474 Dr. Branislav Petionievics oti the 



basal axonost as a sopaiate postaxial radial (in this respect 

 tiie restoration of Broom is oxact). 



Now I consider the posterior bifurcation of the fourth 

 axonost in our specimen as of exceptional importance for the 

 question of homologies. As the pelvic fin of Eustkenopteron 

 is far more reduced than its pectoral fin (comp. fig. 1 of 

 pi. xvi. in Goodrich, 1902, which shows that there is no 

 fourth axonost in tiie pelvic fin — British Museum specimen 

 P. 6704 — and no postaxial processes), we must infer that the 

 paired fins of EnstJieno2yteroti represent a stage far in advance 

 of that stage of the paired fins in its ancestors, whicli was 

 the starting-point for tlie evolution of the paired limbs in the 

 primitive ancestors of the Tetrapoda*. If this inference is 

 a right one, then it is not improbable that the posterior 

 bifurcation of the fourth axonost in our specimen is a remnaiit 

 of a more primitive stage when the fourth axonost was com- 

 posed of two separate ossifications, the paired tins of Eusihenu- 

 pteron being evidently the reduced archipterygium-type of 

 Gegenbaur (a resemblance recognized by Woodward, Tra- 

 quair, and others). So that we have to conclude from this 

 evolution that the axis of the tetrapod lij)ib runs along the 

 humerus, uhia, nlnare, and between the fourth and fifth finger f 

 (comp. text-fig. 2, in which some further hypothetical homo- 

 logi(!3 have been indicated). This conclusion, as one sees, 



* This conclusion is confirmed also by the skull, which in Hnstheno- 

 pteron is simpler tiiau in the more primitive Osteolepidns, whose paired 

 fins are also less reduced ^conip. the tins of iMct/alir/if/ii/s figured by 

 Ed. D. Wellburn in his paper " On the Genus Meyaliclithys" in Proc. 

 YorksliireGeol. & Polytechnic Soc. vol. xiv., 1900). 1 may add in this 

 connexion that the skull of Osteolepis may be considered to approach 

 uearer to the Stegocej halian skull than is sliown by the restoration of 

 Pander (comp. Chr. II. Pander, ' Ueber die Saurodipterinen, &c.,' 1860, 

 pi. i. tigs. 8 & 9), lately reproduced by Gregory (comp. Gregory, 1915, 

 tig. 2, A, B). Pander's restoration was founded on the specimen of 

 Osteolepis microlepidotus figured by him in pi. i. fig. 1 ; but tig. 4 on the 

 same plate represents a specimen in which all the three cliav;icteristic 

 bones of the Stegocephalian skull (supratemporal, intertemporal, po.st- 

 orbital) are present. 



t The pectoral fin of Sauripterun taylori (figured and restored by 

 Gregory, 1915, plate iv. and fig. 9) does not militate against this supposi- 

 tion. This fin, less reduced than that of Ensthcnojiteron, has three 

 elements attached to the third axonost, so that these three elements may 

 correspond with the three digits on the ulnar side of the tetrapod limb. 

 As the two outer of the?e three elements have almost the same lengtli, 

 it may well be supjjosed that the axis runs between the two (and not 

 along the outer one alone, as Gregory hypothetically supposes — comp. 

 Gregory, 1915, p. 3ti0). I sliouid mention that the iir.st to emjihasize 

 the resemblance of the ^ijnuripterus-^n with the tetrapod limb was its 

 discoverer, .lames Hall hiniseh" (comp. J. Ilall, ' Geology of New York,' 

 l)art iv. 1843, p. 2S2). 



