower mantle, Nature, 250, 298-300, 1974. Roy. Astron. Soc., 43, 763-774, 1975.
Lonsdale, P., Regional shape and tectonics of Molnar, P. and L. R. Sykes, Tectonics of the
the equatorial East Pacific Rise, Mar. Geophys. Caribbean and middle America r•gions from focal
Res., __
3, 251-315, 1978a. mechanisms and seismicity, Geol. Soc. Amer.
Lonsdale, P., Near-bottom reconnaissance of a Bull., 8__0,1639-1684, 1969.
fast-slipping transform fault at the Pacific- Molnar, P., and P. Tapponnier, Cenozoic tectonics
Nazca plate boundary, J. Geol., in press, 1978b. of Asia: Effects of a continental collision,
Lonsdale, P., and K. D. Klitgord, Structure and Science, 189, 419-426, 1975.
 Minster and Jordan' Present-Day Plate Motions 5555

Molnar, P., T. Atwater, J. Mammerickx, and S. M. Solomon, S. C., Shear-wave attenuation and melt-
Smith, Magnetic anomalies, bathymetry, and the ing beneath the Mid-Atlantic Ridge, J. Geoph•s.
tectonic evolution of the South Pacific since Res., 78, 6044-6059, 1973.
the late Cretaceous, Geophys. J. Roy. Astrom. Solomon, S.C., and N.H. Sleep, Some simple
$oc., 40, 383-420, 1975. physical models for absolute plate motions,
Morgan, W. J., Plate motions and deep mantle con- J. Geophys. Res., 7__9,255 7-256 7, 19 74.
vection, Geol. $oc. Amer. Men. 132, 7-22, Solomon, S.C., N.H. Sleep, and R. M. Richard-
1972. son, On the forces driving plate tectonics:
Morgan, W. J., P. R. Vogt, and D. E. Falls, Mag- Inferences from absolute plate velocities and
netic anomalies and sea-floor spreading on the intraplate stress, Geophys. J. Roy. Astron.
Chile Rise, Nature, 222, 137-142, 1968. Soc., 4__2,769-802, 1975.
Norton, I. 0., The present relative motion be- Solomon, S. C., N.H. Sleep, and D. M. Jurdy,
tween Africa and Antarctica, Earth Planet. Sci. Mechanical models for absolute plate motions
Lett., 33, 219-230, 1976. in the early Tertiary, J. Geophys. Res., 82,
Olivet, J. L., X. Le Pichon, S. Monti, and B. 203-212, 1977.
Sichler, Charlie-Gibbs fracture zone, J. Geo- Stauder, W., The Alaska earthquake of July 20,
phys. Res., 79, 2059-2072, 1974. 1958: Seismic studies, Bull. Seismol. Soc.
Phillips, J. D., Magnetic anomalies over the Mid- Amer., 50, 293-322,
__
1960.
Atlantic Ridge near 27øN, Science, 157, Stauder, W., Mechanism of the Rmt Island earth-
920-923, 1967. quake sequence of February 4, 1965, with re-
Pitman, W. C., and M. Talwani, Sea-floor spread- lation to island arcs and sea floor spreading,
ing in the North Atlantic, Geol. Soc. Amer. J. Geophys. Res., 73, 3847-3858, 1968a.
Bull., 83, 619-646, 1972. Stauder, W., Tensional character of earthquake
Pitman, W. C., E. M. Herron, and J. R. Heirtzler, foci beneath the Aleutians, J. Geophys. Res.,
Magnetic anomalies in the Pacific and sea-floor 7__3,7693-7701, 1968b.
spreading, J. Geophys. Res., 73, 2069-2085, Stauder, W., Mechanism and spatial distribution
1968. of Chilean earthquakes with relation to sub-
Rea, D. K., Changes in the axial configuration of duction of the oceanic plate, J. Geophys. Res.,
the East Pacific Rise near 6øS during the last 78, 5033-5061, 1973.
2 m.y., J. Geophys. Res., 81, 1495-1504, 1976a. Stauder, W., Subduction of the Nazca plate under
Rea, D. K., Aa%alysis of a fast-spreading rise Peru as evidenced by focal mechanisms and by
crest: The East Pacific Rise, 9 ø to 12 ø South, seismicity, J. Geophys. Res., 80, 1053-1064,
Mar. Geophys. Res., 2, 291-313, 1976b. 1975.
Rea, D. K., and R. J. Blakely, Short wavelength Stauder, W., and G. A. Bollinger, The focal
magnetic anomalies in a region of rapid sea- mechanism of the Alaska earthquake and its
floor spreading, Nature, 255, 126-128, 19 75. aftershocks, J. Geophys. Res., 71, 5283-5296,
Rial, J. A., The Caracas, Venezuela, earthquake 1966.
of July 1967: A multiple-source event, Stauder, W., and L. Mualchin, Fault motion in
J. Geophys. Res., 83, in press, 1978. the larger earthquakes of the Kuril-Kamchatka
Richardson, R. M., S.C. Solomon, and N.H. arc and of the Kuril-Hokkaido corner, J. Geo-
Sleep, Intraplate stresses as an indicator of phys. Res., 81, 297-308, 1976.
plate tectonic driving forces, J. Geophys. Stein, S., and E. A. Okal, Seismicity and tec-
Res., 81, 1847-1856, 1976. tonics of the Ninetyeast Ridge area: Evidence
Rosendahl, B. R., Ipod survey area PT-4, pre- for internal deformation of the Indian plate,
liminary site survey report, Scripps Inst. of J. Geophys. Res., 83, 2233-2245, 1978.
Oceanogr., La Jolla, Calif., 1976. Stein, S., H. J. Melosh, and J. B. Minster,
Rusnak, G. A., R. L. Fisher, and F. P. Shepard, Ridge migration and assnmnetric sea-floor
Bathymetry and faults of Gulf of California, spreading, Earth Planet. Sci. Lett., 36,
Marine Geology of California, Men. 3, edited 51-62, 1977.
by Tj. H. Van Andel, and G. G. Shor, Jr., Sykes, L. R., Mechanism of earthquakes and
pp. 59-75, Amer. Assoc. Petrol. Geol., Tulsa, nature of faulting on the midocean ridges,
Oklahoma, %964. J. Geophys. Res., 72, 2131-2153, 1967.
Schlich, R., Structure et age de l'oc•an indien Sykes, L. R., Focal mechanism solutions for
occidental, M•m. hors S•r. 6, Soc. G•ol. de earthquakes along the world rift system,
Fr. , 1975. Bull. Seismol. Soc. Amer., 60, 1749-1752,
Schlich, R., and P. Patriat, Mise en •vidence 1970a.
d' anomalies magn•tiques axiales Bur la branche Sykes, L. R., Seismicity of the Indian Ocean
ouest de la dorsale m•dio-indienne, C. R. Acad. and a possible nascent island arc between
Sci., 272, 700-703, 1971. Ceylon and Australia, J. Geophys. Res., 75,
Sclater, J. G., R. N. Anderson, and M. L. Bell, 5041-5055, 1970b.
Elevation of ridges and evolution of the Talwani, M., C. C. Windisch, and M. G. Langseth,
central east Pacific, J. Geophys. Res., 76, Reykjanes ridge crest: A detailed geophysical
7888-7915, 1971. study, J. Geophys. Res., 76, 473-517, 1971.
Sclater, J. G., C. Bowin, R. Hey, H. Hoskins, Talwani, M., C. C. Windisch, P. L. Stoffa,
J. Peirce, J. Phillips, and C. Tapscott, P. Buhl, and R. E. Houtz, Multi-channel seis-
The Bouvet triple junction, J. Geophys. Res., mic study in the Venezue%an
basin and the
81, 1857-1869, 1976a. Curagao Ridge (abstract), Eos Trans. AGU, 57,
Sclater, 3. G., B. P. Luyendyk, and L. Meinke, 266, 1976.
Magnetic linearions in the southern part of Tart, A. C., World seismicity map, U.S. Geol.
the Central Indian Basin, Geol. Soc. Amer. Surv., Reston, Va., 1974.
Bull., 87, 371-378, 1976b. Thatcher, W., Secular deformation, episodic move-
5354 Minster and Jordan' Present-Day Plate Motions

ments, and relative plate motions in Southern seafloor spreading north of Iceland, J. Geo-
California (abstract), Eos Trans. AGU, 5__8, phys. Res., 75, 903-920, 1970.
496, 1977. Weber, G. E., and K. R. Lajoie, Late Pleistocene
Thompson, G. A., and D. B. Burke, Rate and direc- and Holocene tectonics of the San Gregorio
tion of spreading in Dixie Valley, Basin and fault zone between Moss Beach and Point Ano
Range Province, Nevada, Geol. Soc. Amer. Bull., Nuevo, San Mateo County, California, Geol.
Soc. Amer.Abstr. Programs,9' 524, 1977.
Tobin, D. G., and L. R. Sykes, Seismicity and Weissel, 3. K., and D. E. Hayes, Magnetic anom-
tectonics of the northeast Pacific Ocean, alies in the south-east Indian Ocean, in
J. Geophys. Res., 7__3,3821-3845, 1968. Antarctic Oceanology II: The Australian-New
Uchupi, E., Eastern Yucatan continental margin Zealand Sector, Antarctic Res. Ser., vol. 19,
and western Caribbean tectonics, Amer. Ass. edited by D. E. Hayes, pp. 165-196, AGU,
Petrol. Geol. Bull., 57, 1075-1085, 1973. Washington, D.C., 1972.
Udfas, A., A. Ldpez Arroyo, and 3. I•ezcua, Seis- Weissel, 3. K., and D. E. Hayes, The Australian-
motectonics of the Azores-Alboran region, Antarctic discordance: New results and impli-
Tectonophysics, 31, 259-289, 1976. cations, J. Geophys. Res., 7.9, 2579-2587,
Van Andel, Tj. H., D. K. Rea, R. P. Von Herzen, 1974.
and H. Hoskins, Ascension fracture zone,
Ascension island, and the Mid-Atlantic Ridge,
Geol. Soc. Amer. Bull., 84, 1527-1546, 1973. (Received November 28, 1977;
Vogt, P. R., N. A. Ostenso, and G. L. 3ohnson, revised April 24, 19 78;
Magnetic and bathymetric data bearing on accepted June 20, 1978.)