Interpretation in Context: Sampling and Analysis in Paleoethnobotany

Heidi A. Lennstrom; Christine A. Hastorf

American Antiquity, Vol. 60, No. 4. (Oct., 1995), pp. 701-721.

Stable URL:
http://links.jstor.org/sici?sici=0002-7316%28199510%2960%3A4%3C701%3AIICSAA%3E2.0.CO%3B2-W

American Antiquity is currently published by Society for American Archaeology.

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at
http://www.jstor.org/about/terms.html. 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/journals/sam.html.

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.

The JSTOR Archive is a trusted digital repository providing for long-term preservation and access to leading academic
journals and scholarly literature from around the world. The Archive is supported by libraries, scholarly societies, publishers,
and foundations. It is an initiative of JSTOR, a not-for-profit organization with a mission to help the scholarly community take
advantage of advances in technology. For more information regarding JSTOR, please contact [email protected].

http://www.jstor.org
Mon Oct 1 14:22:49 2007
 INTERPRETATION IN CONTEXT:

SAMPLING AND ANALYSIS IN PALEOETHNOBOTANY

Heidi A. Lennstrom and Christine A. Hastorf

During the past 20 years the collection ofsamples forjlotation and subsequent paleoethnobotanical analyses havejlourished.
At the same time, archaeologists have become increasingly concerned with both the physical and cultural contexts of mater-
ial remains. With this in mind, we must critical{v examine the sampling schemes used in the field and laboratoty This article
presents a method that can help archaeologists recover the most complete information needed to address both the physical
and cultural conte.xt ofarchaeobotanical remains. By comparingjlotation samples,from/eatures and adjacent locations, we
demonstrate the need for systematic, comprehensive sampling and analytical procedures to delineate the chronological,
stratigraphical, and cultural relationships between and among materials in adjacent areas.

La recoleccidn de muestras dejlotacion y consiguiente analisis paleobotanico han florecido en 10s ultimos 20 060s. A1 mismo
tiempo, 10s arqueologos se han tornado mas preocupados con 10s contextosfisico y cultural de 10s restos materiales. Con esto
en mente, necesitamos e.xaminar criticamente 10s esquemas de muestreo usados en el campo y en el laboratorio. Este articu-
lo presenta un mdtodo que ayuda a 10s arqueologos a recuperar inforrnacion completa necesaria para evaluar 10s contextos
fisico y cultural de 10s restos arqueologicos. A travds de la comparacion de muestras dejlotacidn tomadas en rasgos y local-
idades advacentes, demostramos que se necesita un procedimiento de muestreo sistematico, analitico, y comprensivo para
delinear las relaciones cronoldgicas, estratigraficas y culturales entre materiales provenientes de areas adyacentes.

he increasing use of flotation systems in more precise and contextually sensitive under-

T archaeological projects during the past 20

years has illustrated the important role of
plant remains and other small finds for under-
standing of the past (Case 1973; Hodder 1987:l;
Sharer and Ashmore 1979:561).
The new potential to recover vast quantities of
standing prehistory (French 1971; Struever 1968). material and an interest in more meaningful con-
Until quite recently, archaeobotanical materials textual interpretations, intrasite variability, and
were normally used to solve traditional questions the interpretation of feature-specific plant
concerning domestication and subsistence (e.g., remains necessitates the development of new
Hubbard 197.5; Renfrew 1973). But with the methods and a more sophisticated utilization of
advent oflarge-scale flotation, paleoethnobotanists existing techniques. This can be accomplished
have also been able to study the intrasite spatial with detailed collection schemes in the field and
variability of plant remains, allowing for the defi- careful analysis of specific samples in the labora-
nition of stratigraphy (Asch and Sidell 1988) and tory. We illustrate one simple method that can be
cultural contexts (Jones et al. 1986). In addition, especially useful in increasing the culturally and
archaeobotanical remains have also been used to historically meaningful knowledge recovered
address more complex issues of meaning and con- from different areas of a single site.
textual interpretations of plant distributions (e.g.,
Hastorf and Johannessen 1991; Lennstrom 1992; Sampling Archaeobotanical Remains
Wetterstrom 1978). In this paper we focus on sam- Sampling is necessary in every archaeological
pling strategies used to collect botanical materials. excavation. Typically, the issue centers on which
We suggest that the use of a comprehensive sam- approach to use. Some researchers apparently

..
pling strategy will allow archaeologists to obtain a avoid the question of botanical sampling by sug-
Heidi A. Lennstrom Department of Anthropology. Bernice Pauahi Bishop Museum. Honolulu, HI 9681 7
Christine A. Hastorf Department of Anthropology, University of California, Berkeley, CA 94720

Amer~canA n t ~ q u ~ t60(4).
y, 1995. pp 701-721
Copyr~ght by the Soc~etyfor Amer~canArchaeology
702 AMERICAN ANTIQUITY [Vol. 60, No. 4, 19951

gesting that the matrix from the entire site can be Further. it is evident that plants, like other materi-
processed by flotation (French 197 1 ; Jarman et al. al remains, are most informative when considered
1972). In most cases it is neither economically in combination with other archaeological data.
feasible nor necessary to process all site sediment This "feature bias" arises for a number of rea-
and analyze every flotation sample in order to sons. First. for some, broad issues and explorato-
obtain an accurate representation o f plant ry work materials from features may be sufficient.
remains throughout a site. The goal of maximiz- Paleoethnobotanists themselves often focus on
ing information while minimizing cost and labor features such as pits (e.g.. papers in Scarry 1993).
means that comprehensive sampling schemes This reinforces the idea that features produce the
must be planned prior to excavation. Yet, any only informative archaeobotanical materials. But
sampling scheme must be flexible enough to con- while features are a good source of information,
tend with unexpected events in the field (Adams they are not the only source of information. In
and Gasser 1980; Bohrer and Adams 1977: addition, the only way to evaluate archaeobotani-
Clarke 1989: Pearsall 1989; Toll 1988; van der cal contents of various features is to demonstrate
Veen and Fieller 1982). their independence from other archaeobotanical
The need for systematic recovery and analysis deposition across a site.
of materials has been stated time and again, and The need for more rigorous methods has not
the regular recovery of plant remains continues to gone unnoticed by paleoethnobotanists. Pearsall
improve. Unfortunately, with all the best inten- (1989:95) recently called for comprehensive sam-
tions, archaeologists sometimes continue to bias pling strategies at all archaeological excavations.
the recovery and interpretation of materials from She recommends a "blanket sampling" strategy in
flotation samples. Without an understanding of which sediment for tlotation is collected from all
archaeobotanical materials and their potential, excavated contexts. If the researcher is interested
researchers may employ an unsystematic collec- in interpreting different features or contexts with-
tion strategy. Too often archaeologists focus their in an archaeological site. this strategy is crucial. If
sampling efforts on features defined during exca- all deposits are not sampled systematically for
vation and areas where carbon can be seen or is plant materials it is difficult to formulate valid
expected. Whereas we could find no published interpretations about the botanical contents of any
archaeological manuals or papers that actually given provenience. This archaeobotanical sam-
call for this type of biased sampling. it is preva- pling problem has also been known for more than
lent nonetheless. We, like many other paleoeth- 20 years. Indeed Spector noted (1970: 172) that
nobotanists, have often been asked to be part of some conclusions concerning the botanical con-
ongoing archaeological projects and are subse- tents of pits lack credibility if archaeologists d o
quently told that "there are flotation samples from not discuss the plant remains from other site areas
site features" or "we collected these tlotation outside these features. Again. if the goal of the
samples because we saw plant remains in the research is more than a broad generalization
deposits." This scenario stems from the basic mis- about some of the plants deposited within a site.
conception that the only goal in paleoethnobotany other site areas to be sampled should include both
is to collect great amounts of plant remains. It cultural and noncultural deposits such as roof fall,
must be kept in mind that knowing where inwashed sediment. fill, and areas off-site.
archaeobotanical remains do not occur is just as In general. the impact of comprehensive sam-
important as knowing where they are found. pling schemes on paleoethnobotanical interpreta-
Additionally. the types of features typically sam- tion has not been addressed with empirical evi-
pled for plant remains (e.g., hearths and pits) do dence. One notable exception is the work of
not always produce the greatest amounts of car- Ciasser ( 1985). Using flotation samples from two
bon (Pearsall 1989). In this era. when an increas- Hohokam sites, he compared the botanical con-
ingly strong emphasis is placed on knowledge of tents of features with that of postoccupational fill
precise archaeological contexts and relationships, from abandoned sites. He demonstrated that the
it is imperative to obtain an accurate picture of the botanical con~positionof floor features is nearly
patterns of botanical deposition in space and time. identical to that of postoccupational fill.
 REPORTS 703

Displayed in pie charts, Gasser's samples have quantification schemes commonly used in pale-
similar percentages of material-usually within oethnobotany: percentages and densities.
two percent of one another. Conversely, the densi-
Percentage Di.vtrihlrfion
ties of plant materials show important differences,
with a higher density of materials in the floor Comparisons by plant percentages are expressed
deposits than in the overlying fill. From these dif- in pie charts. Each pie represents a single locus.
ferences Gasser concludes that plant remains have At the site under discussion we used the term
moved from one area to another. Possible causes "locus" to refer to the smallest cultural or natural
for the displacement of plant remains and other event recognized in the deposits (Harris 1979).
artifacts include trampling and mixing by site res- From these diagrams the relative percentages of
idents. animal activity. gravity, and ground water; all taxa can be compared and contrasted with
movement can occur vertically as well as laterally those of the deposits above (later) and below (ear-
(Hughes and Lampert 1977; Matthews 1963; lier). This format standardizes each locus so that
Stein 1983; Stockton 1973). Although the move- samples of different botanical densities can be
ment of materials between adjacent areas of sites compared directly (Renfrew 1973). We are fully
needs further study, Gasser's example is evidence aware of the problem of interdependency of per-
that interpretations based on the botanical con- centage values, i.e.. the effect of one taxon's abun-
tents of features or tloors may be incorrect with- dance on the percentage score of another. But we
out proper sampling and analysis of adjacent are confident this problem is minimal in this data
deposits. set because the number of different taxa per locus
A second important study addressing the issue is high (mean = 12.5; median = 11.5).
of contextual interpretation of plant remains was We present the percentage diagrams as three
conducted by Jones et al. (1986). In their study of pie charts in a vertical column. The overlying
samples from the site of Assiros, Greece. the deposit is at the top of the column, the feature is
authors demonstrated that an intensive sampling in the center, and the underlying matrix is at the
procedure could indicate not only the functions of base. All selected features were examined in this
structures but also the methods of grain storage. format, and 17 representative examples are pre-
Further, they were able to determine which type sented in this work (see appendix). In our com-
of grain was stored in different locations. This parison we predict that when flotation samples
detailed, comparative strategy across the site from features and surrounding areas yield the
enabled Jones et al. to describe additional aspects same taxa in similar percentages, there may be a
of life in prehistoric Greece. problem of contamination (from one to another).
In an effort to emphasize the importance of Alternatively, materials in all locations may
complete sampling of archaeological deposits for derive from the same depositional source.
contextual understanding, we expand on the work Significantly different percentages likely indicate
initiated by Jones et al. and Gasser. To do so we independent sources of plant remains.
will use a series of selected features from our Quantifications of botanical material are made
excavations at a site in the central Andes of Peru. as counts, rather than weight, for two reasons.
The features selected are those often chosen by First. many of the plant remains from this site are
archaeologists for botanical sampling, and small seeds of negligible weight. These taxa are
include a series of well-defined hearths, pits, ani- "lost" when weights were used to display the sam-
mal offerings, and human burials. We contrast the ples. Second, whereas it is true that plant parts are
botanical contents of these archaeological fea- differentially preserved and fragmented, most
tures with their surrounding deposits, including materials from this site are small (.5-10 mm) and
the materials above (as did Gasser). below, or cut there is little variation in the fragment sizes across
by the features. the site or through time. That is, the degree of frag-
mentation for each taxon is similar in most con-
Methods for Interpretation texts and should not greatly distort the relation-
To contrast the botanical contents of features and ships between the taxa. Therefore, we conclude
their neighboring deposits we use two different that using counts is the clearest way to present our
704 AMERICAN ANTIQUITY [Vol. 60, No. 4, 19951

botanical data to illustrate the need for compre- parisons cannot replace more rigorous statistical
hensive sampling. methods, they are a quick and efficient manner to
Abbreviations of the plant types depicted in display information and may indicate to the
the pie charts can be matched to those in the researcher which features and companion
appendix. The category "charfrag" represents deposits warrant further detailed quantitative
undifferentiated burned parenchyma. most likely investigation (Chambers et al. 1983; Tufte 1983;
from Andean tubers (such as the potato) that have Tukey 1977).
a tendency to break into unidentifiable pieces.
The size and morphological characteristics of the An Archaeological Example from Peru
"charfrags" are consistent throughout the deposits To investigate the interpretation of archaeological
and there is little danger in assuming they are features and their associated deposits we utilize a
derived from the same plant types. The category large data set from the site of Pancin. This site is
"other" consists of all taxa that make up less than located in the highlands of central Peru (Hastorf
five percent (each) of the total plant specimens. et al. 1989; Lennstrom 1992) and represents the
The actual composition of the 5 1 loci displayed in remains of an agricultural village that was part of
the pie charts can be seen in the appendix. a small-scale Andean community in the upper
Mantaro River valley (Hastorf 1993). During the
Densities 1986 field season we uncovered the remains of
A second quantification method-density-is four occupation periods dating between A.D. 650
used in tandem with the pie diagrams to help and 1000. Three of the occupation areas were
assess the meaning of similarities and differences characterized by a single walled patio compound
between features and their surrounding deposits. with associated circular structures, whereas the
In our example densities are expressed as the fourth was a poorly defined area of incinerated
number of plant fragments per liter of excavated refuse and midden deposits. One patio compound
site matrix (Popper 1988; Miller 1988). This was dominated by the presence of a large (ca. 5 m
quantification strategy standardizes taxa counts in diameter) structure. which had burned while
across samples of different original size. A com- full of harvested crops. These general features and
parison of different densities of plant material more detailed information about individual
may also help determine the origins of botanical deposits, in concert with specific spatial informa-
materials and the problems of plant movement tion, will show the interpretive value of blanket
within the site. sampling for plant remains.
We explored non-parametric statistical analy- The 1986 field season at Pancan was planned
ses as a means of measuring the similarities and with methodological issues in mind. All excava-
differences between features and their surround- tion teams were instructed to collect at least two
ing deposits, but this did not produce useful six-liter sediment samples for tlotation from every
results. Both Spearman's Rank Order Correlation type of cultural or natural deposit encountered
Coefficient (Freund 1979) and Jaccard's during the excavation (where possible). In this way
Coefficient of Similarity (Sneath and Sokol 1973) the sampling procedure was not biased toward fea-
were tried but neither was well suited to the dis- tures or areas where ash and carbon were visible.
tribution of the botanical data. Most often, each All samples were collected systematically regard-
loci contains five to 10 taxa that are represented less of preconceived notions of which features
by numerous specimens each, in addition to a were most important or interesting.
number of types represented by a single specimen Paleoethnobotanists, as well as zooarchaeolo-
(see appendix). The result, when using either of gists and others, have noted the difficulty in
the statistical routines noted above, is that the taxa assessing whether or not the patterns of archaeo-
represented in small amounts tend to mask the logical remains are merely reflections of sample
similarities and differences between the main size (Allen 1989; Cruz-Uribe 1988; Grayson
components of the samples (see Grayson 1984; Pearsall 1983). Acknowledging such poten-
1984:106).Additionally, even though visual com- tial problems, Hastorf instituted a sampling proto-
 REPORTS 705

col designed to reduce biases produced by sam- Pancan, we focus this presentation on some com-
ples that are too small or by variation in sample mon archaeological features often thought to be
size. Whenever possible, individual flotation sam- intentionally created and filled. These are hearths
ples were a standard size of six liters. Two of these (n = 1I), pits (n = 45), animal offerings (n = lo),
standard-sized samples were collected if the locus and human burials (n = 12).
was sufficiently large. Using this Pancan data set, Due to edaphic conditions at the site we con-
an analysis of three "diversity" indexes (two for sider only charred plant remains to be prehistoric.
richness and one for evenness) demonstrated that Charring occurred due to purposeful burning for
sample size was not a factor if there were more fuel, accidental burning during cooking, discard-
than 25 charred fragments per six liters of soil ing and burning of refuse, accidental burning of
(see Lennstrom and Hastorf 1992. stores, and firing of ceremonial offerings.
To determine if the volume of matrix per locus Uncharred plant tissues are rarely preserved in an
had any influence on our results we also com- open site which is subject to frequent rains and
pared the volume of processed matrix to the num- changes in lake levels.
ber of different taxa per locus. Our results showed In an effort to understand the relationships
there is no significant correlation between these among the deposits at Pancan, we compared the
two variables. material in a number of ways. The most insightful
The outcome of our intensive sampling design level of investigation for these features and their
was the collection of 1,132 flotation samples. surrounding deposits is that conducted on a sim-
Each sample collected was processed with a ple one-on-one basis, whereby the botanical com-
mechanized water flotation system similar to ponent of each feature is directly compared to the
Watson's (1976) SMAP machine. The internal materials located in the deposits above and below
screen size of the machine allowed recovery of the locus. At first we compared entire feature
remains as small as .5 mm, with each sample groups (e.g., all hearths) against entire neighbor-
averaging 441 fragments of charred plant remains deposit groups, but there is as much variability
(Lennstrom 1992:90). within each feature type as there is between con-
Flotation samples were recovered from nearly texts, so we turned to the individual features.
every locus, and each locus was recorded on a Unfortunately, there are neither over-arching
separate data form. We were particularly interest- botanical patterns linked to specific context or
ed in listing all associated deposits that were in feature types, nor are there any plant taxa unique
physical contact with the locus. This allowed us to to certain types of features.
compare materials from neighboring contexts. This data set of 78 features and surrounding
This system is modeled on the Harris matrix deposits comprises more than 450 individual
(Harris 1979), which allows the relationships of flotation samples containing more than 70 plant
all loci from the site to be displayed chronologi- types. It is simply not possible to present the data
cally. At Pancan, more than 700 separate loci are in their entirety here. Fortunately, after compar-
defined. Recorded cultural contexts include fea- ing each of the features to their companion loci,
tures, e.g., floors, hearths, middens, pits, and buri- we discerned a limited number of patterns so it is
als, which are usually sampled during excava- not necessary for all examples to be illustrated
tions. We also collected materials from posthabi- here (for a more complete presentation, see
tational deposits, construction fill, adobe wall Lennstrom 1992).
slump, and roof fall, which some archaeologists
do not consider to be features and therefore do not Hearths
sample for botanical remains. The plant remains recovered from hearths are
deposited in a manner that differs from other fea-
Comparison of Features and Neighboring tures. Hearths are used repeatedly over a number
Deposits of years and the contents are purposefully burned.
Because it is not possible to discuss the relation- However, not all accumulated fill is specifically
ships among all the different loci recorded at intentional. That is to say, people rarely dig, fill,
 AMERICAN ANTIQUITY [Vol. 60, No. 4, 19951

Locus 657 Locus 270

Locus 709 Locus 292

Locus 746 Locus 301

A B
Figure 1. Hearth from Structure 16 with companion deposits (left) and hearth from Structure 7 with companion
deposits (right). See page 719 for a key to abbreviations.
and seal a hearth as they might a burial. We recov- The percentage diagrams of hearths display
ered discrete hearths from all periods at Pancan; distinct patterns. The 10 hearths with independent
they were located in houses, open-air patios, and botanical contents are characterized by two dis-
the area of generalized garbage incineration. tinguishing attributes. First, these hearths usually
Hearths differ from other pit features by display- contain higher proportions of identifiable plant
ing evidence for in situ burning such as fire-red- material than surrounding loci (compare Locus
dened earth and stones. 709 to Locus 746, below). Second, they often con-
The pie diagrams from 1 1 hearths reveal two tain a large percentage of a single type-such as
different patterns. Ten hearths had sample contents wood or seeds of Verbena, sedges, quinoa, or
that were noticeably distinct from the surrounding legumes (see Table 1 ). From these findings we
deposits. An example of this pattern is seen in suggest that plant material found in the Pancan
Hearth 709, located in the earliest level (Figure hearths is usually the in-situ remains of prehis-
la). Over 50 percent of the sample from Hearth toric activities. It also appears that wood, shrubs
709 is wood, whereas the wall fall above it con- (Lfe'prbetia and leguminous plants), and sedges
tains a high percentage of quinoa; the fill below it were used as fuel in this setting. The inclusion of
contains quinoa and many unidentifiable frag- quinoa may indicate that this food was cooked in
ments. Only one example, Hearth 292 (from the the hearths.
latest level), is somewhat similar to its companion Patterns in the density figures are less consis-
deposits (Figure Ib). In this case the contents of tent than the pie charts, but some trends seem
the hearth and its surrounding floor and occupa- apparent. In five of the 1 1 cases the hearth loci
tion zones consist of high proportions of quinoa, contain denser plant remains than either the con-
maize kernels and cobs, and unidentifiable text below or above. An additional four hearths
charred fragments. have denser remains than at least one contiguous
 REPORTS

Table 1. Density of Plant Remains from Hearths and Companion Deposits

Total Most Common % Total Density of Overlying Density of Underlying

Locus Density Plant Type Content Deposit and Context Deposit and Context
Hearth 83 21.0 Wood 44 5 1.3 occupation zone 9.7 prepared floor
Hearth 102 58.3 evhenu seeds 50 32.2 fill 46.2 midden wi wall slump
Hearth 107 15.2 Zeu mnvs cob fragments 40 200.6 midden with ash 154.8 midden with ash
Hearth 137 35.0 Cyperaceae (sedge) seedr 18 10.3 rock fill 36.6 occupation zone
Hearth 16911 72 34.8 Q u ~ n o aseeds 44 56.1 midden wi wall slump 33.0 midden
Hearth 175 65.2 W ~ l dLegume seeds 24 20.6 adobe wall fall 20.6 adobe wall fall
Hearth 276 135.9 bkrbenu seeds 37 46.8 occupation zone 17.8 prepared floor
Hearth 292 4.6 Zen mnla cob fragments 17 17.8 prepared floor 25.8 occupation zone
Hearth 432 16.9 Quinoa seeds 82 13.4 midden wi wall slump 7.2 prepared floor
Hearth 522 53.3 Quinoa seeds 12 16.8 fill 46.3 occupation zone wi wall fall
Hearth 709 38.5 Wood 61 60.8 adobe \\;all fall 2.8 subfloor construction fill
Note: Fragments per liter of processed site sediment.

deposit (Table I), suggesting that the remains of vary greatly, were typically coded in the field by
fuel and food are still in situ. Only two hearths fill type, e.g., midden, clay, ash, or gravel. Pits
contain sparser botanical remains than both com- occurred in all areas of the site: inside structures,
panion deposits. This is not surprising for Hearth in the patios, outside the walled compounds, and
107, located in a midden area characterized by in areas of burned refuse. Our comparison of the
dense burned botanical trash. The other hearth botanical content of the 45 pits to their surround-
with sparse remains is 292. This low density, in ing deposits revealed a number of trends; several
conjunction with the information derived from sources of possible misinterpretation are also
the pie diagrams, leads us to conclude that the apparent. To illustrate the range of variability,
original spent fuel in Hearth 292 may have been Table 2 and Figures 2 4 display data from seven
removed; the hearths were then likely abandoned pits and their companion loci. Unlike hearths, the
and subsequently filled by sediments from the majority of the Pancan pit features are not strik-
surrounding area. ingly different from their companion deposits in
Considering the percentages and densities percentages of plant taxa recovered. Instead most
together we find most of these hearths differ from pit contents appear to be similar to their sur-
those investigated by Gasser (1985). He found rounding deposits. This indicates that nearby
that the percentages of plant taxa in hearths were deposits may have been used as fill or may have
similar to those of overlying postoccupational fill. slumped into the depressions after the pits were
C:onversely, the hearths in our study appear to be cut. Yet, densities of plant remains inside four of
independent of surrounding loci. Density patterns the seven pits illustrated (Table 2) are greater than
from our hearths agree slightly better with in deposits above and below the pits. It is possible
Gasser's data. He consistently discovered that the that some plant remains were transported out of
density of material in hearths exceeded that of the the pits and contaminated the adjacent areas.
overburden; we saw this in 55 percent of the cases Alternatively, the preservation of fragile botanical
in our study (Table 1). The different patterns from remains may have been enhanced by their loca-
Pancan and Gasser's southwestern U.S. sites tion inside the pits.
strengthens the argument that there is no univer- In many cases, comparison of midden-filled
sal pattern for botanical deposition in hearths. pits and their neighboring deposits demonstrates
Archaeologists must recover samples from all that rubbish is not necessarily confined to the pits.
locations and compare contexts within the site. An example is Locus 369 (from Level 2), a typi-
cal midden-filled pit that is cut into a deposit
Pits
(Locus 377) with many of the same plant types
The pit features at Pancan range in size from 1 to and a similar density of botanical remains (Figure
160 liters in volume, with the majority falling 2a, Table 2). This pattern of similarity is not unex-
between 10 and 25 liters. The contents, which pected for Pancan because 12 of the 15 pits filled
 AMERICAN ANTIQUITY [Vol. 60, No. 4, 19951

Locus 333 Locus 582

Locus 369 Locus 593

Charf rag

Locus 377 Locus 584

Charfrag
CharfragJ

Figure 2. Midden-filled pit in Level 2 with companion deposits (left) and pit from burned storage structure with com-
panion deposits (right). See page 719 for a key to abbreviations.
with midden are in the area of the site used exten- of the store house; the ultimate origin of the food
sively for dumping and burning trash. remains may have been the pits, whose contents
The similarity we find between pits and their became scattered during or after the fire. Of all
surrounding deposits is not limited to areas of the pits excavated at Pancan, these two contexts-
midden. Figure 2b shows a pit from a food-stor- midden and burned storage-are where nondis-
age area of a burned structure in Lwel 3 . On its tinct pit contents are most common.
own, this feature (Locus 593) would probably One-to-one comparisons of neighboring
have been interpreted as having special, intention- deposits can also help isolate those pits that do
ally deposited contents. given the high concentra- have unique contents. In such cases our conclu-
tion of maize and quinoa within the pit. However, sions based on pit contents are strengthened
when the pit is compared to its neighboring because we can demonstrate that the plants are not
matrix samples above and below, it is obvious that from the same source nor are they contaminants
all plant remains in the area are similar and the from nearby loci. Examples of this include Pit 735
materials in the pit are merely an expression of a from inside a structure and Pit 309 from an open
burned storage area within the structure. The work area (Figures 3a and 3b). These two features
higher density of plant remains within the feature are unlike their companion deposits in that they
again may suggest that the contents of pits are to contain high proportions of domesticated legumes
some degree protected from degradation. but this and quinoa, respectively. Only by comparing these
protection does not mean the botanical contents features to companion deposits can one demon-
of Pit 593 are of a different origin than the mate- strate that the patterns are indeed unique.
rials in the surrounding matrix. It is probable that In still other cases pit contents considered on
all the materials in this area relate to the burning their own may not appear unique until they are
 REPORTS

Locus 614 Locus 300

Locus 735 Locus 309

V Dom legume

Locus 343

Figure 3. Pit in Structure 17 with companion deposits (11:ft) and pit with hoes in Level 2 with companion deposits
(right). See page 719 for a key to abbreviations.
compared to deposits above and below. An exam- assumed that all remains-including plants-were
ple is Pit 438 that contains mostly unidentifiable purposefully incorporated. Capstones, found cov-
plant remains (Figure 4a). This pit also contained ering some of the pits at Pancan, suggested to us
burned camelid bones, suggesting that it may have that these pits might have had special function; it
been filled with garbage that was removed from seemed probable these pits were purposely exca-
habitation areas of the settlement and purposefully vated filled, and sealed.
buried. Taken by itself this pit might have seemed In our examination of pits with unusual arti-
uninteresting or considered simply an expression facts andlor capstone coverings, we found that
of overall site contents, yet comparison with the most do not appear to have intentionally included
underlying midden and the black lens above the pit plant material. For example, Pit 350 was capped
demonstrates that neighboring contexts are lower with a large stone slab, but the botanical contents
in density and contain higher percentages of wood of the pit are fairly similar to the neighboring mid-
than did the pit. These facts taken together can help den (Locus 288) that was found above it (Figure
us to understand garbage disposal habits of the 4b). Another pit from the same area of the site,
prehistoric inhabitants, who obviously burned both Locus 359, was not only covered by a stone cap
animal and plant rehse and had no aversion to dis- but contained three complete miniature ceramic
posing of them in the same pit. vessels as well (Figure 4c). In this case the pit con-
Many archaeologists define pit functions by the tents are most similar to the plant remains recov-
artifacts they contain (e.g., Binford 1967; Chomko ered from the fill below. In general it appears that
and Gilbert 199 1 ; Winter 1976). When it appears unique artifact content does not signal unique or
that ceramic vessels, bones, figurines, and other intentionally included botanical remains, suggest-
items are intentionally deposited it might be ing that correlations or interpretations of the plants
 AMERICAN ANTIQUITY [Vol. 60, No. 4, 19951

Locus 363 Locus 288 Locus 347

Locus 438 Locus 350 Locus 359

Wood Qu~noa

Verbena

harfrag

Locus 371 Locus 361 Locus 361

W~ldlegume Wild legume

B C
Figure 4. Midden-filled pit in Level 2 with companion deposits (left), pit with capstone in Letel 2 with companion
deposits (middle), and pit with capstone and ceramic tessels in Level 2 with companion deposits (right). See page 719
for a key to abbreviations.

from the pits alone can be spurious. This further matrix, making the archaeobotanical assemblage
suggests that differences in artifact content (such from one indistinguishable from the plant remains
as ceramics and lithics) between pit and surround- of the other.
ing fill may also require additional scrutiny. Other types of artifacts d o not always signal
The conclusions drawn from the analysis of unique plants in pits either. Instead each pit must
Pancan pits differ from those made concerning be evaluated individually and compared to the
hearths. A consistent pattern cannot be discerned surrounding deposits. The more general context
for all pits or even all pits with the same types of of the area in which the pit occurred-such as
fill. That is, the botanical remains from pits vary midden area or a burned structure-also needs to
greatly and individual pit-fill types (clay, ash, be kept in mind as this may affect both the feature
midden, etc.) d o not show patterns that are corre- and nearby deposits.
lated to the fill type. Botanical remains may or
may not be intentionally placed in the pits,
although in these cases it appears most were not. Other features that are of value in interpretation
This is a significant finding and should serve as a of Pancan are animal burials. These features held
warning for drawing conclusions from plant promise for unique archaeobotanical material
remains recovered from pits when adjacent because, ethnographically, animal offerings
deposits have not been san~pled.O n the other (drspacho.~)o ften include plants, burned as part
h a n d it may be that fill was intentionally put into of the associated ceremonies (Allen 1988; Bastien
pits, but that the fill came from the surrounding 1978; lsbell 1985).
 REPORTS

Locus 399 Locus 664

Locus 409 Locus 668

Quinoa

A B
Figure 5. Cuy offering in Level 2 with companion deposits (left) and camelid offering in Level 4 patio with companion
deposits (right). See page 719 for a key to abbreviations.
Pancan contained the remains of 39 animals Therefore, the plant remains cannot be considered
that appear to have been intentionally buried part of the ritual burial. As with other features, the
(Sandefur 1988). Only 10 separate interments occurrence of unusual artifacts accompanying
with associated pits could be defined during exca- animal offerings is not indicative of unique plant
vation, and these are the only animal burials we remains. One such example is Locus 399, a cuy
discuss. Most offerings were of camelids (proba- burial in Level 2. This offering contained the
bly llamas or alpacas) or cuyes (guinea pigs), remains of a single cuy, turquoise and shell beads,
although one wildcat skull occurred as well. The worked bone, mineral remains, and a quartz crys-
burials were found in pits in open patio areas, tal. These unique artifacts are not matched by the
inside houses, and in and around the burned struc- plant remains; botanical materials in the feature
ture containing stored agricultural products. The are quite similar to the deposits both above and
finds were assumed to be special offerings below (Figure 5a). This distribution suggests the
because many of the skeletons were juveniles, objects were laid in the pit with the animal
many were complete and articulated and all the remains whereas the plant remains were simply in
bones lacked butchery marks (Sandefur 1988). nearby sediment used to fill the pit.
Botanical contents of 10 offerings vary widely, Pie diagrams of the floral contents from the
and show some potentially meaningful patterns. three remaining animal offerings show differ-
However, as with the other pit features, many of ences from surrounding site material. The best
these loci do not stand out when they are com- example is Locus 664, a camelid burial recovered
pared to their adjacent deposits. Of the 10 animal from the open patio area of Level 4 (Figure 5b).
offerings we examine& seven appear to have been The high proportion and density of Scirpus (tor-
filled with matrix from neighboring areas. tora or bulrush) seeds in this pit stand out not
 AMERICAN ANTIQUITY [Vol. 60, No. 4, 19951

Locus 454 Locus 645

Locus 452 Locus 614

Charfrag Charfrag

Figure 6 . Double camelid offering in Level 3 patio with companion deposits (left) and camelid offering in Structure 13
with companion deposits (right). See page 719 for a key to abbreviations.

only from the adjacent loci but from all other Level 3, again shows a high proportion of maize
deposits in the site. In this case, it would appear cob fragments (Figure 6b). If these percentages
that Scirpus seeds were intentionally charred as are considered alone, it might appear that this
part of the ceremonial offering and should have offering was similar to Locus 454 (the previous
some special significance. At the present time example). The loci close by contain mostly
Andean peoples sometimes feed this plant to lla- unidentifiable remains, clearly demonstrating that
mas and alpacas (Lawrence Kuznar, personal the botanical remains in this feature are unique.
communication 1993). This prehistoric associa- When the actual density of plant material is con-
tion suggests that the relationship is long term. sidered, however, this conclusion is weakened
Another example of unique plant contents in slightly, as the feature contained very little
an animal offering is Locus 454 (Figure 6a). This charred material (Table 2). We suggest this con-
double camelid grave contains a high percentage clusion the more tentative because small amounts
of maize cob fragments along with small amounts of carbon are more likely to be randomly distrib-
of unidentifiable material. The unusual nature of uted, whereas dense concentrations are probably
these plant remains stands out most clearly in more reliable evidence of past human behavior.
contrast to those of the loci above and below the These examples demonstrate that only about
pit. The high density of archaeobotanical material 30 percent of the Pancan animal offerings had
inside the feature in association with the bones charred plant remains included intentionally.
strengthens the argument that the plants are part From the cases where charred plant remains are
of the offering (Table 2). intentionally interred, it appears that Scirpus and
The third case, a camelid interred during the maize cobs may have played a special role in
building of one of the habitation structures in some rites involving the burial of camelids, such
 REPORTS 71 3

as use as a fuel, symbolic food for the animal, or Table 2. Density of Plant Remains from Features and

wrapping for the animal. In most instances where Companion Deposits.

plant remains are similar in the feature and the

Locus Context Density (dMatrix Volume)
neighboring loci, this comparative method keeps
us from applying erroneous importance to plants Wall fall 60.8 (729i12.0 liters)

or other artifacts in the features. Hearth 38.5 (23116.0 liters)

Fill 2.8 (33112.0 liters)

Hzlman Burials Floor 17.8 (426124.0 liters)

Hearth 4.6 (66114.4 liters)

Human remains were recovered from 16 separate Occupation zone 25.8 (309112.0 liters)

burials in Pancan (Hastorf et al. 1989). We discuss Outdoor patio 58.5 (66711 1.4 liters)

only 12, because in one case no pit could be dis- Pit with midden fill 127.2 (1526112.0 liters)

Midden 103.9 (1372113.2 liters)

cerned and in three others the burials were below

Burned storehouse 92.0 (1 104112.0 liters)

the water table; in these cases no flotation sam- Pit 345.2 (4142112.0 liters)

ples were collected. Burned storehouse 26.7 (320112.0 liters)

The Pancan burials were recovered from many Wall fall 4.4 (189142.6 liters)

different contexts. There is no single pattern of Pit 1.5 (916.0 liters)

Floor 0.9 (1 1112.0 liters)

interment; bodies were buried beneath houses, in

Occupation zone 35.5 (702119.8 liters)

patios, and within storage areas. Some graves are Pit containing stone hoes 116.5 (1398112.0 liters)

shallow, whereas others are elaborate shaft tombs Floor 16.7 (390123.4 liters)

with stone slab linings. Grave goods are not Charred lens 55.9 (839115.0 liters)

many, but include metal objects, stone and bone Pit with midden fill 52 1.3 (594311 1.4 liters)

Midden 18.4 (232112.6 liters)

tools, and ceramic figurines. The distribution of

Midden containing ash 16.6 (199112.0 liters)

these objects is somewhat clustered in that some Pit with capstone 15.9 (191112.0 liters)

burials have no offerings whereas others have as Fill 14.1 (338124.0 liters)

many as 10, yet none of the graves would be Midden 96.9 (2325124.0 liters)

described as rich. Pit with capstone 21.4 (24411 1.4 liters)

Fill 14.1 (338124.0 liters)

As in the pits and animal offerings, no single

Fill 15.3 (183115.5 liters)

botanical "signature" was apparent for the human Cuy (guinea pig) offering 8.8 (5316.0 liters)

burials at Pancan. That is, the botanical contents Fill 10.0 (120112.0 liters)

of all burials are not the same, nor are there plant Open patio 5.4 (84115.6 liters)

types associated only with burials. Primary buri- Camelid offering 64.7 (38816.0 liters)

Gravel fill 4.7 (7011 5.0 liters)

als do not appear to demonstrate more regular

Midden with wall slump 7.3 (87112.0 liters)

botanical patterns than secondary burials. Again, Double camelid offering 13.0 (7816.0 liters)

it is necessary to consider each burial individual- Open patio 11.3 (135112.0 liters)

ly as it compares to its neighboring loci, keeping Fill 5.3 (3216.0 liters)

in mind where it occurred in the site. As in pits Camelid offering 2.0 (1216.0 liters)

Wall fall 4.4 (1 89142.6 liters)

and animal offerings, the botanical remains in

Burned storehouse 1977.8 (1 18,667160.0 liters)

burial fill seem most often to be linked to the larg- Slab-lined burial 165.0 (99016.0 liters)

er, more general site context where they were Midden 38.4 (691118.0 liters)

deposited. For example, graves that were dug into Intrusive burial pit 47.7 (572112.0 liters)

midden or in the area of stored crops have the Burial with capstone 110.8 (1995118.0 liters)

Paved floor 5.8 (69112.0 liters)

same plant materials as the material identified

Wall fall 244.6 (2935112.0 liters)

from the surrounding loci (if we are correct in Subfloor burial 4.5 (2716.0 liters)

postulating that most of the "charred fragments" Floor 3.4 (102130.0 liters)

are tuber remains). (Plough zone: no sample)

Five burials were excavated in and around the Burial in midden 36.3 (458112.6 liters)

burned storage structure. The plant material in Prepared surface 16.7 (390123.4 liters)

four of these five burials is similar not only to their Note: Fragments per liter of processed site sediment.
neighboring deposits, but also to one another (e.g.,
 AMERICAN ANTIQUITY [Vol. 60, No. 4, 19951

Locus 641 Locus 477

Figure 7. Slab-lined burial in burned structure with companion deposits (left) and burial with capstone in burned
structure with companion deposits (right). See page 719 for a key to abbreviations.

Locus 503 in Figure 7a). The pattern of plant Dense quantities of large plant fragments in
remains in the storage area includes a dominance these four features also negate the notion that the
of unidentifiable fragments and the presence of plants infiltrated the burial fill at a later time.
one or more of the four major crops of the area: Large quantities of maize, potatoes, and legumes
maize, quinoa, tubers, and legumes. The similari- do not work their way through sediment as easily
ties of the botanical remains in the graves and as a few small seeds might.
other loci in the storage area suggest most of the In contrast, there is one example of a burial
graves were probably created or reopened after the with distinctive contents that comes from this
burning event. During excavation it had been same area of burned storage (Figure 7b). The
thought that the burials predated the fire, but this botanical content of Locus 660 is unusual in that it
is unlikely as graves would not have been left open contains a very dense fill of well-preserved
for materials to fall into nor would there be such a domesticates, including nearly equal parts maize
high proportion of eroded material (charred frag- kernels, quinoa, and tubers. This pattern is not
ments) if the bodies were purposefully buried with only different than other storehouse loci-which
foodstuffs that became charred during the fire. contain about 75 percent unidentifiable plant frag-
Similar densities of materials inside and out- ments-but this depositional signature is found
side the features infer they are derived from the nowhere else in the entire site. The low percentage
same source, and again we suggest these burials of unidentifiable material hints that this body may
were interred after the structure burned. It is like- have been interred prior to the conflagration and
ly that each grave was dug into an area of burned that these crops were included intentionally. The
crops and then refilled with this same material. actual charring of the material may have occurred
 REPORTS

Locus 441

(No Sample)

Locus 471 Locus 444

Locus 613 Locus 343

A B

Figure 8. Burial in Structure 13 with companion deposits (left) and burial in Level 1 with companion deposits (right)

See page 719 for a key to abbreviations..

as part of the burial ritual or have been caused later It is particularly interesting that the inclusion
by the fire that destroyed the structure. In either of special artifacts and the presence of capstones
event, this may be early evidence of the practice of are not necessarily correlated with intentionally
placing burials (entire or partial) with stored deposited plant material. For example, Burial 444
crops, done today for the symbolic protection of (from just under the plough zone, where a crew
the harvest (Allen 1988:59). member failed to collect flotation samples) was
There is a second burial that shows some dis- associated with a hoe, a piece of metal, two pieces
tinction from surrounding loci, although this case is of slate, and six round stones, yet the botanical
weaker. Locus 47 1 , located inside one of the habi- materials in the fill around the body are not unlike
tation structures (Figure 8a), does not have propor- those from the underlying matrix (Figure 8b).
tions of taxa similar to its adjacent floor or the wall These comparisons of burial matrix and neigh-
fall. This suggests the plants were not derived from boring deposits suggest plant remains were not
neighboring loci. It is difficult to assess the impor- normally intentionally included nor burned as part
tance of these plant remains as they are so sparse. of the burial practice at Pancan. However, only
Because the plants are so few we conclude that through such careful comparisons can we detect
while they may not be contaminants from the and understand ritual use of plants and solve some
deposits above or below, they still may not have chronological or stratigraphical questions.
been included intentionally. Again, we feel more
confident in assigning intent to dense deposits, Summary of Sampling Effects on the
especially of culturally significant plants such as Interpretation of Botanical Remains
crops, than to small amounts of weed seeds, which Several conclusions can be drawn from these
may have been moved by wind or water. comparisons of features and their neighboring
71 6 AMERICAN ANTIQUITY [Vol. 60, No. 4, 19951

deposits. Most important is the graphic illustra- about the natural and cultural depositional history
tion of the necessity for collecting and comparing of archaeological contexts.
flotation samples from all deposits encountered in In addition to addressing these questions con-
archaeological sites. Without these contrasts of cerning the origins of plant remains in archaeolog-
adjacent loci, the functional or symbolic interpre- ical deposits, this exercise also illustrates how care-
tations based on archaeobotanical materials from ful blanket sampling can help unravel questions
specific proveniences are weakened and perhaps concerning chronology, stratigraphy, context, and
even suspect. For example, we have demonstrated meaning. For example, through detailed compari-
that the only way to determine whether plant son of burial features and neighboring deposits
remains recovered from hearth deposits were in from the burned storage area, we found the burials
situ is to sample all other contiguous deposits. took place at different times relative to the destruc-
Because this relationship has been investigated [ion of the structure. This was not evident during
we have more confidence in stating that this is the excavation nor from the other material in the buri-
case for 10 of the 1 1 hearth features excavated at als. Thus the careful use of botanical information
Pancan. can help clarify not only the chronology of the
The comparisons of samples from adjacent burials, but also that of the surrounding deposits.
loci are also of value in the interpretation of pits, This exercise further demonstrates the impor-
animal offerings, and human burials. In each of tance of knowledge about the larger cultural con-
these cases the comparison helps us discern text of the samples in interpreting plant remains
which loci had potentially independent sources of from any given area of a site. In our case at
plant remains and which did not. These compar- Pancan. knowledge of the architectural layout,
isons may also aid in illuminating patterns from midden distribution, and location of the burned
features that might not look unusual until such a storehouse are crucial to interpretation. In many
contrast is made. It is of interest that most of the archaeological situations. knowledge of other arti-
pit features contain materials that are similar to fact distributions is also of assistance, although in
nearby sediment deposits, suggesting that the our examples this proved to be of limited value.
plants may not be related to the use of the pit or to Even in our brief analysis of a small portion of
other intentionally included materials. the loci from Pancan, we can begin to uncover
Gasser (1985) considered only postoccupa- some of the specific, historically meaningful rela-
tional deposits above features as a source or recip- tionships between plants and the Prehispanic peo-
ient of botanical contamination. Here we consid- ple. For example, we found that not all burials and
er the deposits both above and below a feature in offerings were accompanied by charred plant
order to determine whether exchange of plant materials. Those that do show heavy emphasis on
remains took place or whether the original the domesticated foods that were most important
sources were one and the same. Without these in the region, with the interesting exception of
comparisons, we might have considered some of Sciryus associated with one of the llama offerings.
the feature contents to be unique when they are In sum, there are at least six advantages
merely a reflection of the deposit into which the gained from our comparisons of adjacent botani-
feature intruded. Plant remains in features may cal samples. First, by comparing material inside
also have come from slumping or overburden features to those adjacent to them one can
from later activities that do not relate to the fea- strengthen conclusions concerning the plant (or
ture at all. Conversely, floor and occupation other artifact) deposition. Second it may be nec-
deposits might be incorrectly interpreted if not essary to contrast features with surrounding
considered in conjunction with pits or other fea- deposits before their unique characteristics are
tures that penetrate them. It is possible that floral recognized. Third by comparing features with
remains originally deposited in features may have their contiguous matrix, one may prevent spuri-
moved into other areas. With careful sampling, we ous interpretations about the features' contents,
can assess these issues and other depositional for example when plants are simply derived from
queries systematically with no need to speculate areas penetrated by the features or later slumped
718 AMERICAN ANTIQUITY [Vol. 60, No. 4, 19951

the Later Intermediate Period: A Report on the 1986 ed at the 53rd Annual Meeting of the Society for
Field Season Andeat1 Part 2:81-129. American Archaeolog~:Phoenix
Hodder. I. Scarrq, C. M.
1987 Tile A r c ~ h a r o l o g ~o~f ('ot7tr.triiciI .Mcut~it~g.c. 1993 firagitlg cltld Ftzrtiiitlg it1 rile Easrertl Iti)otllat~d.\.
Cambridge Unilersitq Press. Cambridge. Uni~ersityPress of Florida. Ciaines~ille.
Hubbard, R. N L. B. Sharer. R.. and W Ashrnore
1975 Assessing the Botanical Component of Human 1979 Frit7dunlrt1tnl5 o f .ircilueolog~~. Benjarnin
Paleo-Economies. Bullrrltl o f the It~tritritc, of Cummings, Menlo Park. California.
.Archaeologj~12.197-205 (London). Sneath. H. A,. and R. R. Sokol
Hughes, P. J.. and R. J. Larnpert 1973 ,V~rtnrt.iculTuiotlonl>.. Freeman. San Francisco
1977 Occupational Disturbance and Types of Archaeological Spector. J. D.
Deposit. .Journal of Archarological Sciet1c.r 4 135-1 30. 1970 Seed Analysis in Archaeology. If'isc~on.sitl
Isbell, B. J. 4rc~hac~oIogivt 5 1 : 1 6 3 190.
1985 To Defi,tld O1inre11~e.r.W a ~ e l a n d Press, Prospect Stein, J.
He~ghts.Illinois 1983 Earthworm Activity: A Source o f Potential
Jarman. H. N.. A. J. Legge. and J. A. Charles Disturbance of Archaeological Sediments. .Amcric~at~
1972 R e t r i e ~ a lof Plant Remains frorn Archaeological .-lntiquit~,48:277-298.
Sites by Froth Flotation. In Paper.\ it7 Economic. Stockton. E . D.
Prc~histot:~,. edited bq E. Higgs. pp. 3 9 4 8 . Cambridge 1973 Shaw'a Creek Shelter: Human Dlsplacement of
University Press. Cambridge. Artifacts and Its Significance. llatlkit~d9: 1 12-1 17.
Jones. Ci , K . M'ardle. P. Halstead and D. Wardle Struever. S.
1986 Crop Storage at Assiros. S t i o ~ t i f i c ..itnerictrn 1968 Flotatlon Techniques for the Reco~eryof Small-Scale
254:96-103 Archaeological Remains. .Attlc~ric~utl .Atltiqlrih 33:353-362.
Lennstrorn. H. A. Toll, M . S.
1992 Intra~iteSpiitiul luriahilic atld Re.\ource ('tilizatiotl 1988 Flotation Sampling. Problems and Some Solutions.
in rhr Prcilisroric Pc.ru~,iatlH~glllanti.\:At1 Eil,Iorariot~of \r'ith Examples from the American Southtvest. In C'lrrretlt
.llrthoti atlti Tlleor:r it1 PrrIcoc,riltlohorat~>~.
Unpublished Pul~~oethtlohotcit~~~. edited by C Hastorf and \( Popper. pp.
Ph.D. dissertation. Center for Ancient Studies. Unlversitq i&52. University of Chlcago Press. Chicago.
of Minnesota, Minneapolis. Tufte. E. R.
Lcnnstrorn, H A,. and C. A. Hastorf 1983 Tilt li.s~ru1 O~\plu> of' Quut~tituril'eIt1fornrrrt1ot7.
1992 Testing Old Wives' Tales In Paleoethnobotany: A Graphics Press. Cheshire. Connecticut.
Comparison of Bulk and Scatter Sampling Schcrnes T u k e ~J. W
from Pancin. Peru. Jorrrtiul of 4rc~hurologicalScirtlcr 1977 E~plorutor;~Dutir .Atlu!, 51.5. Addison-Wesley.
19:?05-229. Reading, Pennsyl~ania.
Matthe\rss. J. M. \an der \ken. M.. and N. Fieller
1963 Stratigraphic Disturbance: The Human Element. 1982 Sarnpling Seeds. Jotirt~nlo f 4,~hcrrol0gic~al Sc1et1c.r
At~tiyuir~ 3 9 295-298. 9:?87 298.
Miller. N. Watson. P. J.
1988 Ratios in Paleoethnobotanlcal Analysis. In ('rirro~r 1976 In Pursult of Prehistoric Subsistence: A Comparati~e
PaIeoethnohornt7!~, edited by C. Hastorf and \: Popper. Account of Some Contemporary Flotatlon Techniques.
pp. 72-85. Unl~ersityof Chicago Press. Chlcago. .Mid-C'ot~t~t~rt~ttrl ./o~irnulo f .Arc~i~cirolog!~1 :77-100.
Pearsall, D. M. Wetterstrorn. U:
1983 Evaluating the Stabllltq of Subsistence Stratcglea by 1978 ( ' o g n i t i ~ e Systems. Food Patterns. and
the Use of Paleoethnobotanical Data. .lotirnal o f Paleoethnobotanq. In TIICJ.\rrtrrt.c, at7ti Strrtlrh c:f'
Etiinohiolog>. 3: 12 1-1 37. E t h t ~ o h o t i r t ~ ~edited
. by R . Ford, pp. 81-95.
1989 PaIeorrht1ohotut71~. Academic Press. Orlando, Florida. Anthropological Papers No, 67. Museum of
Popper, \: Anthropology. Uni~eraityof Michigan. Ann Arbor.
1988 Selecting Quantitative Measures in Wlnter. M . C.
Paleoethnobotany. In Ctirretlt Ptrlrorthnohorat~>,.edited 1976 The Archaeological Household Cluster In the \'alley
by C . Hastorf and \: Popper. pp. 53-71. Uni~ersitqof of Oaxaca. In Tilr earl^. lIetoumrric.nt7 l'illagr. edited
Chicago Press. Chicago. by K . Flannerq. pp. 25 31 Academic Press. New h r k .
Renfreiv. J.
1973 Ptiltrroc~rirt7ol~ori1n>~. Colurnhla Lnllersity Press.
Neu York.
Sandefur. E. C
1988 rlornrstic At71nluI [;.ye rtl tiic, C'et~rrcrl 4nde.\ Earl,
Intermrdiat~~ Period to tile [.are Hor~zotl.Paper preaent-
 REPORTS

Appendix: Raw Botanical Data.

Names in Figures Names in Appendix Category/Taxon

Locus LOC Locus number
-
VOL Volume of processed soil
N Number of individual samples
Maize Kern KERN Zea ma,vs L. kernel
Cobs CUP Zea mays L. cupules
EMB Zea m a y L. embryos
Quinoa CHEN Chenopodium quinoa Willd.
LUP Lupinus mutabilis Sweet.
Dom Legume DLEG Fabaceae (domesticated)
Charfrag CIIFRCi Charred fragment (cf. tubers)
Scirpus SCR Scirpus
Wild Legume WLEG Fabaceae (wild)
Verbena VER krbena
Sml Grass SPO Poeceae (small)
Lrg Grass LPO Poaceae (large)
Med Grass MPO Poaceae (medium)
LAB Lamiaceae
hlALV Malvasbum
22 1 Unknown # 22 1
CY P Cyperaceae
RUB Rubus
REL Relbunium
COhlP Asteraceae
SOL Solanaceae
AhlB Ambrosia
POT Potamogeton
ELEO Eleocharis
POLY Polygonaceae
OX Oxalis
NIC Nicotiana
HY P Hyporis
PLAN Plan tago
SIS Sisvrinchium
PHAS Phaseolus
OENO Oenothera
224 Unknown # 224
202 Unknown # 202
235 Unknown # 235
GAL Galium
242 Unknown # 242
-
258 Unknown # 258
Unidseed UNlD Unidentifiable seed
Tubers TUB Tuber
-
SPUD Solanum tuberosum L.
Wood N'D Wood
Stalks STK Stalk
TWG Twig
PED Peduncle
AMERICAN ANTIQUITY [Vol. 60, No. 4,19951
aMB KT ELEO F0r:I aY NIC iPIP P W SIS PHAS OENO 224 202 235 GL_L 242 258 SPUD
0 0 0 0 0 0 0 0 0 0 1 4 0 0 0 0 0 0
0 0 1 0 0 0 0 0 0 0 0 3 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 2 0 0 1 0 0 0 0 5 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 1 2 2 0 0 0 0 1 0 0 0 0 0 0 0 0 0
0 0 0 1 1 0 2 1 0 0 0 4 0 0 0 0 0 0
1 0 0 2 1 0 1 0 5 0 1 0 0 0 0 5 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
1 0 0 0 0 0 1 0 1 0 0 0 0 0 0 0 0 0
0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 1 0
0 0 0 0 0 0 0 0 0 0 0 3 0 0 0 0 0 0
0 0 0 8 0 0 0 0 0 0 0 4 0 0 4 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0
2 0 0 0 0 0 3 0 0 0 0 0 0 0 0 0 0 0
0 0 0 1 1 0 1 0 0 0 0 0 0 1 4 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 1 1 0 0 0 3 0 1 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 1 4 0 1 9 0 0 0 0 0 11
0 0 0 0 0 0 0 0 0 0 0 1 2 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 1 6 0 0 0 0 0 0
0 1 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0
0 0 0 0 0 1 0 0 0 0 0 2 9 0 0 0 0 0 0
0 0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 4 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
1 4 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0 0 0