Ward-—The V aviations in Pro port i ons in Bats. 633 
length for each of the several measurements noted. In some 
instances these curves present indentations suggestive that an 
insufficient number of individuals were measured. It will he 
noted, however, that this irregularity of curve is not always 
more marked where the number of individuals is the more re¬ 
stricted. Were the specimens from different localities some of 
the curves with double or multiple apices would suggest geo¬ 
graphical races with differing maxima of measurements; but 
under the circumstances the dearth or occasional absence of 
certain intermediate measurements seems to be fortuitous. 
In this study care has been taken to restrict the specimens 
examined to one locality, and frequently to a single cave, except 
in the case of Nyctinomm hmziliensis which is generally recog¬ 
nized to have a rather wide distribution. There is not over 
75 miles distance between the extremes of the localities repre¬ 
sented in the tabulated specimens of this species. The differences 
in altitude would perhaps be the factor most apt to produce 
variation among these, but the diagram does not seem to indi¬ 
cate that such has been the case. Although for my own pur¬ 
poses I noted the measurements of several other species and of 
some additional small lots from other localities of the species 
herein tabulated, I have not considered it advisable to use these 
measurements in this connection. 
In the present study the eight species are represented by 415 
specimens which afford a total of 6,863 measurements. 
In scanning any of these tables the general similarity of many 
of the graphs is readily seen. Vertices and nadirs are usually 
found to follow their kind in any given specimen though minor 
variations somewhat obscure the apparent correspondence in 
trend of the graphs. Those of length of head and body com¬ 
pared with those of length of tail appear the most dissimilar 
because of the more extensive range in actual length in these 
than in other measurements ; though in examining the propor¬ 
tionate amount of variation it is found to be less than between 
some other measurements. 
In most of the measurements of parts the variation in actual 
length is moderate, though were this computed on the mean or 
average length it would be found in some instances to be pro¬ 
portionally as much as 33 per cent. In a considerable number 
of instances the binomial curves following the graphs show that 
the maximum number of measurements recorded corresponds 
