Plains Indian Studies - Smithsonian Institution Libraries

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