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