Plains Indian Studies - Smithsonian Institution Libraries
Plains Indian Studies - Smithsonian Institution Libraries
Plains Indian Studies - Smithsonian Institution Libraries
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
160 SMITHSONIAN CONTRIBUTIONS TO ANTHROPOLOGY<br />
sent alongside the indigenous wild taxa. Bone<br />
fragments from these domestic species can easily<br />
be confused with small fragments of bison, antelope,<br />
or deer. Because of the more complex zoological<br />
picture, it is more difficult to assign a bone<br />
fragment to a taxon purely on the basis of size.<br />
As a result, bone counts are likely to be reflective<br />
of historic and prehistoric situations in different<br />
ways. In general, the number of bone fragments<br />
identified as belonging to a species varies with the<br />
number of potentially misleading forms in the<br />
sample, the ability of the zooarcheologists to observe<br />
the distinctions, and the variability of the<br />
distinctions themselves.<br />
Figure 24 is an attempt to illustrate the interplay<br />
of all these biasing factors and the correspondence<br />
between Cowgill's model of cultural<br />
information and Clark and Kietzke's discussion<br />
of fossil assemblages. Overall, the information<br />
content of a collection tends to decrease with<br />
time, while the interplay of cultural (C) data and<br />
natural (N) data becomes more complex. Since<br />
the features observed in a faunal sample must be<br />
partitioned between cultural and natural factors<br />
in order to interpret past behavior, the transformation<br />
processes that change one level into another<br />
are of central interest to zooarcheologists.<br />
Sullegic Bias<br />
It has been amply demonstrated (Clason and<br />
Prummel, 1977; Thomas, 1969; Watson, 1972)<br />
that the technology of collection—screening, water<br />
sieving—can markedly affect the number and<br />
size of bones collected (Semken, 1971:111). To<br />
estimate sullegic bias, it is necessary to determine<br />
what part of the potential finds are not being<br />
adequately collected. Watson (1972) demonstrates<br />
that each collection technique is associated<br />
with what he calls a "critical size." A critical size<br />
is the smallest size a fragment can be and still be<br />
certain to be recovered by the collection technique<br />
employed at the excavation. Critical size<br />
can be determined empirically by examining a<br />
histogram that shows the frequency distribution<br />
of the different bone fragment sizes found in an<br />
excavation. Above the critical size, the distribution<br />
will form a smooth curve; below, it will drop<br />
to much lower values and the curve will be<br />
sharply irregular. Another important variable in<br />
analysis is the "minimum size." This is the smallest<br />
a fragment may be and still be identified as to<br />
what kind of bone it was and to what kind of<br />
animal it belonged. For a single species this<br />
trephic variable changes from collection to collection<br />
based on the number of potentially confusing<br />
forms encountered. Obviously, minimum size is<br />
larger with larger animals.<br />
It should be possible to use the concepts of<br />
critical size and minimum size to design an excavation<br />
sampling scheme. If on the basis of<br />
historical information relating to the site, ethnographic<br />
considerations, or previous archeological<br />
experience in the region, it is possible to determine<br />
what species are likely to be important for cultural<br />
interpretation, a screen size can be chosen<br />
that is small enough to capture a critical size less<br />
than the minimum size for the species of interest.<br />
For example, if the quantitative relationship between<br />
antelope and bison is deemed to be the<br />
primary zooarcheological statistic of interest, a<br />
screen size would be chosen that is larger than<br />
would be required to detect reliably the quantitative<br />
relationships between rodent taxa. Adoption<br />
of this strategy, however, while it may satisfy<br />
an immediate goal of speeding up excavation,<br />
carries with it the responsibility of explaining in<br />
the future why potentially retrievable information<br />
was ignored.<br />
More importantly, the same relationship can<br />
be used to salvage information from collections<br />
obtained under less than ideal sampling conditions.<br />
If the collections of interest are sorted in a<br />
nested set of graded screens, it would be possible<br />
to determine their critical sizes. A collection of<br />
unsorted faunal remains can be passed sequentially<br />
through a set of graded screens. The resulting<br />
sized subsamples can be analyzed quantitatively<br />
to determine empirically the collection's<br />
critical size. Information about smaller species<br />
would be lost, but at least the bone counts reported<br />
for the larger forms would be comparable