2g6 E. MITSUKURI. 



the invagination cavity becomes exactly co-extensive with the epi- 

 blastic embryonic shield, before that act takes phice (II Art. figs. 6fo, 

 7h, 8h). In Platy dactyl us it is said to be only slightly less. In 

 Chelonia caouana, which I have studied, the invagination cavity 

 breaks through, as I have stated in the foregoing article, when it is 

 quite small compared with the epiblastic shield. Will accounts for this 

 discrepancy by assuming that Cistudo and Platydactylus on 

 the one hiind and Chelonia on the other are really different in this 

 respect. (7/ Jrt., Nach sell rift. Also in a note " U. d. iTüstndation 

 r. Cistudo a. Chelonia,'' Anal. Anz., Ill J Jahnj., No. ltl/19). 

 In Tri onyx, I possess several embryos which are like Fig. 11). of 

 Contrib. I. or Fig. (3 of the foregoing article, so that I think I am 

 justified in concluding that Trionyx is like Chelonia in this 

 respect. In Clemmys there seems to be individual variations as 

 to- this point. For instance, if we comj^are Fig. A. in this postscript 

 with that given in Fig. 1 of my Contril). Ill, Ave find that in the 

 latter, the inviigination cavity must have advanced farther forwards, 

 nearer the anterior end of the embryonic shield than the former. iSo 

 that it is an actual fact that there are variations in different species 

 ov within tVie same species in the proportion of the invagination cavity 

 to the shield. For the j)resent, I am therefore willing to accept 

 Will's assumption as the correct explanation of the disagreement be- 

 tween his statements and mine. And yet I can not help having some 

 doubts lurking in my mind that his Figs. 6/>, 7/>, and <S/> (II Art.) are 

 expressions of something other than the breaking through of the in- 

 vagination cavity. That in Cistudo the invagination cavity be- 

 comes both in length and breadth exactly coextensive 

 with the embryonic shield — not one whit more or 

 less — seems to me very extraordinary. The figures Will copies 

 from Clarke do not certainly show the laterjil extension of the in- 



