Plains Indian Studies - Smithsonian Institution Libraries

NUMBER 30
between sites. Since so much of Plains zooarcheology
is bison dominated, such a procedure could
generate a significant amount of information
from older collections, information that could be
compared to data from the most intensively excavated
site.
Both Watson (1972) and Thomas (1969) describe
how adjustments can be made where part
of the site is still available for intensive excavation.
These techniques use the frequencies derived
from small, intensively collected subsamples to
create recovery factors relevant to different sizes
of bones or animals. The recovery factors are then
used to adjust the frequencies of the non-intensively
collected sample. One should be cautioned
that this procedure makes strong assumptions
about the homogeneity of the deposit. Falk
(1977:154) cites some examples from the Plains
where this kind of adjustment would be an unwise
procedure:
Frequency distributions for major classes plotted by provenience
unit (outside house, roof/floor fill, subfloor pit etc.)
at Jake White Bull revealed a "differential distribution of
both fish and amphibian remains, with the great majority of
each class occurring in subfioor pits . . . ." Evidence from the
VValth Bay and Bower Grand sites demonstrate similar
patterns at comparable recovery levels ....
The data from Mowry Bluff (Falk 1969:45, table
8) also show definite intrasite variability, since
one deposit, feature 7, has a very different assemblage
compared to the rest of the site.
Estimating the Finds Population
The most common statistic employed in zooarcheology
is the Minimum Number of Individuals
(MNI) (see the discussions in Bokonyi, 1970; Casteel,
1977; Chaplin, 1971; Daly, 1969; Grayson,
1973, 1978; Holtzman, 1979). MNI is equal to
the frequency of the most abundant bone type
(humerus, radius, etc.) in the subsample of bones
assigned to a particular taxon. It represents the
smallest number of animals necessary to produce
the sample of bones observed. MNI has often
been cited as the best estimator of species frequency
in the finds population (e.g., Butler,
161
1974:97; Falk, 1977:153), and it is the statistic
implied in the passage that introduces this paper.
Nevertheless, some aspects of the use of the statistic
in the analysis of the finds populations of
habitation debris can be called into question.
Unfortunately, the exact method of calculating
MNI is not often spelled out in reports. An important
exception is contained in the excellent
report of Parmalee (1977) on Plains avifauna.
Because it illustrates well important points about
MNI, his procedure is cited here in detail (Parmalee,
1977:193):
The minimum number of individuals represented by each
species was determined by selecting the element which was
the most numerous in the site sample, e.g. a right humerus,
a left femur, and so forth. Once this number was obtained,
the elements were then compared with those of the opposite
side to determine if any were paired; those which could not
be paired were added to the total MNI. In some instances
when the MNI was based on a particular element from all
adult birds and a juvenile was represented in the sample but
not by that particular bone, it was obvious that another
individual should be added to MNI. The assumption was
made that the individuals represented at each site were
unique to that site and that parts of one bird would not, in
all probability, be encountered in two or more sites. To
illustrate, remains of the turkey vulture occurred in seven
sites and it was assumed, therefore, that at least seven
individual birds were represented. If all 18 elements had
been considered together, however, the MNI would have
been three based on the carpometacarpus.
Several objections can be raised about the use of
MNI. The variance associated with MNI is larger
than that associated with other estimators of finds
populations (p. 166), mostly because it uses only
part of the bone elements assigned to a taxon to
estimate frequency (Holtzman, 1979:86). Much
more important is the point that by focusing
attention on the relative abundance of whole
animals, the MNI statistic makes an assumption
about the relationship of the finds population to
the consequences population. It can be argued
that a few bones do not necessarily represent the
original deposition in the site area sampled of a
whole animal (what Wedel (1970) was wrestling
with in his review of the Mowry Bluff sample).
Use of rates for whole animals implies an assumption
that the finds collection represents a taphon-