Co., Kansas; No. 2757, Smoky Hill River, Trego Co., Kansas; No. 23026, no data.

The three abnormal specimens vary in width/height ratio from 1.83 to 3.14. In the 37 normal turtles measured, the corresponding ratio is 4.64 to 7.85. The ratio of 4.64 is possibly subject to correction since the shell tends to warp in some specimens, especially in those retaining the skin about the periphery of the shell. The warping does not produce a marked convexity in transverse section, but does so in longitudinal section. Accordingly the height as here measured is little effected, and the comparison with width rather than length of shell provides for the lesser error from warping. There appears to be no close correlation of proportions with either size or sex.

It is of interest that Amyda ferox is the most distinctive in proportions of the carapace. Its carapace is longer in relation to its width than that of any of the other species. The average relative length of the carapace of A. emoryi is intermediate between that of A. ferox and the averages of A. spinifera and A. mutica, but the overlap in range with the latter two is complete.

The cause for kyphotic anomalies is unknown. That it is accompanied by a greater degree of growth in the vertebral column than in the periphery of the costal plates is obvious. There seems to be no well-established accommodation for the difference in growth, since the hump produced by it varies considerably in form.

There is no trend from small to large specimens in size of the hump; large and small humps occur in both large and small specimens. Accordingly it seems that the humped condition is developed in the late embryo or early post-embryonic life, and does not later change.

An apparently reasonable hypothesis is that the costal plates ankylose distally with the ribs early enough in embryonic life so that any differential in growth rate between them and the vertebral column is translated into abnormal contortions of the body. Agassiz and others have shown that the costal plates normally do not fuse with the ribs by the time of hatching; the fusion then does not normally occur in the embryonic stage. Presumably, once fused, the costal plates and vertebral column normally have equal growth rates, since the height/width ratio does not change significantly with increased size. It is well known that fusion takes place in young specimens soon after hatching; in all skeletons examined of this genus, from the smallest (62 mm. in length) to the largest (295 mm.), the fusion has occurred. Therefore, the normal time of fusion must be approximately at the time of hatching.

Although costal plates and the vertebral column grow in direct proportion to each other throughout life from a period shortly after hatching, the vertebral column apparently grows more rapidly than the costals shortly before and possibly also shortly after hatching, at least in kyphotic and probably also in normal specimens. An exceptionally early date of fusion of costal plates and ribs would thus result in a kyphotic condition, and it may well be assumed that the earlier the fusion, the greater the hypertrophy would be. Whether or not this hypothesis correctly accounts for kyphosis in turtles can be ascertained only by further study.

Stejneger (Bull. Mus. Comp. Zoöl., 94: 12, 1944) regards the presence of 8 neurals as opposed to 7 as an important peculiarity of A. mutica. The 42 specimens for which the number of neurals is recorded reveals, however, that there is greater variation than previously supposed: in 16 A. mutica more than half (9) have 7 neurals and the remainder (7) have 8. Eight neurals were recorded also in 2 of 18 spinifera, and in 1 of 7 A. emoryi. Seven neurals are present in the single specimen of A. ferox examined.

It is of interest also that the number of costals, which has been reported to be consistently 7 in New World species and 8 in