Download as pdf or txt
Download as pdf or txt
You are on page 1of 13

Society for American Archaeology

Downslope Movement and Archaeological Intrasite Spatial Analysis


Author(s): John W. Rick
Source: American Antiquity, Vol. 41, No. 2 (Apr., 1976), pp. 133-144
Published by: Society for American Archaeology
Stable URL: http://www.jstor.org/stable/279164
Accessed: 02/06/2009 15:38

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at
http://www.jstor.org/action/showPublisher?publisherCode=sam.

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.

JSTOR is a not-for-profit organization founded in 1995 to build trusted digital archives for scholarship. We work with the
scholarly community to preserve their work and the materials they rely upon, and to build a common research platform that
promotes the discovery and use of these resources. For more information about JSTOR, please contact support@jstor.org.

Society for American Archaeology is collaborating with JSTOR to digitize, preserve and extend access to
American Antiquity.

http://www.jstor.org
DOWNSLOPE MOVEMENT AND ARCHAEOLOGICAL
INTRASITE SPATIAL ANALYSIS

JOHNW. RICK
Erosion can produce heterogeneous distributions of cultural materials of differing densities, weights, and
shapes within an archaeological site. A large, provenienced surface collection from a preceramic site in highland
Pert is examined for evidence of downslope movement as an explanation for observed heterogeneity in artifact
distribution. Statistical methods of spatial analysis are used to define the extent of movements of surface
materials in this site. For this particular area of Perui, critical angles for downslope movement will limit the site
areas in which analysis of horizontal distributions can be expected to produce culturally significant information.

ARCHAEOLOGISTSCURRENTLYEMPHASIZEintrasite spatial analysis when identifying


non-randomdistribution of artifacts. When localized clusters of artifacts are found on a living
floor, they are usually interpreted as the fossilized remains of human activities. Recurrent
covariationin frequencies of tool types within these activity areashas been used to define tool
kits. These tool kits are thought to have been groups of tools used together to performspecific
tasks. In sites containing architecturalfeatures, rooms or other walled areas can be used to
partition the site into units which can be examined for such spatial information.In Paleolithic
sites, which usually lack architecturalremains,the observedclusteringof artifacts and featuresis
interpreted as the significant cultural structure. These areas are then used as the basis of any
inferenceconcerningbehaviorpatterns.
Since these spatial distributionsplay such an important role in archaeologicalanalysis, it is
surprisingthat very little attention has been paid to the human and natural processeswhich can
destroy or change culturally created patterns of artifact distribution.Both the post-depositional
activities of man, and variousforces of nature can act to bias or destroy the culturalpatternsof
artifacts. In well-stratifiedsites, which are most promisingfor spatial analysis, post-depositional
human activity can often be identified. Ditches, road cuts, or other limited soil disturbancewill
stand out in sharp contrast to undisturbed strata. Post-depositional changes caused by easily
recognizablenaturalforces include modernerosion, rockfalls,or landslides.Othernaturalagencies,
especially those acting on artifacts soon after their deposition, may not leave obvious signs, but
still can completely alter artifactdistributions.
The most common of these forces is probablyerosion. Erosionis defined as "the wearingaway
of the earth's surface and the transportationof detritus to lower levels, either on land or in the
sea" (Gilluly, Waters,and Woodford1968:61). For archaeologicalpurposes,erosion is a downward
movement of culturalmaterial,the impetus for which is providedby gravity,precipitation,wind,
streams,oceans, or such organicforces as plants and animals.In archaeology,some attention has
been paid to the role that streamsand washingmay play in artifactdistributions.Isaac(1968) has
demonstratedthat size sorting of large and small stone objects is characteristicof washing,and
that perhaps certain alignments of artifacts may be caused by water currents. In a controlled
study, he found that fluvial forces moved stone flakes more quickly than larger and heavier
concrete replicasof handaxes. Davis (n.d.) has describedrelationshipsbetween stream forces and
the surface area of artifacts such that objects with greatersurface area will move at lower stream
velocities, if densitiesare equal. For fluviallyinduced movementmore energyis requiredto move a
more dense object than a less dense one. Thus, specific distribution patterns of artifacts with
different surface areas and densities could serve as indicators of fluvial erosion. Both of these
studies are concerned with Lower Paleolithic sites in eastern Africa, where sites tend to appear
along stream margins or lakeshores. In these locations, fluvial forces may have been quite
important. On the other hand, erosionalmovementat other sites may requirea differentmodel-I
refer primarilyto sites on steep slopes, where the force of gravity plays the majorrole in moving
artifacts downslope, and wind and water are lesser factors in this movement. This paper is a
discussionof downslopemovementin archaeologicalsites.
133
134 AMERICAN ANTIQUITY [Vol. 41, No. 2, 1976

DOWNSLOPE MOVEMENT: A CASE EXAMPLE


During the summer of 1973, a group of archaeologists from the University of San Marcos
(Lima, Peru) and the University of Michigan (Ann Arbor) made preliminary studies of preceramic
and later sites in the area of Junin, Peru. Junin is a small town located about 140 km (linear
distance) to the northeast of Lima at an altitude of 4175 m (13,325 ft). The town of Junin lies
just south of Lake Junin, the second largest lake in Peru. The plateau in which the lake lies has a
tundra or puna environment, characterized by gently rolling hills covered with bunch grasses and
low herbs, which are closely cropped by llama and sheep.
During the field season, more than a dozen preceramic sites were examined. Dr. Ramiro Matos
Mendieta of the University of San Marcos has carried out excavations at a number of these sites,
and preliminary observations indicate that human occupation began at least 7,000 years before the
first appearance of ceramics at 1800 B.C. The greatest quantity of preceramic occupation debris is
found in caves and rockshelters in the limestone bedrock, which outcrops on hillslopes or in bluffs.
Quite often there is only a small, level cave chamber or sheltered area above a much larger slope
below. Cultural remains are distributed down the slope over distances of 20 to 300 m, with the
angles of the slopes varying from less than 10? to 44?.
It might seem that erosion should take place at sites with such steep gradients. However, the
extent to which it would have taken place depends upon both the slope and the vegetation cover.
In Junin, most slopes have a dense growth of bunch grasses which typically covers at least 50% of
the ground surface. These grasses, if present in the past, would have done much to prevent
downslope movement of archaeological materials. Therefore, although downslope movement
would be expected at slopes above about 25?, it cannot be assumed to occur on lesser gradients.
For this study, a large provenienced surface collection was taken from a rockshelter site called
Ccurimachay, which has a long, concave slope spanning the typical range of slope angles (see Fig.
1). Six contiguous lines of 33 3 X 3 m squares were collected from the top to the bottom of the
site, while three lines of collection units were placed at right angles, approximately following the
contour of the slope (Fig. 2). All cultural materials-bone, chipped stone, and ceramic
fragments-were collected from a total of 252 3 X 3 m units. Pottery, betraying the presence of
Formative components, which presumably overlie the preceramic, was encountered on the
surface-a situation typical of sites of this type. These collections were washed, labelled, counted,
and weighed and all lithics were examined for a series of technological and morphological
attributes.
The data for this study have been analyzed through the use of the IBM 360 computer at the
University of Michigan. The Michigan Interactive Data Analysis System (MIDAS) program
developed by the Statistical Research Laboratory of the University of Michigan was used for the
statistical computing, data analysis, and graphic display of results.

ANALYSIS OF THE MATERIAL DISTRIBUTIONS


For the majority of these analyses, only the six vertical rows of collection units will be used,
since they span the complete slope. Seven measurements were made for each unit: the slope in
degrees, the absolute frequencies of bone, ceramic, and lithic fragments, and the average weight
(total weight/count) of each piece for each category (bone, ceramics, and lithic material). The
measurements for each row of six squares across the vertical lines were averaged in order to give a
more uniform picture of material distributions. This pooling seems reasonable, since the site
profile is unchanging within the area considered, and therefore all the units which are pooled have
the same slope and reflect similar downslope movement conditions. Visual inspection revealed no
significant variation in counts or weights across the vertical rows. An examination of Figs. 3 and 4
will give an idea of the nature of this slope. Fig. 3 is a profile of the site surface and Fig. 4 is a
graph of slope angle over distance. One should note that the peaks on this graph do not represent
slope reversals, but are steep slopes followed by less steep ones.
When the seven measurements are plotted from top to bottom of the slope, three general
patterns are suggested. First, all three frequency measurements (lithic, bone, and ceramic) show
Rick] DOWNSLOPE MOVEMENT 135

Fig. 1. View of site Ccurimachay. Dark spot near center of photograph is the actual shelter, while the slope
collected for the purposes of this study lies below.

positive correlation with slope angle, and, therefore, among themselves: the greater the number of
pieces observed, the steeper the slope, and vice versa (compare Figs. 4 and 5). Second, the average
weights of lithics, bone, and ceramics plotted against slope angle show a tendency toward negative
correlation: the greater the slope, the smaller the average piece of each material and vice versa (see
Fig. 6a,b,c). Third, as a logical consequence of the first two observations, counts and average
weights are negatively correlated: when the number of pieces of any material is large, they tend to
be light, and when the number is small, the average weight is greater. Another tendency is observed
in the graph of the material frequencies (Fig. 5). Bone is found in greatest numbers near the top of
the site, where slope angle is larger; lithic material tends to be found in larger numbers near the
bottom of the site, where slope angle is smaller. Ceramics have frequencies which lie between those
of bone and lithics.
The frequency curves for all the materials show a number of marked peaks that violate the
general upward or downward trends. Most of these peaks represent the effects of recent field
boundaries-vertical cuts 30-40 cm in height made at three points up the slope. The increased
artifact frequency at these points is due to the erosion of materials out of the unprotected, vertical
face of the cuts. According to local inhabitants, the cuts are not more than seven years old. It is
therefore probable that the overall material distributions reflect conditions of downslope
movement previous to the construction of the boundaries.
Since much of the lithic material predates all of the ceramics from this site, the earlier lithic
materials would have had more time to be transported through downslope movements, and greater
136 AMERICAN ANTIQUITY [Vol. 41, No. 2, 1976

L7LJL BLUFF
EDGEK

9METERS

MODERN ROAD Fig. 2. Plan of 3X3 m collec-


tion units at site Ccurimachay.

representation of these materials might e expected on the lower parts of the site. The frequency
curves (Fig. 5) show such an increased frequency of lithic material downslope, while the ceramic
frequency is greaterin the upperareasof the slope. I doubt that age is the primaryexplanationfor
this difference, however. It is doubtful that any of the materialspresentlyobservedon the surface
moved downslope before 1800 B.C. (the date of the first appearanceof ceramics).I would expect
that earliermaterialshave been covered by accumulationof organicmaterialor soil washeddown
from above. The presence of artifacts 25 cm below the surface in the profiles of cuts near the
bottom of the site tends to confirmthis conclusion.
Decomposition of bone duringthe process of downslope movement might tend to reduce the
quantities of bone found in the lower parts of the slope. The frequency of bone does decrease
toward the bottom of the site, althoughceramicsalso decreasein similarproportions.However,if
bone decomposition were a dominant process, it would also follow that averageweight of bone
should decrease downslope, reflectingthe breakdownof bone fragments.This, however,is not the
case. The averagebone weight does decrease over a limited area of the slope, but the average
Rick] DOWNSLOPE MOVEMENT 137

weight of bone fragments near the bottom is as large as that of those from the middle of the slope.
Therefore, although some bone decomposition may occur, it is not the determining factor for
bone surface distribution.

SHELTER

CCURIMACHAY:
SLOPE PROFILE

IOM

VERTICALEXAGGERATION:
TIMES 1.6
1IOM

Fig. 3. Profileof site Ccurimachay.

At each step in this analysis, carewas taken to identify any culturalinformationpresent.Cross


plots were made among all seven of the variables,and the plotted points were examinedfor linear,
curvilinear,or bilinearcovariationbetween the two plotted variables.These patterns,as well as any
discrete clusteringsof points could requireculturalexplanations.In five of these cross plots the
plotted points formed two lines; that is, bilinear relationshipswere suspected. These pairs of
variableswere therefore transformedinto ratios by dividingone variableby the other. They were
then plotted againstthe five remainingvariablesto see if the bimodalnatureof their ratio related
to variationin any of the other variables.In all pronouncedcases, the bimodalityof the ratio was
explained by a slight bimodality in the distributionof slope angles;the remainingcases did not
suggestany culturalinformation.
The next step in analysiswas to compute least squareslinearregressionsof the frequenciesand
average weights of bone, lithics, and ceramics against slope. This procedureshould reveal any
signifilcantinformation remainingin the materialvariablesafter the variationexplained by slope
has been removed. A series of regressionswere run to determineif usingthe logs of the dependent
(counts and averageweights) or of the independent(slope) variables,or if usingmovingaveragesof
the slope would improve the fit of the linear model. For all six cases, the best fit was achieved
using a moving averageof the slope, which averagedthe slopes of the collection unit concerned
with the units immediatelyabove and below it. The next best fit was achievedby leavingboth the
dependent and independent variables unaltered. Table 1 shows the R-square, F-ratio, and
significancestatistics for the best fit regressions.As can be seen, the regressionsfor the frequencies
all had quite high F-ratios,and are all highly significant.Slope thus explainsa high percentageof
the variation. The regressionfor the averageweight of lithic materialslikewise explains a high
138 AMERICAN ANTIQUITY [Vol. 41, No. 2, 1976

percentage of the variation, but slope accounts for much less of the variation of ceramic average
weight. The regression for bone is still worse, and does not reach significance even at the 0.20
level. Therefore, slope is an accurate or fairly accurate predictor of the average weight of lithic and
ceramic fragments, but is much less accurate for average weight of bone.

Table 1. Regression values for dependent variables when regressed individually against slope angle.

Dependent Variable R-Square F-Ratio Significance

Lithic Frequency .463 19.8 .0002


Bone Frequency .794 88.8 .0000
Ceramic Frequency .901 211.0 .0000

Lithic Av. Weight .570 30.5 .0000


Bone Av. Weight .065 1.6 .2175
Ceramic Av. Weight .169 4.7 .0408

.30

25

.eo

2O

Fig. 4. Slope angle from top (left) to bottom (right) of site Ccurimachay.

Cross plots of the residuals from these regressions were made and examined for variation that
might suggest cultural patterning. The plots in general showed a random scatter, suggesting that no
significant information remained unexplained. It is reasonable to hypothesize that downslope
movement was responsible for distribution patterns noted in the data.
Rick] DOWNSLOPE MOVEMENT 139

THE PROCESS OF DOWNSLOPE MOVEMENT


The geomorphological literature identifies many processes by which materials move downslope.
For Junfn, the majority of these can be eliminated. Processes characteristic of glacial or periglacial
climates do not apply to this area; climatic data from nearby Morococha (Bowman 1916:171-76)
suggest that the ground of this region normally does not freeze.
Soil creep as described by Sharpe (1938:22-29) may occur at Ccurimachay, but this process
would carry superficial materials along more or less in situ, and would not differentially distribute
artifacts according to materialand weight. Similarly,downslope movement caused by expansion
and contraction of materials when heated and cooled, or dampened and dessicated, does not
provide an explanation for the observed distributions.

/00

75

/-\ i!\ \ "5

.....^ ^J \ ..25

Fig. 5. Absolute frequency of lithics (A), bone (B), and ceramics (C) from top (left) to bottom (right) of site
Ccurimachay.

Processes where material is transported by water, air, or ice do not seem to apply.
Climatological data allow us to eliminate ice and wind, and the presence of heavier materials at the
bottom of the slope differs from the data of Isaac (1968) and the model of Davis (n.d.), both of
which suggest that the heaviest materials would be moved the least by the passage of water.
Therefore it is doubtful that fluvial processes were important at Ccurimachay.
The processes which carried the materials downslope at Ccurimachay must have depended on
gravity for most of their motive force. At the time of deposition, gravity may move an object, or it
may need a further impulse to start it downslope. The inertia of objects at rest on the slope may
be overcome when they are undermined by eolian or fluvian action, and subsequently drop a
shortdistance; or humans or animals may dislodge objects and start them downslope.
Once in motion, the objects will continue until their kinetic energy of motion is dissipated by
friction or an obstruction. Kirkby and Statham (n.d.) have argued that larger rock fragments travel
further down scree slopes because they are less likely to be caught in spaces between rocks at rest
on the slope. Ccurimachay, being covered with soil and plants, is not a scree slope, and does not
have much non-artifactual rock on its surface. Small depressions or flat spots at Ccurimachay may
act like the pockets of a scree slope, and trap downward moving materials. If so, then larger pieces
140 AMERICAN ANTIQUITY [Vol. 41, No. 2, 1976

would be expected near the bottom of the slope.


The grass on the slope of Ccurimachay is undoubtedly responsible for stopping much of the
materialmoving downslope. Objects passingthroughsuch a resistingmediumwill travela distance
proportional to their initial velocity, density, and diameter (M. J. Kirkby, personal communica-
tion). In this case, largerand denser materialswill be found nearerthe bottom of a slope if their
initial velocities were equal. Objectsof any size or density will have nearlyequal initial velocities if
started by falls of equal height. Largerand denser objects started by a fixed impulse will have
slower initial velocities than smallerand less dense ones. However,variationsin size and density
will not greatly alter the initial velocity of objects which are very small comparedto the mass that
impartsthe startingimpulse. At Ccurimachay,all the animalswhich might kick objects downslope
are very largein relationto the size of the materialI collected.

35yi 8

M ;',1
01

\,

Fig. 6a. Average weight per piece of lithics (solid line) compared with slope angle (dotted line) from top (left)
to bottom (right) of site Ccurimachay.

In considering downslope motion at Ccurimachay, a number of practical problems must be


faced. First are the shapes of the objects. A round object is considerablymore apt to roll than a
flat one. Since rolling entails less friction than sliding, round objects will tend to travel farther
before their energy is lost. A bounding object, touching the ground less than a rolling one, loses
even less energy. From empirical observation, very few artifact shapes collected at Ccurimachay
could roll on this slope, if merely dislodged by a passing foot. Any of the objects, however, if given
a faster initial velocity, will roll or bound regardless of shape. However, it is difficult to imagine
that many of the artifacts were given initial velocities equivalent to a strong throw or kick by a
human. Objects given a slower initial velocity may slide or perhaps make a few somersaults before
being stopped by the vegetation on the slope.
Quantification of the size of materials recovered from Ccurimachay also represents a problem,
since more than 14,000 pieces were recovered from the surface. A much simpler method of
evaluating the average size of a given material is the average weight per piece (weight/count).
Within a class of material which has unvarying density, average weight measurements will be an
accurate reflection of average size.
Rick] DOWNSLOPE MOVEMENT 141

The three different materialsfrom Ccurimachayare surprisinglysimilarin shape. Both bone and
ceramicfragmentstend to be slightly curvedbut basicallyplanar.Lithic artifactsare more variable,
but the vast majority are flake or flake fragments,which are mostly flat. Very few lithic artifacts
approach roundness;understandablythose that do tend to be found at the bottom of the site.
However,very few objects appearsufficiently round for rollingto have affected their distribution.

35r 5

I\ 1

30L.
/I \ A 4
'\ C!(:)
/i
^ \
, A
Kz

/I
C,) /
'I5
rT- 1 I.
I /
I0o I
I
I
I __ _ i
Fig. 6b. Average weight per piece of bone (solid line) compared with slope angle (dotted line) from top (left)
to bottom (right) of site Ccurimachay.

DOWNSLOPE MOVEMENT AT CCURIMACHAY


The actual patterns of material distribution from Ccurimachay can now be compared with the
model outlined above. Perhapsthe best-definedrelationshipis between slope and frequencyof the
different materials.The regressionsdemonstratethat frequenciesof bone and ceramics,and slope,
vary directly with each other. This covariation results from a number of causes.
First, it appears that the majority of the midden deposit was originally near the top of the site;
material frequencies should decrease at greater distances from the site's center. Since the slope
angle also decreases away from the upper part of the site, a positive correlation would be expected.
It should be noted, however,that lithic frequenciesdo not decreasetowardthe bottom of the site.
Another factor which may be responsible is exposure and depositional burying of materials.
The surface materials moving down the steepest slopes are deposited and subsequently disappear
on lower slopes when organic matter and silt wash down from above.
A third explanation for this relationship requires consideration of the variable of average
weight. It has been shown that averageweight per piece of all the materialsis inverselycorrelated
with slope, although the averagebone weight/slope regressionwas notably less significantthan
those of the other two materials.If the slope angle controls the averageweight of the materials,
this would explain the slope/frequency correlation.Histogramsof the weight distributionsof all
three materials have unimodal distributions with right skews, indicating that there are relatively
fewer heavy objects than lighter ones. Therefore, gentler slopes which select uniquely for heavier
142 AMERICAN ANTIQUITY [Vol. 41, No. 2, 1976

objects will have less material, since there are fewer heavier objects. Since slope angles decrease
downslope at this site, it follows that frequencies should decrease, and weights increase downslope.
It was noted that lithics tend to be found in relatively larger numbers at the bottom of the site
than ceramics, and that ceramics are found at the bottom in relatively larger numbers than bone.
The differences between bone and lithics may be duedue to weight differences: the overall average
weights for lithics and bone are, respectively, 3.8 and 2.2 grams. Selection for heavier objects on
smaller slope angles may thus account for the presence of more lithics near the bottom of the
slope compared with bone. However, ceramics have an average weight of 2.4 grams-similar to that
of bone-yet significantly more ceramic fragments are at the bottom of the site than bone. This
may reflect the tendency of materials with greater density to move to lower and gentler slopes.
The regressions of lithics and ceramics against slope suggests that the average weights of these
materials are inversely correlated with slope angle, although the average bone weight/slope
regression is not definitive. Average lithic weight increases where slope decreases (Fig. 6a);
consequently average lithic weight increases from the top of the site toward the bottom. Ceramics
follow this pattern for the upper two-thirds of the site (Fig. 6c); below this, the inverse
relationship seemrnsto disintegrate. Bone reflects the inverse relationship for only the uppermost
portion of the site (Fig. 6b), after which no particular relationship can be observed between the
two variables. In summary, this inverse relationship holds for the heaviest and densest material
from the whole site, for the material with intermediate density and weight from the steeper
two-thirds of the site, and for the material with the lowest average weight and density only for the
very steepest slopes. Slope angle seems to control distributions of the different weights of
materials, but requires a minimum weight or density, combined with a minimum slope, for the
operation of this control.

35
-4.0
/\

/ . -
\ \3.5

'/5. , ~ / ' .2.0


/ ^\

Fig. 6c. Average weight per piece of ceramics (solid line) compared with slope angle (dotted line) from top
(left) to bottom (right) of site Ccurimachay.

Because of the concave slope, slope angles at Ccurimachay decrease from top to bottom. For
any moving object, the greater the slope, the smaller will be the amount of kinetic energy that
friction absorbs from a moving object. Therefore, the distance downslope that a disturbed object
Rick] DOWNSLOPE MOVEMENT 143

moves is decreasedas slope decreasesuntil the point where an object will not continue downslope
to a significant degree. Lighterobjects will come to rest on steeper slopes, while heavierobjects
will continue toward gentler slopes. If there is a finite upperweight limit in a given material,then
there is a slope anglebelow which that materialwill not be moved.
A negativecorrelationbetween slope angleand averageweight should thereforeexist in all areas
affected by this type of downslopemovement.Below a certainangle,however,the correlationwill
cease. On a concave slope, the point where the correlationstops will be representedby an average
weight peak. If all of the materialis transporteddown from above, then the frequency of the
materialshould drop off below this point. I choose to call such points criticalangles.
The averageweight of bone shows such a peak where the negativecorrelationwith slope stops
(Fig. 6b). Below this point, slope and averageweight behave independentlyof each other. Bone
frequency drops off sharplybelow this peak (Fig. 5); this explainsthe sharppeaksand depressions
at the right side of the averageweight graph.Since there are only an averageof 2.7 pieces of bone
in the collection units in this area, one largefragmentcan account for a very largeaverageweight.
If the peak on the bone curve representsthe deposition point for the heaviest fragments,then
bone has a critical angle of 23-25?. Using the same techniques, the critical angle for ceramic
deposition appearsto be 18-20? (see Fig. 6c). In general,averageweight of lithics increasesto the
bottom of the collected area(Fig. 6a). Since the collection was terminatedat the edge of the lithic
debris, the right-handpeaks probably represent the critical angle for lithic movement: about
16-17?.
The regressionof averageweightsagainstslope showed the greatestsignificancewith lithics, the
least with bone, and an intermediatevalue with ceramics.Since downslope movement does not
control averageweight of bone below 23-25?, a significantareaof the site should lack a correlation
between slope and bone weight; hence the lack of significancein the regression.Slope seems to
control the averageweight of lithics over all the area considered,causingthe regressionbetween
averagelithic weight and slope to be quite good. Ceramics,with an intermediatecriticalangle of
18-20?, have a regressionwith expected intermediatedegreeof significance.
These are the critical angles at which the heaviest objects of a given materialare deposited.
Lighter pieces deposited at higher slopes will have correspondinglygreatercritical angles. Spatial
analysis of these lighter objects would be worthwhile on slopes of greater angles. To test this
proposition, frequencies of lithic tool types of greatly differingweights were recordedfor four
classes of slope. These classes representedsegments of the total rangeof slope angle, divided so
that one-fourth of the collection units fell into each class. Of the 121 projectilepoints (average
weight: 3.4 g), 87, or 72%,were found in the two steeper slope classes.In contrast,21 of the 33
cores (averageweight: 57.5 g), or 64%,were found in the squareof the two classesof lesserslope.
Choppers-tools of intermediateweight-had intermediatevalues: approximatelyhalf were found
in units with less than the medianslope, and half were found on slopes greaterthan the median.
Therefore,heaviertools are found on lower angle slopes,
an d lighterones tend to be found on
high angle slopes. Projectile points are not moved as far downslope as cores, and therefore their
critical angle is larger.It appearsthat no significantnumber of projectile points are deposited in
squareswith less than 20? slope.
It would be foolish to underestimatethe degree to which downslope movementscan create
patterned distributions of cultural material. For instance, a chi-squaretest of independence
between cores and projectile points at Ccurimachayrevealsa lack of co-occurrenceof these two
tools significantat the .12 level. Similarly,bone frequencieshigherthan the medianand projectile
points co-occur with a chi-squaresignificanceof less than .001. Whilethese simple statisticsclearly
do not define activity areas, they could be misconstruedand used as the basis for furtheranalysis.
Such interpretationwould be unwarrantedin the light of the nonculturalprocessesthat produced
this patterning.

CONCLUSION
This study has archaeologicalapplication in both specific and general circumstances.From
144 AMERICAN ANTIQUITY [Vol. 41, No. 2, 1976

Ccurimachay with its uniform grass cover and consistent site morphology it may be possible to
hersites. Using these critical angles, excavation could be
extrapolate the critical angles for use in other
avoided in parts of puna sites which would very likely be eroded. For these angles to be truly
useful, a knowledge of paleoclimate will be essential, both for determining the extent of past
ground cover, and also for understandingchangesin precipitation.The stratigraphiccharacteristics
of each site will determine which layers may have been subject to erosion. Also the increased
thickness of the main occupation deposits at the top of some slopes may have caused strata to dip
more steeply as time passed.
For at least one type of downslope movement, heavierand denser objects are moved further
downslope than lighter and less dense ones. This is especially important, since it is exactly the
opposite of the effects of fluvial erosion. Sloping sites subject to fluvial action may have been
affected by both downslope movement and fluvial action, and would requireadditional study.
When concerned solely with the process of downslope movement, the lightest and least dense
materials will better reflect the original distribution of materials than heavy, denser materials.
Other archaeologists should take the time necessary to look for signs of erosion to help expand our
knowledge of this process, for it not only can destroy cultural data, but also create data of its own.

Acknowledgments. The data analyzed here were collected during fieldwork made possible by Kent V.
Flannery, Ramiro Matos Mendieta, and Jeffrey R. Parsons. The following friends and colleagues aided me greatly
in the actual data collection: Michael Brown, Victor Chang, Daniel Morales, and Emilio Ramos. Robert Whallon,
Jr., Charles M. Rick, Jr., William R. Farrand, and Michael J. Kirkby read and made very helpful comments on
previous versions of this paper. However, I alone bear responsibilities for persisting deficiencies.

Bowman, Isaiah
1916 The Andes of southern Peru. Henry Holt.
Davis, David D.
n.d. Some procedures for the delineation of activity areas on paleolithic occupation floors. Dept. of
Anthropology, Yale University. Mimeographed.
Gilluly, James, Aaron C. Waters, and A. 0. Woodford
1968 Principles of Geology. W. H. Freeman, San Francisco.
Isaac, Glynn L.
1967 Towards the interpretation of occupation debris: some experiments and observations. Kroeber
Anthropological Society Papers 37:31-57. Berkeley.
Kirkby, Mike, and Ian Statham
n.d. Surface stone movement and scree formation. Journal of Geology. (In press, ms. 1975.)
Sharpe, C. F. Steward
1938 Landslides and related phenomena. Columbia University Press, New York.

You might also like