8 Dfi. W. G. KIDiEWOOD oii tSe [Jan. 1 8, 



Stage 3. Specimen measuring 19 mm. from snout to root of fail. 

 Length of tail 29 mm. Length of hind limb, extended, 21 mm. 

 Length of fore limb, extended, 10 mm. (Plate II. fig. 3.) 



The ceratohyals are more massive than before, and have acquired 

 a distinct backward slope, Exauiined from the ventral surface, 

 the two ceratohyals are seen to meet in the median line and to be 

 overlapped (ventrally) by the tapering anterior end of the larger 

 copula, although in the two preceding stages they were separated 

 from one another by a distance equal to one-half of the total 

 width of the pars reuniens. The pars reuniens itself is less 

 conspicuous than before. The anterior copula has disappeared, 

 and the hyoglossal sinus (hgs) thus makes its first appeai'ance. 



The posterior copula is now thicker than the hypobranchial 

 plates ; the two are flush above, but the copula projects ventrally. 

 The future thyrohyals are assuming shape and are thicker than the 

 surrounding cartilage. In fact, the cartilage immediately external 

 to the middle part of the rod is already so much resorbed as to 

 present an incipient foramen (tf), the " thyroid foramen " of the 

 previous communication (9. p. 586). The fenestration does not 

 begin exactly at the region of attachment of the thyroid bodies, but 

 more posteriorly ; the absorption, however, continues in a forward 

 direction and also externally (see figs. 3 and 4). Since both the 

 developing thyrohyals and the copula are thicker than the sur- 

 rounding cartilage, the former appear to be processes of the latter, 

 for the line of junction is no longer to be seen on the ventral 

 surface, and is barely visible above. 



Considerable i-eduetion has occurred in the branchial skeleton, and 

 it is chiefly this which is responsible for the new aspect which the 

 whole hyobrancbial skeleton has assumed. The ceratobranchials 

 are not only thinner, but shorter than before, judging by the 

 diminution in the length of the branchial clefts, so that a shrinkage 

 of cartilage must occur as well as absorption, a fact already pointed 

 out in the case of Pi^m (8. p. lOo) and Pelodijtes (9. p. 588). The 

 distal end of the first ceratob ranch ial has separated from its 

 commissural cartilage, but the second and third clefts still remain 

 enclosed. The spicula have practically disappeared, and the warty 

 outgrowths on the ceratobranchials are mostly absorbed. 



Stage 4. Specimen measuring \S mm. from snout to cloaca. Tail 

 reduced to ^ mm. Length of hind limb, extended, 2\ mm. Length 

 of fore limb, extended, 10 mm. (Plate II. fig. 4.) 



The hyoglossal sinus is wider than before, and the ceratohj-als 

 slope more backwardly and are much more slender, especially at their 

 distal or posterior ends. Here the surfaces of articulation with 

 the palatoquadrate cartilage are no longer distinguishable. The 

 pars reuniens has entirely disappeared, and the two ceratohyals can, 

 in a dorsal view, be seen to unite in the median line. In Stage 3 

 this was only visible ventrally. A central oval area is differen- 

 tiating in the middle of the hyobrancbial skeleton. Its outline, 

 though faint and ill-defined in frojit, is sharply marked behind, and 



