TESTUDINID^. 371 



The metacarpal bones are very short, usually broader than long. The whole foot is short 

 and heavy, resembling that of an elephant. 



The pelvis is constructed on the plan of that of the Emydidae. 



The femur is distinguish! from that of the Emydidae in the union of the two trochanters 

 by a ridge which runs nearly on the level of the head of the bone, when the femur is held perpen- 

 dicularly. Between this ridge and the head there is a pit of some depth. The metatarsals and 

 the phalanges are not so short as the corresponding bones of the forelimb; but they are much 

 shorter than those of the Emydidae. There are only 4 digits, the fifth metatarsal being reduced 

 to a rudiment. No digit has more than 2 phalanges. 



Little is known regarding the feet of Hadrianus, but we may confidently expect to find only 

 2 phalanges in each digit. While the structure of the digits of Stylem ys is not certainly deter- 

 mined all the known facts point to conditions similar to those found in the other genera of the 

 family. 



The most striking differences between the skulls of the Testudinidae and those of the Emy- 

 didae are found in the excavation of the roof of the mouth of the members of the former 

 family and in the closure of the stapedial notch. The palate rises high above the level of the 

 crushing-surfaces of the upper jaws and the vault thus formed is carried back to between the 

 quadrates. The skull of HaJrianus is wholly unknown. The skull of Stylemys presents the 

 same vaulting as that of Testudo. 



The geographical distribution of the Testudinidae (fig. 14, p. 33) within historical times is a 

 most important and interesting question, and it has within recent years provoked much dis- 

 cussion. Dr. Albert Giinther (Proc. Linnean Soc. Lond., ex, 1898, p. 26), as an appendix to 

 his presidential address on this subject, has given a list of 51 papers which have a bearing on 

 the question. The most difficult question to settle is how the ancestors of the gigantic tortoises 

 which once abounded on the Galapagos Islands and on various groups in the Indian Ocean 

 reacht those islands. Originally it was assumed that these ancestors were accidental arrivals, 

 borne thither by oceanic currents; but in 1891 Dr. George Baur (Amer. Naturalist, xxv, 

 pp. 217-229; 307-326) promulgated the theory that these islands were only the volcanic peaks 

 of a great tract of land which had at one time connected these islands with one another and 

 with the western coast of America. In a number of papers succeeding this announcement he 

 defended this theory. For Dr. Baur's views the reader is referred to his writings, a list of which 

 was publisht by Dr. W. M. Wheeler in 1899 (Amer. Naturalist, xxxni, pp. 23-30). 



The present writer does not intend to enter upon a full discussion of the question, but he 

 records it as his conviction that in general terms Dr. Baur's position is correct. Probably the 

 simplest way to account, on the old theory, for the presence of these tortoises on these remote 

 islands is to suppose that, some time in the distant past, a gravid female was borne by currents, 

 swimming or floating on a raft of fallen timber; that she landed on one of the islands; and that 

 her descendants were in some similar way distributed to the other islands of the group. But 

 these Testudinidae are probably of all turtles the least adapted for such transportation. In the 

 water they would almost certainly drown within a few hours. Nor is it probable that a raft of 

 wood would hold together long enough or well enough to carry one of these turtles, wholly incap- 

 able of clinging to such objects, the distance of 700 miles. If this almost miracle of transpor- 

 tation had been accomplisht, it must have been repeated on a smaller scale many times, in 

 order to populate the various islands. And all this work must have been accomplisht long ago, 

 and long ago have ceast; have been accomplisht long enough ago to permit the evolution of 

 a distinct species on each island, and have ceast long enough ago that new arrivals should 

 not have disturbed this differentiation. It is difficult to imagine why the distribution of these 

 turtles between the various islands should not have continued, if it had ever been possible. 



Probably the chances for lizards to reach the Galapagos Islands from the American coast 

 and from one island to the other are much better than those favoring the turtles. Yet we find 

 the same variation among the lizards of the islands that we find among the turtles. Four 

 genera are represented on these islands, of which two, Tropidurus and Phyllodactylus, are 

 represented on the American coast. Two others, Conolophus and Amblyrhynchus, are peculiar 

 to the islands in question. It seems that the opportunities for the accidental transportation of 

 lizards are to-day as good as they could ever have been, yet no species are common to the 

 American coast and the Galapagos Islands. We would be justified in expecting something 



