42 H. M. BERNARD. 



We conclude^ tlien^ that in all youug amphibian eyes, in 

 which the rod layer seems to consist mainly of cones ending 

 at some distance from the pigment, the apparently vacant 

 space between the truncated cone-tips and the pigment is, in 

 life, filled up by a compact mass of swollen vesicles. These 

 vesicles are, however, so exquisitely delicate that the process 

 of fixation and hardening destroys them almost completely. 

 But I should add that now that I have seen the vesicles in 

 the Table Mountain specimens, I have been able to discover, 

 in sections of our native forms, several cone-tips running out 

 into faint divergiug threads. 



Another peculiarity in the retinas of these Table Mountain 

 tadpoles deserves mention. In Part I, p. 34, I remarked in a 

 note that the only long-necked elements which I could find in 

 froo-s' retinas at all resembling the long-necked ''cones" figured 

 by van Genderen Stort (' Quain's Anatomy,' 10th ed., vol. iii, 

 part 3, p. 48) were those which appeared in each case as one of 

 the so-called twin or double cones (see PL 3, fig. 5). Besides 

 these, the only elements with long inner limbs were Schwalbe's 

 rods, in which the refractive globule had, as a rule, already 

 disappeared and the outer limbs had already become 

 cylindrical (Part I, PI. 3, fig. 4, r). But in the retinas of 

 these Table Mountain tadpoles, cones with striking refrac- 

 tive globules like those figured by van Genderen Stort are 

 very plentiful, close down against the pigment layer. The 

 greater toughness of the walls may account for the persis- 

 tence of the shape in a phase where it is quite lost in our 

 native forms (see the phases Part I, fig. 4, c^ and r). The 

 fact that the refractive globule is not so quickly absorbed 

 may be referred to the great quantities of pigment to be 

 dealt with (for the origin of this globule see Part II, p. 463). 



The transformation of these long-necked cones into rods is, 

 in some cases, very easy to follow. The conical portion 

 thickens and shortens, while the swollen vesicle at the tip 

 becomes cylindrical and the refractive globule disa])i)ears. 

 In fig. 9 elements like those on the right and left hand are 

 very comuiou ; that on the right shows a division in the 



