360 JOHNS HOPKINS UNIVERSITY MORPHOLOGICAL MONOGRAPHS. 



undoubted answer would be " that found in the most widely separated 

 species." This would indicate that the gland arising from the peri- 

 branchial chamber is the more primitive, since it is found in one of the 

 ascidians (Phallusia mammillata) and also in salpa. Prof. Herdman, in 

 his paper already referred to and in other more recent papers [5], makes 

 an interesting suggestion that " if the hypophysis cerebri was, as E. van 

 Beneden and Julin think, a renal organ in the anterior part of the body 

 of the primitive chordata, its ducts supposing it to be formed of several 

 pairs of nephridia would originally open on the sides of the body, and 

 might have then become implicated in the sinking-in of the epiblast to 

 form the atrial involutions, and so would come to open into the peri- 

 branchial cavity (a condition found in some ascidians)." He says, "I 

 would regard then the connection between the duct, or one of the ducts, 

 of the subneural gland and the dorsal tubercle as being secondary." 

 According to this hypothesis, Phallusia mammillata shows a tran- 

 sitional phase; most ascidians have passed beyond and have lost the 

 peribranchial portion of the gland. Salpa would then have reverted 

 to the condition found in one of the more primitive ascidians having 

 a well developed peribranchial gland. It is probable, however, that 

 the presence in salpa of the ventro-lateral outgrowths from the brain 

 toward the hollow disks (of the peribranchial gland) indicates the par- 

 tial retention of a process similar to that by which the subneural 

 gland of ascidians is formed at the expense of the thick ventral wall 

 of the visceral portion of the larval neural canal. The innervation 

 of the large-celled pair of these outgrowths (Plate L, Fig. 4, &') may 

 indicate that they were formerly, if not at present, functional as ganglia 

 in connection with a well developed gland which arose partly from the 

 wall of the peribranchial chamber and partly from the nervous system. 

 On the other hand, if a gland of neural origin be the more primitive, 

 Phallusia mammillata would still represent the transition from this con- 

 dition to the gland derived from the wall of the peribranchial chamber. 

 I incline to the former hypothesis; but, be this as it may, salpa has 

 retained or reverted to the condition found in a transitional form like 

 Phallusia mammillata, since it shows traces of a gland of double origin. 

 Molgula ampulloides has two well developed lateral chambers, one on 

 each side of the gland, which open through the duct of the gland into 

 the pharynx by way of the ciliated funnel. These are of interest in this 

 connection, for they may correspond to the hollow disks of salpa's gland. 

 Molgula ampulloides would then represent a second transitional form 



