Old World species, varies markedly. In New World specimens, one A. mutica has 7 on one side, 8 on the other, and 8 occur on both sides of one other (of a total of 16 examined). One of twenty A. spinifera, and one of eight A. emoryi have 8; the single A. ferox (Schneider) has 7. Accordingly the suggestion by H. M. Smith (Field Mus. Nat. Hist. Zoöl. Ser., 23:19, 1939) that Platypeltis Baur be resurrected for the American soft-shelled turtles on the basis of the occurrence of only 7 costals, is untenable.

The generic allocation of American soft-shelled turtles has varied considerably in recent years: Smith (loc. cit.) uses Platypeltis; Pope (Turtles of the United States and Canada, p. 343, 1939) uses Trionyx Geoffroy; and Stejneger (op. cit., p. 8) uses Amyda Geoffroy. As stated above, use of Platypeltis at the present time is unwarranted, since no constant difference has been discovered that would support generic separation of Asiatic and American members of this group. New World turtles should be placed either in Trionyx or in Amyda, depending upon the interpretation of type designation for the latter name. Malcolm Smith (Bull. Raffles Mus. 3:2, 1930) and others have considered that, as a part of the original description, Geoffroy (Ann. Mus. Hist. Nat. Paris, 14:20, 1809) designated the type species of his new generic name Trionyx as aegypticus E. Geoffroy (= triunguis Forskål a well-recognized species). Stejneger argues that Geoffroy did not adequately designate a type from among the many species he treated in his genus Trionyx, and that it remained for Fitzinger (Syst. Rept., p. 30, 1843) to select one of these as a type; he chose coromandelicus Geoffroy, which is a synonym of granosa Schöppff, a species belonging to a different genus (as now recognized) from that to which triunguis belongs, although Geoffroy had made the mistake of considering both groups as members of his genus Trionyx. Now if Fitzinger's type designation is accepted, the name Trionyx is to be applied to that group containing granosa (only one other form is known in the genus, and both forms occur only in India and Burma), whereas the name Amyda of Geoffroy (op. cit., p. 1) is applied to the genus (as now recognized) which includes triunguis and some 20 other species of Asia and North America. The type of Amyda is a typical member from Asia (cartilagineus Boddaert). On the other hand, if Geoffroy's type designation is accepted, the American forms (and the others of that genus) would take the generic name Trionyx, of which Amyda would be a synonym, and the genus to which granosa belongs would take the name Lissemys Malcolm Smith (Fauna Brit. India, Rept. Amph., 1:154, 1931).

Stejneger discussed the various aspects of this problem (op. cit., pp. 6, 7), and I can add nothing to his discussion. His arguments for the acceptance of Fitzinger's type designation rather than that of Geoffroy are well founded upon the statement of the International Rules of Zoölogical Nomenclature, while those of Smith are not. In weighing these two alternatives, the practical value of maintenance of the "status quo" is not here important, for the whole system of nomenclature in this field is completely upset; any conclusive decision would be of great practical value and one alternative holds no special, practical advantage over the other. Accordingly, it seems reasonable to consider the matter closed with Stejneger's analysis, retaining Amyda for the American and related species of soft-shelled turtles. That this assemblage contains natural subgroups that may warrant subdivision into other genera is obvious, but to none of these will the name Trionyx be applicable.