No. i.] MORPHOLOGY OF VERTEBRATE HEAD. 8 1 



the same stage of development. These median cavities are 

 probably the homologues of the anterior cavities in Torpedo, 

 recently described by Dohrn (No. 8), in which he thinks he sees 

 the rudiments of several lost somites. I find in Acanthias little 

 evidence to favor such a view. The median cavities lack defi- 

 nite walls, are irregular in number and in outline, and are 

 surrounded by tissue which appears to be disintegrating rather 

 than passing through a process of normal development. Sooner 

 or later these median cavities fuse with one another and with 

 the posterior of the paired cavities, forming together the pre- 

 mandibular cavity. 



Kastschenko (No. 9) says that the entire premandibular cavity 

 arises " aus der . . . praechordalen Hohle, welche einer ganzen 

 Reihe verschiedener Umgestaltungen unterworfen wird," and 

 that the resulting cavity cannot be homologized with a true 

 head somite. Views so at variance as those of Kastschenko, Van 

 Wijhe, and Dohrn show that the true significance of the pre- 

 mandibular cavity is still extremely problematic. For while 

 Kastschenko claims that the cavity is non-somatic, Van Wijhe 

 thinks it is formed by the fusion of one pair of somites, and 

 Dohrn by the fusion of many. 



Of the paired lateral cavities above described, the anterior 

 are first to appear. They are found shortly before the breaking 

 through of the first visceral clefts, and their appearance is im- 

 mediately followed by that of the lateral portions of the pre- 

 mandibular cavity, after which the median cavities arise. The 

 fusion by which the entire premandibular cavity is formed takes 

 place just after the first visceral clefts appear. 



I have spoken of these cavities as arising from the mass of 

 cells in which the notochord ends. This is true ; but at the 

 time when the cavities appear, the original mass of cells in 

 which the chorda ended has been increased by the addition of 

 cells proliferated from the walls of the mandibular cavity. 

 Exactly what part of the resultant tissue is derived from each 

 of these two sources, I am unable to say. I believe, however, 

 that the lateral cells in which the paired cavities arise, are in 

 great part, if not entirely, the result of proliferation from the 

 walls of the mandibular cavity. 



Although the anterior pair of cavities, and the lateral portions 

 of the premandibular cavity, arise independently in Acanthias, 



