Professional Documents
Culture Documents
World Archaeology
Publication details, including instructions for authors and subscription
information:
http://www.tandfonline.com/loi/rwar20
To cite this article: Geoffrey W. Conrad (1977) Models of compromise in settlement pattern studies: An
example from coastal Peru, World Archaeology, 9:3, 281-298, DOI: 10.1080/00438243.1973.9979704
Taylor & Francis makes every effort to ensure the accuracy of all the information (the
“Content”) contained in the publications on our platform. However, Taylor & Francis, our
agents, and our licensors make no representations or warranties whatsoever as to the
accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views
expressed in this publication are the opinions and views of the authors, and are not the views
of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon
and should be independently verified with primary sources of information. Taylor and Francis
shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses,
damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in
connection with, in relation to or arising out of the use of the Content.
This article may be used for research, teaching, and private study purposes. Any substantial
or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or
distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and
use can be found at http://www.tandfonline.com/page/terms-and-conditions
Models of compromise in settlement pattern
studies: an example from coastal Peru
Geoffrey W. Conrad
Introduction
Downloaded by [Universite Laval] at 04:52 29 December 2014
Methodological background
The method to be followed in developing a model of compromise is derived from
Trigger's (1968: 72) 'principle of hierarchical resolution of conflicting tendencies'^ This
principle states that in cases where the selection of a settlement pattern requires a
compromise among opposing considerations, the resulting configuration will reflect the
relative importance of the factors involved. ;
World Archaeology Volume 9 No. 3 Landscape archaeology
282 Geoffrey W. Conrad
Archaeological background
The archaeological region known as the North Coast of Peru is a 350-km. strip of
desert interrupted by a series of alluvial river valleys. The standard North Coast
chronology assigns the period c. A.D. 200-700 to a cultural phase known as Moche
(Donnan 1973: 1; Day 1976: 36-7). Five Moche subphases have been defined on the
basis of detailed changes in the characteristic ceramics of the period (Larco Hoyle 1948).
Archaeological investigations have identified the cultural phenomenon represented by
the Moche phase as a military expansionist state whose history can be outlined as follows.
During subphases I and II the state seems to have been confined to the Moche and
Chicama Valleys (fig. 1). It subsequently expanded to encompass the nine valleys from
ZANA
JEQUETEPEQUE
CHICAMA
PACIFIC OCEAN
50 100 NE PEÑA' 77 e
Figure i Map of the North Coast of Peru showing the valleys dominated by the Moche state
at the height of its power
Models of compromise in settlement pattern studies 283
Lambayeque in the north to Nepeña in the south, a distance of about 300 km. The bulk
of this expansion took place in subphases III and IV. Presumably most of the sites, and
certainly all of the large centres, considered in the following pages were established
during subphases III and IV, c. A. D. 300-600 (Donnan 1973:125,131 ; Proulx 1973:48).
Throughout subphases I-IV the capital of the state lay at the type site of Moche
(Huacas del Sol and de la Luna) in the Moche Valley (Donnan 1973: 125-6; Day 1976:
37). In subphase V, a relatively brief period, a new capital was established at the site of
Pampa Grande in Lambayeque (Day 1976: 40). Shortly thereafter the state collapsed.
The reasons for the shift of capital and ultimate collapse remain uncertain.
Published studies of Moche settlement patterns are available for three valleys: Viru
(Willey 1953), Santa (Donnan 1973), and Nepeña (Proulx 1968, 1973). The Viru data
were gathered by the Viru Valley Project of 1946, which produced a number of impor-
tant reports (Bennett 1950; Strong and Evans 1952; Collier 1955), including the first
Downloaded by [Universite Laval] at 04:52 29 December 2014
systematic archaeological settlement pattern study (Willey 1953). (Members of the Viru
Valley Project used the name 'Huancaco' to identify the local manifestation of the Moche
phase; this term will not be employed here, except as the designation for one large site.)
Information from Santa and Nepeña will be used for comparative purposes, but I
have chosen to base my arguments on Willey's Viru data because, while they were the
least recently gathered, they are the most extensive and probably the most nearly
complete. A total of 106 Moche sites were recorded in Viru, compared to 85 in Santa
and 22 in Nepeña (Willey 1953: 178; Donnan 1973: 2; Proulx 1973: 40). Furthermore,
60 of the Santa sites and 13 of the Nepeña sites functioned solely as cemeteries (Donnan
1973: 11; Proulx 1973: 40); although the total is probably slightly higher (see below),
Willey (1953: 234) classified only 23 of the Viru sites as cemeteries. Hence Viru, which
contains the greatest number of habitation sites and public centres, provides the best
basis for discussion.
deposits separated by distances of a few hundred metres or less. The single exception
is the Huancaco group ^—88/89/90/91/92/93), a large site that was arbitrarily sub-
divided for the purposes of specimen collection and cataloguing (Willey 1953: 205-6).
Second, several sites with corporate labour architecture are treated here as villages
(V-59, 95, 152/153, 167). In all of these cases except V-95 the only Moche remains are
village refuse; the public architecture definitely belongs to earlier periods (Willey 1953:
205, 213, 218). V-95 contains an undated pyramidal mound of earth or adobe bricks,
along with habitation refuse of several periods, including the Moche phase (Willey 1953 :
90, 218). Since the only definite traces of Moche occupation at V-95 are refuse deposits,
I have classified the site as a village.
Finally, I am not considering sites that served only as cemeteries and had no living
population. Therefore, I have eliminated four sites (V-155, 258, 272, 308) in which,
despite the presence of architecture or domestic debris from earlier periods, all Moche
Downloaded by [Universite Laval] at 04:52 29 December 2014
materials were found in intrusive graves (Bennett 1950: 47; Willey 1953: 128, 214,
222-3; Collier 1955: 87-8).
Classification of centres
Centres are divided into the subcategories primary, secondary, and tertiary. (These
terms are used in a provincial sense only; obviously, even the primary centre in Viru was
subordinate to the capital of the state.) The basis for classification is corporate labour
investment as inferred from the size and extent of public architecture. This emphasis on
public architecture is grounded in an analogy to the Inca Empire (A.D. 1438-1532),
which dominated the entire Central Andean area at the time of the Spanish Conquest.
Ethnohistoric sources indicate that the Inca economy was based on a system of labour
taxation (mit'a) administered by a hierarchical state bureaucracy. The mit'a required
taxpayers to contribute a certain amount of labour to the state each year; all Inca public
projects were constructed through this system (Rowe 1946: 267-8; Murra 1958).
Moseley (1975) has been able to demonstrate that mii'a-like labour taxation was an
ancient Central Andean pattern, and specifically a characteristic of the Moche state.
Accordingly, my classification of centres is based on the assumption that in the Moche
state, as in the Inca Empire, a more important centre would control more taxpayers, and
would therefore receive a greater investment of corporate labour, than a less important
centre.
During the Moche phase there was one predominant focus of labour investment, and
hence one primary centre, in Viru. This site is the large Huancaco group (V-88/89/90/
91/92/93) on the south side of the valley; in fact, Huancaco may have been the largest
site ever established in Viru (Willey 1953: 205, 382).
There are two sites whose ruins reflect a labour investment less than that of Huancaco,
but clearly much greater than that of any other Moche site. These two sites, classified
here as secondary centres, are V-67 (Kroeber 1930: 77; Willey 1953: 226) and V-280
(Willey 1953: 214-5, 382).
Finally, six remaining sites with lesser public architecture are treated as tertiary
centres (V-51, 130, 166,168, 245, 276/288). There may be some finer distinctions among
them, but I cannot make any on the basis of the available data.
286 Geoffrey W. Conrad
even identify, all such factors; instead, I will isolate three fairly obvious considerations
and treat each in turn as the sole determinant of an optimal distribution of the thirty-five
sites in the lower and middle valley. These factors are:
1 maximization of agricultural land (resource maximization) ;
2 minimization of the work required to cultivate that land (minimization of effort) ; and
3 maintenance of social and political control (administrative efficiency).
Resource maximization
In order to develop predicted settlement patterns for resource maximization and effort
minimization, I must make one further assumption: that the maximum limits of
prehistoric cultivation in Viru (figs. 2, 3, 5, 6) had been attained by the Moche phase.
The Viru Valley Project did collect some information on prehistoric irrigation and field
systems, but the data are not very detailed. In general, the available information suggests
that the maximum limits of prehistoric cultivation were approximated in Moche times,
but that some sections of land within these boundaries, particularly in the lower valley
near the coast, may not have been in use (Willey 1953: 364-5). Hence the assumption
stated above is probably to some degree incorrect, but the data at hand are insufficient to
permit a more precise delimitation of Moche farmlands in Viru.
Given these problems, the question remains: How should the thirty-five sites in
lower and middle Viru be distributed in order to maximize agricultural land? The
answer is obvious: if agricultural land is to be maximized, sites should not occupy
arable locations, and settlement should be restricted to zones outside the limits of culti-
vation. Since there are many ways in which sites can be distributed in the zones beyond
those limits, there is a large number of alternative settlement patterns that would serve the
purpose of maximizing agricultural land. Figure 3A illustrates one alternative, in which
settlements are evenly distributed along the margins of the cultivated part of the valley.
There are three points to note about the pattern shown in fig. 3A. First, this particular
distribution represents a compromise in which minimization of travel time to farmland
is subordinated to the desire to maximize that land, but not completely ignored. That is,
of all the patterns that remove sites from cultivable land, this one keeps them as close as
Models of compromise in settlement pattern studies 287
PACIFIC OCEAN
Downloaded by [Universite Laval] at 04:52 29 December 2014
I I I | K m
O 12 3
Figure 2 Moche phase settlement pattern in the lower and middle Viru Valley (modified from
Willey 1953). Centres, combined sites, and reclassified sites are numbered in order to facilitate
comparison with the original report
possible to that land. Second, for the purpose of maximizing agricultural land the specific
locations of public centres are irrelevant, provided that they, like villages, lie outside the
limits of cultivation. Therefore, centres may be randomly distributed among the thirty-
five site locations shown infig.3A. Third, and most important, comparison of figs 2 and
3A shows that this predicted settlement pattern is very different from the actual one.
Unless one makes the undoubtedly erroneous assumption that only a small fraction of the
Viru Valley was farmed during the Moche phase, it seems clear that many sites lay
within the limits of cultivation on potentially arable land.
Minimization of effort
A settlement pattern can optimize agricultural effort by minimizing the amount of time
a farmer must spend travelling from home to fields. In other words, the problem here is
one of dividing the cultivated part of the valley among thirty-five sites in such a way that
the inhabitants of each site live as close as possible to the land they must farm.
Classical Iocational theory is based on the fact that the hexagon is the most economical
geometrical form for the equal division of an area among a number of points (Christaller
1933; Haggett 1966). The advantages of the hexagon are two-fold. First, it is one of
several regular polygons that can be packed in such a way as to cover all of the area in
288 Geoffrey W. Conrad
question; a hexagonal grid contains no interstices. Second, of the packable polygons, the
hexagon is the one that most nearly approximates a circle - that is, the one in which all
points on the perimeter are as close as possible to being equidistant from the centre.
Accordingly, during the Moche occupation of the Viru Valley the optimal settlement
pattern for minimizing travel time from home to farmland would have been one in which
the thirty-five sites were arranged so that each lay at the centre of an equal hexagonal
field. This pattern is shown in fig. 3B.
Downloaded by [Universite Laval] at 04:52 29 December 2014
Figure 3 Optimal settlement patterns for maximization of arable land (A) and minimization of
agricultural effort (B) in lower and middle Viru
Models of compromise in settlement pattern studies 289
Several aspects of this predicted distribution merit further consideration. Again, if the
sole determinant of settlement pattern is minimization of agricultural effort, the specific
locations of centres are irrelevant. If one assumes that at least a portion of the population
of all settlements was engaged in farming, centres may be randomly distributed among
the thirty-five locations shown in fig. 3B. More importantly, while the predicted settle-
ment pattern for maximization of farmland was very different from the actual configura-
tion, the expected pattern for minimization of agricultural effort is much closer to reality
(figs 2, 3A, 3B).
Administrative efficiency
The optimal settlement pattern for maintaining social and political control is a hexagonal
central place hierarchy (Christaller 1933). In accordance with Christaller's administrative
Downloaded by [Universite Laval] at 04:52 29 December 2014
principle, each central place in this ideal hierarchy has a K-value of 7. That is, each
central place has uncontested control of its dependent places. No dependent place is
shared by two or more central places, since sharing of dependent places among central
places would create confusion, undercutting administrative efficiency. (For more
complete discussions of K-functions see Haggett 1966: 118-25; Smith 1975; Blanton
1976.)
In the specific case of the Viru Valley, which has a three-tiered hierarchy of centres,
the ideal site distribution would be similar to the one Marcus (1973) has identified
among the lowland Classic Maya: secondary centres should be distributed hexagonally
around the primary centre, tertiary centres should define smaller hexagons around the
primary and secondary centres, and villages should cluster around the tertiary centres.
Since the lower and middle reaches of the Viru Valley contain only two secondary and
six tertiary centres, it is obvious that the hexagonal cells would be incomplete. None the
less, it is possible to derive a predicted hexagonal central place hierarchy in a form
truncated by the limits of the valley.
The first step is to determine the predicted locations of six secondary centres defining
a hexagonal cell around the primary centre at Huancaco. In order to do so I must
compute the radius of the hexagon, which will also be the length of each of its sides.
Note that the two existing secondary centres are nearly equidistant from Huancaco
(fig. 2): V-67 n e s 8 km. from Huancaco, while V-280 is 7-2 km. from Huancaco; the
average of these distances is 7-6 km.
I have not included the distance separating V-67 from V-280, 11-7 km., in the
computation. The ratio of this distance to the average distance separating V-67 an< i
V-280 from Huancaco is approximately 1*5:1. Furthermore, lines drawn connecting
V-67 a n d V-280 to Huancaco form an angle of about ioo°. In a hexagon the ratio of
the distance between two adjacent apices, whose radii form an angle of 6o° at the centre,
to the distance between apices and centre,- in other words, the ratio of side to radius - is
1:1 (fig. 4A). However, the ratio of the distance between two apices whose radii
intersect at an angle of 120o to the distance between apices and centre is V3 •' 1, or
approximately 1*7 : 1 (fig. 4B). Since the geometrical configuration of V-67, V-280, and
Huancaco more closely approaches the latter case, the distance between the two
secondary centres should not be treated as a measure of the radius of the hexagon.
290 Geoffrey W. Conrad
V3" r
Downloaded by [Universite Laval] at 04:52 29 December 2014
Figure 4 Comparison of the ratios of the distance between two apices to the length of the radius
of a hexagon. In Case A, radii drawn to the apices in question define an angle of 60° at the
centre. The distance separating the two apices (the length of a side of the hexagon) is equal to
the radius; the ratio is 1:1. In Case B, where radii drawn to the apices intersect to form an
angle of 120o, the ratio of the distance between the apices to the length of the radius is -v/3 :i.
The latter value can be verified by the Pythagorean theorem
as some of them would be shared. However, if they lie at the midpoints of the sides,
none are shared, and Christaller's administrative principle is satisfied. Furthermore,
tertiary centres located in this manner are still arranged hexagonally around the primary
and secondary centres (fig. 5.4). Accordingly, the expected sites of the tertiary centres
Î /,
/ í c
\ 1 • '
\ 1
jxrX
\ /
1
\
A
J
1
Downloaded by [Universite Laval] at 04:52 29 December 2014
'l I
/ 1
/ 1
Figure 5 Derivation of a predicted hexagonal central place hierarchy for lower and middle Viru.
Heavy dashed lines indicate margins of valley ; lighter dashed lines identify segments of hexagons
lying outside valley; other symbols same as fig. 2
1 Predicted locations of secondary centres define a hexagonal cell with a radius of 7-6 km.
around the primary centre ; four of the six secondary centres lie outside the limits of the valley
and would not be expected to exist.
2 Maximal packed hexagonal cells (fields of uncontested spatial control) are constructed around
the primary centre and the two expected secondary centres. The perimeter of the field
surrounding the primary centre consists of perpendicular bisectors of the radii of the hexagon
created in Step 1.
3 In order to satisfy Christaller's administrative principle (no sharing of dependent places),
smaller hexagons are inscribed in the hexagons of Step 2 by connecting the midpoints of
their sides.
4 The predicted sites of the tertiary centres are the midpoints of the sides of the hexagons
created in Step 3. Tertiary centres located in this manner define still smaller hexagons about
the primary and secondary centres.
292 Geoffrey W. Conrad
are those depicted in fig. 5.4. Since there are only six such centres in the lower and middle
valley, in some cases pairs of predicted sites must be viewed as alternative locations for
a single tertiary centre.
Figure 6 compares the spatial arrangement of secondary and tertiary centres in this ideal
hexagonal central place hierarchy with their actual distribution. As may be seen, the
expected and observed configurations are quite similar. Most of the centres lie within a
kilometre of their predicted location.
Villages would then be expected to cluster around tertiary centres. Villages do tend to
be arranged in this manner, but the pattern is not rigid (fig. 2), probably because tight
clustering of villages would greatly increase travel time to farmlands. None the less, of
the three predicted settlement patterns that have been presented, the one that optimizes
social and political control of the valley is clearly the closest to the actual site distribution.
Downloaded by [Universite Laval] at 04:52 29 December 2014
Synthesis
Moche settlement in middle and lower Viru did not optimize any single consideration:
none of the three predicted patterns corresponds exactly to the observed pattern.
Instead, the available data support Flannery's (1972: 418-19) more general argument
that in complex societies service functions outweigh other factors as determinants of
Models of compromise in settlement pattern studies 293
1
settlement. The predicted site distribution for resource maximization is very different
from the actual pattern, the expected distribution for minimization of effort is closer to
reality, and the predicted pattern for optimum administrative efficiency most nearly
approaches the observed configuration. This evidence indicates that the settlement
pattern imposed by the Moche state in lower and middle Viru reflects a compromise in
which maintaining social and political control of the local population was the most
important consideration, minimization of agricultural effort was a secondary factor, and
maximization of arable land was the least important determinant of the three. (This last
statement means only that the state did not attempt to restrict settlement from agri-
cultural land. I am not implying that increasing the amount of cultivable land was deemed
totally insignificant by the Moche state, or that it was not a highly important factor in
other state concerns - canal construction, for example.) Hence a model of compromise
derived from the Viru settlement pattern would rank the three determinants discussed
Downloaded by [Universite Laval] at 04:52 29 December 2014
Figures 7 and 8 depict the known distributions of Moche sites in the Santa and Nepeña
Valleys. The two patterns are not identical, nor does either seem to be the result of the
same compromise made in Viru. While the Santa pattern (fig. 7) shows some similarities
to that of lower and middle Viru, there are also some significant differences. In Santa,
'"I
N
Downloaded by [Universite Laval] at 04:52 29 December 2014
Figure 7 Moche phase settlement pattern in the Santa Valley (modified from Donnan 1973).
Key same as fig. 2, except that dashed lines indicate the limits oí modern cultivation. Sites are
numbered to facilitate comparison with the original report
Models of compromise in settlement pattern studies 295
unlike lower and middle Viru, secondary centres are not almost equidistant from the
primary centre. Their spacing probably reflects the topography of the Santa Valley,
which is tightly constricted by Andean foothills for nearly all of its length. Furthermore,
the Santa sites exhibit a greater tendency to lie along the valley margins. This pattern is
probably due, at least in part, to the destruction of bottomland sites by erosion or
cultivation (Donnan 1973: 11), but it may also be the result of narrow topography, as in
upper Viru. Finally, in contrast to the Viru settlements, which tend to be uniformly
distributed throughout the valley, the Santa sites occur in discrete clusters (Donnan
1973: 13). This clustering may be a defensive arrangement necessitated by Santa's
location near the southern frontier of the Moche state (fig. 1).
Moche settlement in the Nepeña Valley, the southern boundary of the Moche state
(figs 1, 8), displays even greater differences from Viru. The only administrative sites in
Downloaded by [Universite Laval] at 04:52 29 December 2014
PACIFIC
OCEAN
Villages
• Moche Ceramics Present
o Moche Ceramics Absent
Maximum Limit of Modern
Km
Cultivation
I I I I I l
0 1 2 3 4 5
Figure 8 Moche phase settlement pattern in the Nepeña Valley (modified from Proulx 1973);
except where indicated, key same as fig. 2. Sites are numbered to facilitate comparison with the
original report
Nepeña are a primary centre in the middle of the valley and a possible tertiary centre a
few kilometres upstream. This distribution neither restricts settlement from arable land
nor minimizes travel time to that land, and it certainly appears inadequate for maintain-
ing tight social and political control over presumably contemporaneous villages in the
upper and lower valley. Instead, the Nepeña pattern seems to mirror a primarily defen-
sive strategy in which nearly all state personnel were housed in a single protected location.
DWA
296 Geoffrey W. Conrad
While this emphasis on defence is not surprising in a frontier zone (Proulx 1973: 48),
it would not be predicted by the specific model developed in Viru.
Conclusion
I have attempted to show that the settlement pattern imposed by the Moche state in the
lower and middle Viru Valley did not optimize any single factor; instead, it was the result
of a compromise among a number of determinants. Three of those factors have been
identified and ranked as follows, in order of increasing importance: maximization of
arable land, minimization of agricultural effort, and maintenance of sociopolitical
control. This ranking was achieved by comparing the optimal settlement pattern for each
consideration to the actual configuration ; the closer the correspondence between expected
and observed patterns, the more important the factor in question as a determinant of
Downloaded by [Universite Laval] at 04:52 29 December 2014
settlement. Although only three considerations have been discussed here, other factors
should be subject to similar treatment.
Comparative data have been taken from three other provincial regions - the upper
Viru, Santa, and Nepeña Valleys. None of the four Moche settlement patterns that have
been examined is identical to any of the others. Each reflects a compromise, but those
compromises involved varying sets of determinants. Where determinants were shared,
their relative importance often differed from case to case.
The implication to be drawn from these results is that the sub-optimal behaviour
inherent in the compromises (and satisficing decisions) made by complex societies
cannot be foretold with certainty. The understanding of one prehistoric compromise
does not automatically convey knowledge of all the considerations involved in another,
nor does it necessarily indicate the relative importance attached to similar factors in
different cases. Cowgill (1975: 515) has summarized the matter neatly:
The problem is not that human behavior is unintelligible, at least after the fact, but rather
that we cannot always predict which of several reasonable choices will be made, or, if there
are contending interests, which will prevail in a specific instance.
Acknowledgements
I would like to thank Drs Gordon R. Willey and Garth L. Bawden for their comments
on a draft version of this paper. I am also grateful to Barbara Westman, who prepared
the illustrations.
2.VÍÜ.1977 Dept. of Anthropology
Harvard University
References
Adams, R. McC. 1969. The study of ancient Mesopotamian settlement patterns and the
problem of urban origins. Sumer. 25:111-24.
Bennett, W. C. 1950. The Gallinazo Group: Viru Valley, Peru. Yale University Publications in
Anthropology. 43.
Models of compromise in settlement pattern studies 297
Publications in Anthropology. 8.
Flannery, K. V. 1972. The cultural evolution of civilizations. Annual Review of Ecology and
Systematics. 3:399-426.
Haggett, P. 1966. Locational Analysis in Human Geography. New York: St Martin's.
Keatinge, R. W. 1975. Urban settlement systems and rural sustaining communities: an example
from Chan Chan's hinterland. Journal of Field Archaeology. 2:215-27.
Kroeber, A. L. 1930. Archaeological Explorations in Peru, Part II: the northern coast. Field
Museum of Natural History Anthropological Memoirs. 2, no. 2.
Lamberg-Karlovsky, C. C. 1975. Third millennium modes of exchange and modes of
production. Ancient Civilization and Trade, pp. 341-68 (eds J. A. Sabloff and C. C. Lamberg-
Karlovsky). Albuquerque : University of New Mexico Press.
Larco Hoyle, R. 1948. Chronología Arqueológica del Norte del Perú. Buenos Aires: Sociedad
Geográfica Americana.
Leach, E. 1973. Concluding address. The Explanation of Culture Change: models in prehistory,
pp. 761-71 (ed. C. Renfrew). London: Duckworth.
Marcus, J. P. 1973. Territorial organization of the lowland classic Maya. Science. 180:911-6.
Morris, C. 1972. State settlements in Tawantinsuyu : a strategy of compulsory urbanism.
Contemporary Archaeology: a guide to theory and contributions, pp. 393-401 (ed. M. P. Leone).
Carbondale: Southern Illinois University Press.
Moseley, M. E. 1975. Prehistoric principles of labor organization in the Moche Valley, Peru.
American Antiquity. 40:191-6.
Murra, J. V. 1958. On Inca political structure. Systems of Political Control and Bureaucracy in
Human Societies, pp. 30-41 (ed. V. F. Ray). Seattle: University of Washington Press.
Plog, F. and Hill, J. N. 1971. Explaining variability in the distribution of sites. The Distribution
of Prehistoric Population Aggregates, pp. 7-36 (ed. G. J. Gumerman). Prescott College Anthro-
pological Reports. 1.
Proulx, D. A. 1968. An Archaeological Survey of the Nepeña Valley, Peru. University of
Massachusetts, Department of Anthropology, Research Report. 2.
298 Geoffrey W. Conrad
Willey, G. R. 1953. Prehistoric Settlement Patterns in the Viru Valley, Peru. Bureau of American
Ethnology Bulletin. 155.
Abstract
Conrad, G. W.
Models of compromise in settlement pattern studies: an example from coastal
Peru
The settlement patterns of complex societies require compromises among numerous deter-
minants. In the case of the lower and middle Viru Valley, a provincial region of the Moche state
of coastal Peru (A.D. 200-700), three of these considerations can be identified and ranked in
order of increasing importance as follows: maximization of agricultural land, minimization of
agricultural effort, and maintenance of sociopolitical control. This ranking is achieved by
deriving the settlement pattern that would optimize each factor and then comparing those
optimal site distributions to the actual configuration; the greater the similarity, the more
important the factor in question as a determinant of settlement. Finally, comparative data from
other provincial regions of the Moche state are presented in order to show that their settlement
patterns also reflect compromises, but not necessarily the one made in Viru.