320 THE ORIGIN OF VERTEBRATES 
This latter gland undoubtedly arises posteriorly to the hypophysial 
tube, or Rathke’s pouch (as it is sometimes called), and, as already 
mentioned, is supposed by Kupffer to be formed from the posterior 
wall of this pouch. More recently, as pointed out in Haller’s paper, 
Nusbaum, who has investigated this matter, finds that the glandular 
hypophysis is not formed from the walls of Rathke’s pouch, but from 
the tissue of the rudimentary connection or stalk between the two 
premandibular cavities, which becomes closely connected with the 
posterior wall of Rathke’s pouch, and becoming cut off from the 
rest of the premandibular cavity on each side, becomes permanently 
a part of the ‘ Hypophysis Anlage.’ 
The importance of Nusbaum’s investigation consists in.this, that 
he derives the glandular hypophysis from the connecting stalk 
between the two premandibular cavities, and therefore from the 
walls of the ventral continuation of this cavity on each side. 
This may be expressed as follows :— 
The cceelomic cavity, known as the premandibular cavity, divides 
into a dorsal and a ventral part; the walls of the dorsal part give 
origin to the somatic muscles belonging to the oculomotor nerve, 
while the walls of the ventral part on each side form the connecting 
stalk between the two cavities, and give origin to the glandular 
hypophysis. 
Now, as already pointed out, the premandibular cavity is homo- 
logous with the 2nd prosomatic ccelomic cavity of Limulus, and this 
2nd prosomatic ccelomic cavity divides, according to Kishinouye, into 
a dorsal and a ventral part; and, further, the walls of this ventral 
part form the coxal gland. “Both in the vertebrate, then, and in 
Limulus, we find a marked glandular tissue in a corresponding 
position, and the conclusion is forced upon us that the glandular 
hypophysis was originally the coxal gland of the invertebrate an- 
cestor. As in all other cases already considered, when the facts of 
topographical anatomy, of morphology and of embryology, all com- 
bine to the same conclusion as to the derivation of the vertebrate 
organ from that of the invertebrate, then there must be also a struc- 
tural similarity between the two. What, then, is the nature of the 
coxal gland in the scorpions and Limulus? Lankester’s paper gives 
us full information on this point as far as the scorpion and Limulus 
are concerned, and he shows that the coxal gland of Limulus differs 
markedly from that of Scorpio in the size of the cells and in the 
