Description of the Notochordal Invagination. 9 



evidence for this conclusion will appear later. All other series of 

 sections in my possession disclose only one layer of entoderm 

 below the notochordal canal. On the posterior floor of this canal 

 (Plate VII. Fig. 32) is situated that solid mass of cells which is 

 described as the commencing mesoderm. Anteriad this compact 

 mass gradually passes into a vacuolated condition, as is seen in 

 sagittal sections (Plate VII. Figs. 32-34). I have repeatedly 

 observed in cross sections this same vacuolated condition, and 

 have figured it in Plate VIII. Figs. 36 and 37, and in Plate X. Figs. 

 49 and 50. 



Another point of contention between Will and Mitsukuri con- 

 cerns the length and width of the blastoporic canal at the time of 

 its opening toward the yolk. Views of the ventral surface of two 

 embryos — Chelydra serpentina (Plate II. Fig. 10') and Chrysemys 

 picta (Plate I. Fig. i' ) — illustrate the variability in the widtJi of 

 the canal in different genera. Transverse sections of the canal of 

 the embryo of Chelydra serpentina (Figs. 10 and 10') show that its 

 width nowhere exceeds one half that of the shield, whereas in 

 a transverse section of the Chrysemys picta embryo (Plate IX. 

 Fig. 45) it is seen to be almost coextensive with the width 

 of the shield. If now we compare surface views of two shields 

 of Chrysemys picta (Plate 1. Figs, i and i', and Figs. 3 and 3'), 

 we shall at once see that great variability in regard to the width 

 of the notochordal canal exists even in the same species. Chely- 

 dra serpentina also seems to show some variability in this respect, 

 as a comparison of Figures f and 10' (Plate II.) will show. 



Concerning the length or anterior extent of the canal, I am able 

 to give evidence for three genera of American turtles ; Chelydra 

 serpentina, Chrysemys picta, and Ozotheca odorata. I have 

 already had occasion, in considering vacuolated entoderm, to call 

 attention to a series of sagittal sections of an embryo of Che- 

 lydra serpentina (Plate VII. Figs. 32-34). Figure 32, which passes 

 through the axis of the embryo, gives little information in regard to 

 the length of the canal, for the whole anterior part of the floor of 

 the canal has already disappeared. A view of the ventral surface 

 of this embryo (Plate II. Fig. 7'), however, gives one the impression 

 that all of the lateral portion of the canal has not opened below. 

 If we examine a parasagittal section made along the inner margin 

 of this lateral unbroken area (Plate VII. Fig. 33), we find that the 

 posterior ventral thickened area extends much farther anteriad than 



