Professional Documents
Culture Documents
The Archaeological Visibility of Storage
The Archaeological Visibility of Storage
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted
digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about
JSTOR, please contact support@jstor.org.
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
http://about.jstor.org/terms
Society for American Archaeology is collaborating with JSTOR to digitize, preserve and extend access to
American Antiquity
This content downloaded from 132.248.9.8 on Tue, 09 May 2017 14:31:12 UTC
All use subject to http://about.jstor.org/terms
THE ARCHAEOLOGICAL VISIBILITY OF STORAGE:
DELINEATING STORAGE FROM TRASH AREAS
Susan Kent
Despite the importance attributed to the study of storage behavior, little research has been conducted to determine whether it
is even possible to distinguish storage areas from refuse areas. Archaeologists routinely separate storage pits from trash pits,
but few have systematically investigated the defining characteristics of each. This study suggests that there is an archaeolog-
ically visible signature that can help researchers correctly interpret these loci. Research at occupied and recently abandoned
camps among the now sedentary residents of Kutse in the Kalahari Desert of Botswana shows that refuse areas have a more
homogeneous artifact inventory, regardless of the number of objects present. In contrast, non-trash activity areas at the same
camps have a more heterogeneous, or diverse, inventory. The applicability and utility of this finding to the archaeological
record is evaluated through the analysis of a Pueblo II Anasazi archaeological site from the southwestern United States.
Patterns first recognized ethnoarchaeologically also appear to be recognizable in the archaeological record using the same
methods. The results indicate that the statistical tests described here are applicable to distinguishing trash from other activ-
ity areas at archaeological sites.
A pesar de la imporancia atribuida al estudio del almacenamieno, no se ha conducido suficiente investigaci6n para determinar
si es posible distinguir entere areas de almacenamiento y areas de desecho. Los arqueologos rutinariamnente separan pozos de
almacenamiento de los de desecho, pero pocos han estudiado sistematicamente las caracteristicas que definen cada tipo de p
Este estudio sugiere que existe una huella visible arqueologicamente, la que puede ayudar a investigadores a interpreter cor-
rectamente estos rasgos. Investigaciones conducidas en campamentos ocupados o abandonados recientemente por los residentes
sedentarios de Kutse en el Desierto Kalahari en Botswana indican que las dreas de desecho tienen un inventario artefactual mds
homogeneo, independientemente del narmero de artifactos presentes. En contraste, las dreas de actividad y almacenamiento
tienen un inventario mds heterogineo o diverso. Se investiga la aplicabilidad y utilidad de este hallazgo a travis del andlisis de
un sitio arqueolkgico. Anasazi del periodo Pueblo II en el suroeste de los Estados Unidos. Los patrones reconocidos etnoarque-
ologicamente tambijn se reconocen en el registro arqueolkgico utilizando los mismos metodos. Los resultados indican que las
pruebas estadlsticos descritos aquf son aplicables para distinguir descho de otras dreas de actividad en sitios arqueologicos.
79
This content downloaded from 132.248.9.8 on Tue, 09 May 2017 14:31:12 UTC
All use subject to http://about.jstor.org/terms
80 AMERICAN ANTIQUITY [Vol. 64, No. 1, 1999]
will be examined, I am intentionally not including include huts, windbreaks, formal and informal stor-
the storage of food in this study because agricul- age areas, trash/ash areas, kraals, and so on). This
tural storage bins, granaries, and similar features study includes 20,799 objects located at 247 indi-
tend to be archaeologically recognizable. However, vidual features. At the time of recording, most
nonfood storage areas are as common or more objects inventoried at non-refuse areas were in
common than trash areas in many societies with informal or formal storage. Only a very small num-
various economic emphases (Kent 1998). ber of objects were in actual use at the time of the
Kutse inventories.
Ethnoarchaeological Storage
To test the applicability of the ethnoarchaeolog-
and Trash Patterns
ical patterns in an archaeological context, the
There are two types of storage: informal and for- observations obtained from the data collected eth-
mal. All modem human societies use informal stor- noarchaeologically at Kutse are applied to data
age areas. Informal storage is the placement of recovered archaeologically from a mid-Pueblo II
objects on or in areas that are not specific in func- Mesa Verde Anasazi site located in the southwest-
tion for storage. Putting a spear or metate by the ern United States. The site has well-defined and
side of a hut when not in use is an example of visible activity areas, so that midden or trash
informal storage. Informal storage areas consist of deposits can be readily distinguished from storage
hut or windbreak walls or ceilings, trees, and loci. This makes the Anasazi site an ideal archaeo-
bushes. The use of informal storage areas is ubiq- logical test case.
uitous, but is rarely visible in the archaeological Because I want to compare material assem-
record. Sedentary groups, however, also have for- blages from different sites, it is important to keep as
mal storage areas, such as storage rooms or parts of many factors constant as possible. The Anasazi site
rooms, where objects not in use are placed (Kent examined here is a relatively small, sedentary, sin-
and Vierich 1989). Formal storage is the placement gle-component habitation camp located in a semi-
of objects in facilities used specifically for stor- arid region where there is no constraint on space.
age-e.g., storage platforms, pits, bowls or jars, Likewise, the Kutse sites examined are sedentary,
and boxes. They are called formal storage areas single-component habitation camps occupied by
because they consistently are used to store tools or one or more nuclear families located in a semi-arid
other material culture, whereas informal storage region where there is no constraint on space (as
areas, such as a tree, may store an object temporar- there may be in densely forested areas). In contrast,
ily but are not specific in function for storage. large towns and cities may have different patterns
Formal storage loci are almost exclusively found in of refuse disposal as might sites located in other
semi- to completely sedentary societies and are areas of constrained space such as rockshelters
only rarely found in highly mobile groups (Kent (e.g., Mayan centers; Deal 1985,1987). Such sites
and Vierich 1989; Murray 1980). Because mobility are probably not appropriate to apply the ethnoar-
has an important influence on storage practices, I chaeological research discussed here-targeted
chose sedentary groups to study both formal and studies appropriate for urban and other areas need
informal storage areas. to be conducted first.
Since 1987, 1 have spent part of each year study- The two questions addressed below are:
ing storage practices at Kutse, Botswana, a recently * Is the material culture from trash areas and stor-
sedentary community located just south of the age areas consistently different in ethnographic
Central Kalahari Game Reserve. Kutse residents sites where the function of a locus is known?
are not agriculturalists and, hence, do not have spe- * Are the same differences visible in the archaeo-
cialized food storage facilities for the storage of logical record?
grain or other agricultural produce. Complete fea-
Observations of Storage and Trash in an
ture and surface object inventories were compiled
Ethnographic Context
for each Kutse camp, some during site use, some
after abandonment, and others both during and Kutse residents, who are Central Kalahari hunter-
after occupation. All surface objects, regardless of gatherers, occupy camps from two months to two
size, located on features were recorded (features or three years (descriptions of the people and their
This content downloaded from 132.248.9.8 on Tue, 09 May 2017 14:31:12 UTC
All use subject to http://about.jstor.org/terms
REPORTS 81
community can be found in Kent 1989, from any single piece of blanket. The latter is anal-
1993a,1993b,1993c, 1995a, 1995b, 1996a,1996b). ogous to the way in which sherds are counted
Excluding visitors, camps are usually inhabited by (noted individually or by weight, depending on the
five or more individuals (range 1 to 25). G/wi, analysis).
Gllana, and a few Kua Basarwa (Bushman or San) It should be pointed out that there may still be
and Bakgalagadi Bantu-speakers live at Kutse. some questions concerning the particulars of this
Interviews with adult residents, including former classification. However, in all cases, there was con-
occupants of abandoned camps, covered a variety sistency in how objects were classified. The same
of topics, including how long they planned to classification system was used at all Kutse camps,
occupy a camp and why. Intensive participant- abandoned or occupied, and the same person,
observation fieldwork and time allocation data also myself, categorized each item to enhance consis-
provide information on the use of space and storage tency. Because the focus of this article is not on the
facilities. classification of objects, only a brief summary of
Camps at which inhabitants plan a moderate-to- methods is presented here. The data must be valid
long occupation have more formal middens than for ethnographic and archaeological test cases, and
sites where short occupations are planned. Sheet transcend hunter-gatherer or agriculturalist eco-
trash (a thin distribution of trash throughout a nomic orientation, time period (contemporary or
camp) is most common at camps at which inhabi- prehistoric), group ethnicity (Basarwa,
tants anticipate a short occupation. Refuse areas at Bakgalagadi, or Anasazi), geographical location,
semi-sedentary and sedentary Kutse camps also and object material type (whether Western materi-
serve as ash dumps where accumulations of ashes als or aboriginal). The measure selected was the
from hearths are disposed. Refuse/ash areas are diversity of artifact categories. Although some
usually scattered in and around the periphery of the archaeologists may not agree that the diversity of
camp. It is possible to distinguish between formal categories at various features is the best measure-
storage areas and formal trash areas based on arti- ment to distinguish storage from trash areas, it best
fact content at the Kutse sites. The data below show suits this analysis.
that a relatively lower number of artifact categories Many investigators divide the concept of diver-
is consistently associated with trash areas, regard- sity into two dimensions: richness and evenness.
less of the number of objects. This creates less het- "Richness is the number of different nominal
erogeneous artifact assemblages at trash areas than classes of items observed in a sample regardless of
occur at storage areas. their individual frequencies. Evenness is the rela-
tive proportion of representation of each of the
Difficulties in Classification
classes present. The combined effect of richness
There is no such thing as a "perfect" classification and evenness usually corresponds to most people's
system. How one decides which object attributes intuitive idea of diversity" (McCartney and Glass
and categories to use depends on the reason for 1990:522; original emphasis).1 In this study I
grouping the objects in the first place. Because the employ statistical techniques that combine both
intention was to compare the Kutse material culture richness and evenness to obtain a measure of diver-
with artifacts found elsewhere in an archaeological sity.
context, the categories used were based on object
Difficulties in Measurement
material, form, and function. For example, a tin can
regularly used as a cooking pot was placed in a dif- I first used a simple descriptive statistic to deter-
ferent category than a smashed can used as a pot mine if there were any patterning present in the
holder. Cast iron pots used for cooking were placed data that would warrant a more sophisticated analy-
in a third distinct category. Arrows were classified sis. I compared the mean number of objects per cat-
as different from spears or other hunting equip- egory per locus. The number of objects in a feature
ment, including snares, digging sticks, and bows. divided by the number of categories at the feature,
Consistent with classification systems used in resulted in a mean number of objects per category
archaeology, blanket threads were coded individu- for each feature. This simple statistic is problematic
ally when completely unraveled and spatially apart for a variety of reasons, and I did not include fea-
This content downloaded from 132.248.9.8 on Tue, 09 May 2017 14:31:12 UTC
All use subject to http://about.jstor.org/terms
82 AMERICAN ANTIQUITY [Vol. 64, No. 1, 1999]
that although kraals and windbreaks each have a Formal Storage Area 198 0.5244
Goat Corral 867 0.7348
Table 1. Objects Located by Locus at Occupied and Non- Informal Storage Area 1,030 0.5754
Occupied Camps at Kutse. Features with Less Than a Total TOTAL NUMBER OF OBJECTS 27,391
of 100 Objects from All Kutse Camps Were
...............................................................................
This content downloaded from 132.248.9.8 on Tue, 09 May 2017 14:31:12 UTC
All use subject to http://about.jstor.org/terms
REPORTS 83
51.8
50-
40
m
0
0'- 30-
z 20 1.
z 1.
11.5 11.0 99
10- 7.3
5.3
3.4
0
Refuse/ Hut Kraal Wind- Informal Roasting Hearth Storage
Ash Areas break Storage Pit Platform
Area
CAMP LOCI
Figure 1. Bar diagram of mean number of objects per category at Kutse camps, Botswana.
ing the number of categories present at different Since I wanted to compare the ethnoarchaeolog-
loci. ical data to archaeological data, I thought it prudent
These patterns appear to be the result of behav- to examine separately the average number of
ior. If they were not, one would expect the loci with objects per category from occupied and abandoned
the most objects to have the same number or more Kutse camps. Many of the objects recorded at cur-
categories than ones with fewer objects. The num- rently occupied camps were perishable materials
ber of rare categories should increase with the size that would not likely be left or be archaeologically
of the assemblage. However, that is not the pattern visible after camp abandonment (Schiffer 1987). I
found here. The mean is the same at different loci inventoried 24 abandoned camps, which had a total
with very different numbers of objects. Table 1 of 84 features. Camps had been abandoned from
shows that ash areas have more than double the one to seven years. Most features were still visible,
next largest mean number of objects per category but extremely fragile objects did not preserve.
figure (Figure 1). Whereas refuse/ash areas and Table 2 shows that ash/trash areas at abandoned
huts have roughly the same number of objects camps, like those at still occupied camps, have the
(7,299 for the trash areas and 7,063 for huts), they greatest (over double) mean number of objects per
have significantly different numbers of categories category (Figure 3). To determine whether the
(141 categories for trash areas and 416 categories mean number of objects per locus was significantly
for huts). The result is very different mean objects different, t-tests were calculated. The mean of the
per category values (Figure 1). The absolute num- non-trash areas is significantly different from the
ber of categories decreases at some loci where mean of refuse/ash areas (i.e., p is less than 0O.05).2
object number increases. This observation cannot To facilitate comparisons with a single archaeo-
be attributed simply to differences in the number of
logical site, the data were separated by camp, locus,
objects across loci. Loci with roughly similar num- and type of object. I wanted to determine whether
bers of objects have very dissimilar mean numbers the patterns delineated from a group of sites (i.e.,
of objects, again indicating disposal behavior is aggregated data) were still visible when examining
responsible for this patterning. a single site by itself. The mean number of objects
This content downloaded from 132.248.9.8 on Tue, 09 May 2017 14:31:12 UTC
All use subject to http://about.jstor.org/terms
84 AMERICAN ANTIQUITY [Vol. 64, No. 1, 1999]
43.7
40
C,,
0 30
LL
0
= 20-
z
z
11.8 11.2
10- 8.5
5.7 5.4
2.0 1.
01
Figure 2. Bar diagram of mean number of objects per category at only abandoned camps at Kutse, Botswana.
0.8-
0.7-
0.6-
w
=) 0.5l
C,,
w
0.4-
I-
cc 0.3-
w
00.2-
0.1
Figure 3. Bar diagram of the Teachman statistic measure of diversity only at occupied camps at Kutse,
This content downloaded from 132.248.9.8 on Tue, 09 May 2017 14:31:12 UTC
All use subject to http://about.jstor.org/terms
REPORTS 85
per category was calculated for each locus at all probability of observing any category would be 1/1.
camps. Even on the single camp level, ash/trash If there was no difference in the diversity of objects
areas have the largest mean number of objects per across loci, any category would have the same
category at 88 percent of these sites. The few probability of being observed in any locus, which
camps where ash areas do not contain the greatest would, in turn, produce the same index.
number of objects/category tend to have a small In addition, the Teachman's statistic has two
total number of artifacts located at one or more of practical advantages over the Shannon's
the features. This pattern may be due to the fact that Information Statistic. First, it is normalized against
the loci all have less than 100 objects and often less the theoretical maximum wherein all categories are
than 50. observed and objects are evenly distributed among
all categories. Therefore, it generates a value of 1
The Teachman Information Statistical Measure of
when objects are evenly distributed, indicating
Diversity
maximum diversity, and a value of 0 when all
To further evaluate the material culture patterning, objects are concentrated in one single category,
I employed a modification of the information theo- indicating minimum diversity. Second, the
retic statistic adapted from Teachman Teachman statistic always generates a positive
(1980:340-344). Information theoretic statistics value, with a larger value indicating greater diver-
are not completely without problems, as docu- sity. This makes the interpretation of the statistic
mented by a number of archaeologists (e.g., Jones more practical and intuitively sensible. Thus,
and Leonard 1989; Leonard and Jones 1989). Most whether a category is observed in a locus is not
of these archaeologists correctly criticize statistics, related to sample size but reflects "something else"
such as the Shannon statistic of diversity, for being that is influencing the distribution of categories.
3
influenced by sample size differences. Even so, The Teachman statistic measures diversity and
none of the alternatives suggested thus far are with-
was applied to objects located at recently aban-
out their own problems, some with difficulties in doned Kutse camps using data recorded in different
assumptions or use that are equal to or greater than field seasons from 1988 to 1992. The Teachman
those of the Shannon and other diversity measures. statistic of diversity (Teachman 1980:344) is:
Because there is no perfect measure, I looked at
DM = A pi (log (1/pi))
two different ones: mean number of objects per cat-
egory which seems to be correlated with the num-
ber of different types of activities performed at a where DM is Diversity Measure and p is the
locus, and the Teachman diversity statistic probability of objects within a category 4.
described below. Both statistics reinforce the other,
To have the DM vary between 0, indicating
demonstrating that lower means per category and minimum diversity, and 1, meaning maximum
higher diversity values identify storage and similar diversity, it is necessary to standardize or normalize
non-trash areas. DM. The following equation was used:
The particular statistic used here, which I refer
DM' = DM / log (I)
to as the "Teachman" statistic, is not subject to the
sampling size biases that Leonard and Jones (1989) where I is the theoretical maximum or the total
point out as a major limitation of the Shannon number of possible categories and log (I) is the the-
Information Statistic. First, the Teachman statistic oretical maximum value of DM. By using the
is based on a probability distribution, which is not probability of observing each category instead of
affected by sample size. Second, since all activity the absolute number of objects in each category, the
areas were inventoried at a site, what I observed Teachman diversity statistic takes into account the
and consequently used in the analysis is the uni- different number of objects located at each area,
verse of all objects from all categories at all fea- thereby controlling for different numbers of objects
tures at camps-not samples from specific loci. per locus. A full description of the statistic and its
Therefore, the number of categories recorded con- application is in Teachman (1980). The statistic
stitutes the theoretical maximum, designated as I. was used to calculate a diversity measure at loci
In theory, if objects were evenly distributed, the that have 50 or more objects per camp. The major-
This content downloaded from 132.248.9.8 on Tue, 09 May 2017 14:31:12 UTC
All use subject to http://about.jstor.org/terms
86 AMERICAN ANTIQUITY [Vol. 64, No. 1, 1999]
0.7-
LU0.6-
uw 0.5-
4 0.4-
I-0.3-
> 0.2-
0.1
0-
In-between Windbreak Hut Ash/Trash Formal Goat Informal
Features Area Storage Corral Storage
Area Area
Figure 4. Bar diagram of the Teachman statistic measure of diversity at all camps located at Kutse, Botswana.
ity of the huts and windbreaks have a diversity fig- trash/ash areas that it was noticeable while in the
ure around 0.400 whereas most ash/trash areas field conducting the camp inventories. One reason
have a diversity figure of about 0.340. for the difference among areas may be that storage
Table 3 shows that areas of known trash accu- areas are often located near activity areas of various
mulations, such as in-between features (i.e., "sheet types. For example, at Kutse, storage platforms were
trash") have the lowest diversity measure value by built to store items, but people also sat under the plat-
far. They are followed by ash/trash loci. It is evi- form in the shade to perform various activities. In
dent, then, that trash areas consistently have the huts, the back area opposite the door was a formal
lowest values of all the loci regardless of whether storage area but sleeping, eating, and other activities
all the camps are examined together, only the occu- occurred nearby. Informal storage areas, by defini-
pied camps are compared, or only the abandoned tion, are areas at which storage is not the sole activ-
camps are analyzed by themselves. Consistently, ity. Examining the archaeological record alone, are
refuse areas have the least amount of diversity trash areas recognizably different from storage in the
while all other loci, from huts to windbreaks to same way that they are ethnographically?
informal and formal storage areas, have a higher
Archaeological Observations of
value (Figure 4). Note the agreement between the
Storage and Trash
values for the diversity measure and the mean num-
ber of objects per category figure at, for example, A cross-cultural examination of societies suggests
ash/trash areas and huts at all camps or at various that the pattern observed at Kutse is not tied to a
features and ash/trash areas at abandoned camps particular culture (Kent 1990). The study reveals
alone (Figures 2 and 5). similarities or regularities in storage patterns that
Regardless of whether the Teachman diversity suggest the possibility that the findings may be
statistic or various simple descriptive statistics are generalized to different societies. Are the
used, refuse/ash areas are significantly different fromTeachman measurement of diversity and the mean
storage and other activity areas. In fact, the differ- number of objects per category visible and corre-
ence from other loci is such a strong attribute of spond in the same direction in archaeological data
This content downloaded from 132.248.9.8 on Tue, 09 May 2017 14:31:12 UTC
All use subject to http://about.jstor.org/terms
REPORTS 87
0.5-
'~0.4-
w
C,,
*'< 0.3-
w
C,,
cc 0.2-
w
0.1
0
In-between Windbreak Hut Ash/Trash Formal Goat Informal
Features Area Storage Corral Storage
Area Area
CAMP LOCI
Figure 5. Bar diagram of the Teachman statistic measure of diversity only at abandoned camps located at Kutse, Botswana
Note: Not enough objects were recorded from formal storage areas in the abandoned camps; therefore, no bar is plotted on this fig-
as that documented at Kutse? To answer this ques- were not included in this study. The context of the
tion, I applied the ethnoarchaeological findings to artifacts from these randomly selected units is
Gnatsville, a small, single component, Mesa Verde ambiguous. Some probability units may represent
Anasazi mid-Pueblo II habitation located in south- portions of formal storage areas that were unde-
western Colorado (Kent 1992). Site configuration, tected since the entire horizontal area was not exca-
artifact inventory, and object distribution all indi- vated. At the same time, other probability units
cate that the single component site was occupied likely represented sheet trash, complicating the
year-round between circa A.D. 1030-1050 (for a interpretation of their use and their reliability for
description of the artifacts, stratigraphy, dates, testing
type the applicability of the diversity and mean
of sedentism represented, and site structure, see objects/category statistics in the archaeological
Kent 1991, 1992). record.
With the exception of formal storage areas such A number of formal storage loci were identified
as storage pits, storage had to be inferred as more at Gnatsville. These included a very large bell-
likely within the rubble mound than in the midden shaped pit and a slab-lined storage cist. Consistent
(as huts at Kutse were the locus of formal storage). with the ethnoarchaeological data presented below,
Material culture from five areas at Gnatsville was these storage loci were not found at other nearby
analyzed. These areas were the rubble mound sites of comparable date that were occupied for a
(which contained several rooms); kiva; midden; a shorter duration (Kent 1992). The reason I selected
secondary, more temporary midden with a hearth Gnatsville to test the archaeological consistency of
located stratigraphically above it (Area A); and ran- the ethnoarchaeologically derived observations was
domly placed units excavated outside these features because the activity areas were recognizable, dis-
(see Kent 1991:Figures 2, 3, 5). Since probability crete, and general function was identifiable. The
units were areas not associated with a feature, and site analysis can be seen as an independent test of
since I wanted data comparable with that collected the methods.
at Kutse, Gnatsville probability sampling units At Gnatsville, we can assume that storage areas
This content downloaded from 132.248.9.8 on Tue, 09 May 2017 14:31:12 UTC
All use subject to http://about.jstor.org/terms
88 AMERICAN ANTIQUITY [Vol. 64, No. 1, 1999]
all part of the Dolores Archaeological Project to All Objects and All Categories (Lithics, Faunal Remains,
further ensure comparability with other sites in the and Ceramics)
This content downloaded from 132.248.9.8 on Tue, 09 May 2017 14:31:12 UTC
All use subject to http://about.jstor.org/terms
REPORTS 89
22.8
Cn
-U 20-
0
U-
0
z
z 10-
w
3.9
1.1 0.7
SITE LOCI
Figure 6. Bar diagram of mean number of objects per category by features at Gnatsville, Colorado, USA.
density of objects, except for ceramics, at the lower from the surface and was located southwest of the
levels below a hearth was higher at Area A than any- rubble mound and kiva. The area contained fill that
where else at the site, with the exception of the for- was found stratigraphically below a hearth located
mal midden (Tables 4-5). Area A was not visible at the margin of the feature's midden deposit, about
0.4-
w
w 0.3-
CD
~-0.2-
0.1 1
SITE LOCI
Figure 7. Bar diagram of the Teachman statistic measure of diversity at features located at Gnatsville, Colorado, USA.
This content downloaded from 132.248.9.8 on Tue, 09 May 2017 14:31:12 UTC
All use subject to http://about.jstor.org/terms
90 AMERICAN ANTIQUITY [Vol. 64, No. 1, 1999]
Table 5. A. Objects Per Cubic Meter of Excavated Soil Per Locus at Gnatsville, A Small Pueblo II Anasazi Site (Surface
Artifacts Not Included). B. All Objects at a Feature Divided by the Number of Categories Located at a Feature Per Cubic
Meter of Excavated Soil.
B
Locus Objects/Cubic Meter/Categories
Combination of All Objects Divided by All Categories per Cubic Meter
Midden 22.8
Area A 3.9
Rubble mound 1.1
Kiva 0.7
2 m to the south of the kiva. The fill was very ashy pied for more than a few decades and that the locus
and contained the second highest concentration of was last used for non-refuse activities. That is, the
artifacts found at the site, rivaled in number only bymaterial culture patterning in the feature more
the large, formal midden. Because the clayey matrix closely reflects the last activities performed before
of the kiva floor was uncovered in a part of this fea- abandonment than the different activities initially
ture, it is suggested that Area A served as the initial performed at the locus (as also noted by Stevenson
midden for the site during the kiva construction 1985, 1991). Whether one examines each material
(Kent 1991:64). Study of the stratigraphic profile class separately (e.g., lithics or ceramics) or as a
and analysis of the cultural material and features at group, the formal midden consistently is more
Gnatsville supports the conclusion that Area A had repetitive in categories or has a lower category
originally been a small midden that was later con- diversity according to the Teachman statistic than
verted into a living area (see stratigraphic profile in that found at any other feature (Figures 5 and 6).
Kent 1991:Figure 5).
Sedentismn and Trash Areas
Because Area A had two distinct functions-ini-
tially a midden and later an activity area of some Trash at Basarwa camps at which inhabitants plan
kind with an associated hearth-the values for the to stay for a brief period, usually less than three
objects per category figure should fall between months, is scattered as sheet trash. Refuse is rarely
those for midden and non-midden areas. This is deposited at a specific locus. At semi-sedentary or
precisely what occurs. The Teachman statistic indi- recently sedentary communities such as Kutse, dis-
cates that Area A is intermediate in its category crete trash areas are located throughout a camp, as
diversity and resembles numbers from non-refuse long as occupants plan to stay more than a short
loci at Gnatsville (which actually was its last func- period of time. Several camps have an ash/trash
tion) more than figures from the midden (Figure 7). area adjacent to the central windbreak hearth at
The same conclusion is appropriate when view- which most activities are performed. It is interest-
ing the mean number of objects per feature. Even ing that I observed a Kutse family not use trash
when the amount of dirt excavated per feature is areas when they anticipated a brief stay of less than
taken into account, the number of all objects three months in the community-that is, most
refuse
together per category (and per cubic meter), Area A was sheet trash. However, the same family
is again indeterminate, but closer to the figure for did use discrete ash/trash areas when they planned
non-refuse areas (Table 5). This finding is perhaps to stay more than six months at a different Kutse
explicable given that the entire site was not occu- camp several years later. My research with other
This content downloaded from 132.248.9.8 on Tue, 09 May 2017 14:31:12 UTC
All use subject to http://about.jstor.org/terms
REPORTS 91
societies more accustomed to sedentism, such as In contrast, storage areas are often located near, if
the semi-sedentary to short-term sedentary Navajo not part of, areas with more than one activity. The
who today occupy camps for six months to less presence of more than one activity performed at
than five years, indicates that Navajos place mid- these other loci is reflected in the material culture
dens away from the main activity areas, although in terms of a lower mean number of objects per cat-
not necessarily on the periphery of the camp. Long- egory. More "categories" of behavior occur, result-
term, established sedentary societies, such as the ing in more categories of material culture present.
Highland Maya and other Mesoamerican groups, This results in a higher diversity value and lower
tend to place trash areas on the perimeter of a plaza mean number of objects per category. Research is
or site (Deal 1985). These trash areas are quantita- needed to determine whether a highly specialized
tively different from those found at Kutse (see activity area, such as a manufacturing locus for one
Hayden and Cannon 1983; Deal 1985; Arnold type of object, will mimic refuse areas.
1990). Informal storage, while probably always
Conclusion
present for anatomically modern humans is,
according to my research, probably not archaeo- While one may have expected pronounced differ-
logically visible. ences in the pattern of object diversity characteris-
Although not very common at the Kutse sites, tic of functionally different loci from such
abandoned parts of the community at many other dissimilar cultures and time periods as modern
longer-term sedentary villages or cities are used for Kutse residents of the Kalahari Desert, Botswana,
trash disposal. Not only has this been ethnographi- and Pueblo II Mesa Verde Anasazi from Colorado,
cally observed in various societies from and Navajo Indians from Utah (although I only
Mesoamerica to the North American Southwest have qualitative data for the latter), the pattern is
(e.g., Deal 1985), but also prehistorically inter- notably consistent. Such consistency suggests that
preted at a variety of archaeological sites. It is the pattern of low diversity at trash areas in com-
interesting, therefore, that the Teachman diversity parison to other loci at a site is both cross-culturally
statistic allowed us to interpret the ambiguous Area and cross-temporally valid and therefore poten-
A as a locus at which refuse was deposited and was tially useful to archaeologists in various regions
later reused as an activity area with a hearth. and time periods. Trash area artifact assemblages
The Teachman statistic is a relative measure that are more redundant in the object types represented
allows archaeologists to infer which loci were most than are storage areas which are more heteroge-
likely middens and which were more likely storage neous. Although this finding appears to be reliable
at a site. Rather than providing absolute numbers for various sedentary sites as those included here,
for classifying loci within or between sites, the the same pattern does not appear to be equally valid
areas with the lowest diversity tend to be, in most for sites occupied by nomadic peoples. Instead, at
cases, trash deposits in contrast to loci at the same nomadic camps occupied for a relatively brief
site with high diversity figures. I suggest that a period, refuse tends to be scattered throughout as
higher diversity of categories found at storage areas sheet trash. While observations conducted in the
in contrast to trash areas may occur because some Kalahari indicate that trash is not usually deposited
categories are deliberately not deposited at refuse at a separate locus when the occupation is antici-
areas. Categories of items that have religious/ritual, pated to be brief, further assessment is needed to
sociopolitical, and similar meaning tend not to be determine its cross-cultural validity and reliability.
discarded unless they lose their symbolism. Objects However, evaluating the material culture patterning
and/or raw material that are hard to procure may at sedentary sites show that storage and trash areas
similarly be curated and not often disposed of in a are recognizably different in the same ways at
midden. At the same time, I suggest that a larger archaeological sites as they are at modern seden-
number of different types of activities are per- tary occupations.
formed at or near storage areas. Trash areas tend to Diversity of activities leaves a diversity of mate-
be characterized by one activity, dumping trash, at rial culture, as observed through participant-obser-
any one time, assuming the function of the area vation fieldwork, inventories of occupied and
does not change through the life history of the site.abandoned camps, and interviews with former
This content downloaded from 132.248.9.8 on Tue, 09 May 2017 14:31:12 UTC
All use subject to http://about.jstor.org/terms
92 AMERICAN ANTIQUITY [Vol. 64, No. 1, 1999]
occupants about the use of loci at camps. Through Xi made the archaeological research described here possible,
and funding from the Swan Fund, Fulbright Foundation,
artifacts, this diversity is expressed at Kutse in the
Wenner-Gren, and Old Dominion University made the eth-
lower mean number of objects per category figure
noarchaeological research possible.
and also by a higher diversity statistic value at loci
where multiple activities occurred. The above
References Cited
analyses indicate that it is possible to differentiate
trash areas from storage areas in the archaeological Arnold, P.
1990 The Organization of Refuse Disposal and Ceramic
record. Because the distinction can be made, we
Production within Contemporary Mexican Houselots.
can study various facets of storage behavior using American Anthropologist 92:915-932.
age that is neither based on Western culture nor 1984 Dolores Archaeological Program Ceramic Dating:
Justification and Procedures. Technical Reports No.
specific to any one group or one time period. A
DAP-144. Dolores Archaeological Program, USDI
model of storage behavior is presented elsewhere Bureau of Reclamation, Engineering and Research Center,
(Kent 1998). The above is one attempt to distin- Denver.
Blinman, E., D. D. Wilson, R. Waterworth, M. Errickson, and
guish trash from non-trash loci at sites. It is the first
L. Hart
step that permits the second step of investigating 1984 Additive Technologies Group Laboratory Manual.
differential trash disposal and the sociopolitical Technical Reports No. DAP-149. Dolores Archaeological
Program, USDI Bureau of Reclamation, Engineering and
organizational changes of storage through time. Research Center, Denver.
Both steps are ultimately necessary to understand Cowgill, G.
how and why sites are organized. 1989 The Concept of Diversity in Archaeological Theory. In
Quantifying Diversity in Archaeology, edited by R.
This study does not provide the definitive state-
Leonard and G. T. Jones, pp.131-141. Cambridge
ment on the archaeological visibility of storage University Press, Cambridge.
which I have data. The analyses presented here are 1987 Household Pottery Storage and Site Structure: A View
from the Maya Highlands. Paper presented at 52nd Annual
examples of methods to differentiate loci by func- Meeting of the Society for American Archaeology,
tion and the resulting patterning of material culture. Toronto.
Dohm, K.
1988 The Household in Transition: Spatial Organization of
Acknovledgments. A preliminary draft was greatly improved
Early Anasazi Residential- Domestic Units, Southeastern
by valuable comments from Xiushi Yang, Michael Deal, Ben
Utah. Ph.D. dissertation, Washington State University.
Nelson, Martha Graham, Sarah Schlanger, Randy Gainey, University Microfilms, Ann Arbor.
Robert Leonard, and Stephen Plog. However, none of these Euler, R.
people should be held responsible for any inadequacies that 1988 Demography and Cultural Dynamics on the Colorado
may remain. I am grateful to Lynne Goldstein for allowing Plateaus. In The Anasazi in a Changing Environm2nent,
me to revise and resubmit this manuscript. I greatly appreci- edited by G. Gumerman, pp. 192-229. Cambridge
ate the following statisticians for their important help with University Press, Cambridge.
Gilman, P.
the manuscript: Xiushi Yang, Jay Teachman, and Dayanand
1983 Changing Architectural Forms in the Prehistoric
Naik. Each provided valuable suggestions concerning the
Southwest. Unpublished Ph.D. dissertation, University of
quantitative portion of the manuscript, ODU Graphics for
New Mexico, Albuquerque.
drafting, and Denise Elliott and Marion Blue provided edito- 1987 Architecture as Artifact: Pit Structures and Pueblos in
rial advice. the American Southwest. Amnerican Antiquity 52:538-564.
I am most grateful to the government of Botswana for Gould, R.
granting me a research permit to conduct the study which 1982 To Have and Have Not: The Ecology of Sharing
forms the basis for much of this paper. I especially thank the among Hunter-Gatherers. In Resource Managers: North
Office of the President, Botswana, Ministry of Local American and Australian Hunter-Gatherers, edited by N.
Williams and E. Hunn, pp. 69-91. Selected Symposium
Government and Lands, and the National Museum and Art
67. American Association for the Advancement of
Gallery of Botswana. I will always be indebted to the people
Science, Washington, D.C.
of Kutse who allowed me to live in their camps and share in
Gross, G. T.
a part of their lives. I thank the U.S. Bureau of Land 1987 Anasazi Storage Facilities in the Dolores Region,
Management for permission to excavate on BLM land. Colorado: A.D. 600-920. Ph.D. dissertation, Washington
Funding from the American Philosophical Society and Sigma State University. University Microfilms, Ann Arbor.
This content downloaded from 132.248.9.8 on Tue, 09 May 2017 14:31:12 UTC
All use subject to http://about.jstor.org/terms
REPORTS 93
This content downloaded from 132.248.9.8 on Tue, 09 May 2017 14:31:12 UTC
All use subject to http://about.jstor.org/terms
94 AMERICAN ANTIQUITY [Vol. 64, No. 1, 1999]
our knowledge of the past, rather than promote it. It may be population with the greater number of categories should have
possible that sample size and diversity are related in some the higher diversity. Conversely, combining two or more cat-
circumstances because of behavioral reasons, and not just egories should not increase the measured diversity
intrinsic quantitative reasons associated with the statistic (Teachman 19;342-343)
itself. 5My interpretation of Area A is that it served briefly as
4 Pi is defined as if N population elements are dividedmidden
into Iduring the initial habitation of the site while the kiva
arbitrary categories with the proportion of events falling in was being built. The orange-colored clayey soil for the pre-
the ith category is pi with pared kiva floor was first dumped on the small midden when
it was brought to camp. Most of the clayey soil was placed in
the kiva to construct the floor, but what was not used was left
in the formula described as the Teachman statistic (that is: covering the midden beneath it. After this time, the area was
not used as a specific activity area. During the occupation of
DM = EPi (log (1/pi))
the camp, Area A was made into a very different non-refuse
where DM is diversity measure and p is the probability of activity area with a hearth. Stratigraphically, it is possible to
objects within a category). So that if all events fall into one see that the people were not walking on exposed refuse when
category, the measure should take a value of zero indicating the hearth was built at Area A because the clayey soil capped
the refuse and because enough time had elapsed to have a
no variation, if all pi = 1/I, indicating that population events
are equally distributed among all categories, the measure thin layer of soil on top of the clayey soil (anywhere from a
should reach its theoretical maximum (log I). Additionally, it few years to a decade or two). The hearth was one of the last
is desirable to norm such a measure so that it equals 1 indi- features built at the site, which had an entire occupation
cating the highest degree of diversity and 0 indicating the spanning only around 20 to 30 years (see Kent 1991 for a
least diversity. If 1+1, 1+2, etc., exists, additional categories detailed and a stratigraphic profile of Area A).
to which no population members belong should leave the
measure unaffected. In other words, given two populations, Received March 18, 1996; Accepted Febtr-uaY 12, 1998;
with elements evenly distributed across all categories, the Revised Marc/ 4, 1998.
"A seminal work. It helps dispel the and his colleagues have woven together an
stereotypical view of life on the plains admirable tapestry that supersedes Wedel's
by examining the rich diversity of cul- seminal volume Prehistoric Man on the
tures and lifestyles that have existed in Great Plains." - R. BRUCE MCMILLAN,
the region. The excellent maps convey Director, Illinois State Museum.
a great deal of information." - LIBRARY 528 pages, illustrated. $29.95
JOURNAL. "Provides a much needed
of Plains archaeology."
SUSAN C. VEL-UK, Associate Pro- / -
fessor of Anthropology, Uni-
versity of Oklahoma. "Wood
3A
=_ S
This content downloaded from 132.248.9.8 on Tue, 09 May 2017 14:31:12 UTC
All use subject to http://about.jstor.org/terms