332 LVA iL.S' xrnv york \r\r>r:MY of scfEXCES 



margin. The tliird segment is also subdivided ventral ly; the attachment 

 of the quintal ganglion (3) extends close to its cranial border. These 

 two ganglia are thus approximated and on the point of union which is 

 ultimately effected in embryos of twenty-six somites. The third seg- 

 ment is followed by a slight prominence (14) confined to the ventral 

 region of the neural tube and less clearly marked off from the myelen- 

 cephalon tlian in the preceding embryo. Above it is the acoustico-facial 

 ganglion (^). The relief of the following neuromeres is faint and to 

 be made out only with great difficulty in this embryo. The model of 

 the lumen (Plate XLI) corroborates the description we have given of 

 the external relief, but in addition presents one or two details which are 

 not perceptible from the surface. The plica ventralis encephali by its 

 broad summit forms the floor of the midbrain. Its anterior angle, the 

 tubereulum postero-superius of authors, juts forward prominently. Im- 

 mediately below is the recess of the mammillary region (17) botmded 

 ventrally by the tubereulum postero-inferius (.2). Ventral to this again 

 is the large triangular cavity of the infundibular region (IS). Its 

 ventral wall shows a moderate thickening, torus postopticus, in front of 

 which is the shallow preoptic recess, from which the lamina terminalis 

 extends forward and upward to a point at which the ectoderm (7) is 

 adherent. 



We have now completed the record of our objective findings on the 

 basis of which we propose to discuss the problem stated in our opening 

 paragraph; we shall endeavor in the 'course of this discussion to compare 

 our results with those of other students only in so far as they have dealt 

 with mammalian embryos of corresponding stages of development. Un- 

 fortunately the number of detailed descriptions of such embryos is not 

 large. We have not, therefore, attempted any general comparison of 

 the ontogeny of the mammalian neuraxis with that of better known 

 forms, except in a few instances when it has a direct and important 

 bearing upon our problem. 



Gaxglioxic Crest 



It has already been stated that the medullary plate primitively lacks 

 a precise boundary and passes by a gradual thinning into the somatic 

 ectoderm (Plate XXII. Fig. 1, 2S). Prior to closure, however, and this 

 is true of the heail as well as the trunk, an abrupt demarcation is estab- 

 lished and the somatic ectoderm joins the medullary plate at its dorso- 

 median angle (Plates XX'\'TIT-XXX). This remodeling of the neuro- 

 somatie juiiclion is jirogressive ce]>linli)-( iuulad. and is eoin])le;ed in each 



