Much along these lines can be learned by examination of large random samples such as that afforded by the Topeka accident. Although only a few species in this sample were measured, the results secured seem to show further the need for segregation of age classes in taxonomic work with some species. Figure 2 shows the variation in the lengths of wing and tail in the Nashville Warbler. It is evident from the figure that in both of these characters the four sex- and age-classes differ significantly. An accurate idea of geographic variation in this species could not be obtained without separating these classes in comparisons. Age classes in spring and summer, long after the skull is completely ossified, can be segregated only if differences in plumage can be found. In the Nashville Warbler, such differences occur in fall (see annotated list) but these probably are obliterated by the partial prenuptial molt. These facts emphasize the importance, for taxonomic studies, of aged specimens collected in late summer or early fall on their breeding ground and in fresh winter plumage. Figure 3 shows the variation in length of wing in the Lincoln Sparrow in which age seems to be of much less importance than in the Nashville Warbler. Males and females of the Lincoln Sparrow differ significantly in wing-length, but adults and immatures are of nearly the same size. It would seemingly not be necessary to separate age classes in studies of geographic variation in the Lincoln Sparrow. Size data for some other species are given in the annotated list. WEIGHTS.--Little seems to have been done to determine the correlation of weights with sex- and age-classes. Weight may be the best single index of over-all size and is especially useful to students of the physiology of migration. Weight, however, is subject to the considerable variable imposed by fat condition. In large and comparable series, this variable is probably removed insofar as comparability of means is concerned, but the high variability of weight in most cases naturally results in more overlap (less separability) between populations with means close together than is found with most linear measurements. [Illustration: FIG. 2. Statistics of variation in length of wing and tail in the Nashville Warbler. The solid lines represent the observed ranges in millimeters. The stippled boxes represent two standard errors ([Greek: s]m) to each side of the means (vertical lines). The open boxes represent one standard deviation ([Greek: s]) to each side of the means.] [Illustration: FIG. 3. Statistics of variation in length of wing in the Lincoln Sparrow.]