MEMOIRS OF THE NATIONAL ACADEMY OF SCIENCES. 81 



proliferated from the anterior end of the archenteron, and the power "f producing the same 

 seems to diminish gradually toward the posterior end of the blastopore lips in those gastrula 

 where the round blastopore is just beginning to close up (figs. li </. I>. e, d, e,f). 



The mesoderm cells are \ ei'y amoeboid in character and are often seen in living specimens and 

 sometimes in sections sending out long pseudopod-like prolongations, which become attached to 

 the walls of the gastrula. l'>\ means of these amoeboid movements thej are able to crawl up the 

 walls of the blastocoele. 



We were unable to make out any structure in Phoronis architecta which could be interpreted 

 as "archenteric diverticula," such as figured by Caldwell (3a) and [keda (9). Fig. 16c might be 

 interpreted as showing these diverticula, but the condition there is hardly different from the 

 arrangement of the mesoderm cells, which are being pushed out into the blastocoele in front of 

 the blastopore (figs. L6, !'>/'). Caldwell first observed these structures in the gastrulse of Pho- 

 ronis kowalevskii, hut Lone-champs (12), who has recently carefully' studied the same species, has 

 been unable to rind them. Ikeda (9), however, rinds very definite diverticula in the gastrulse of 

 Phoronis ijimai, but he figures them as being in the region of the blastopore, while, according 

 to Caldwell's (3a) figures, they are found posterior to the blastopore. 



Let us return again to the mesoderm cells which lie anteriorly to the blastopore. These 

 amoeboid cells undoubtedly multiply while in the blastocoele, and in a gastrula where the blasto- 

 pore lips have closed up somewhat so as to give an oval outline to the blastopore (tie-. 18/) these 

 cells have become arranged into a definite sac (figs. 19, 20a), which i- later to form the lining of 

 the preoral lobe. In no case were we able to find the least indication of an anterior unpaired diver- 

 ticulum, which Masterman (Hi) says exists in the gastrula of l'h<>r<>nis l>uskii. At this stage 

 the cavity of the sac is small and is present only in front of the blastopore. The walls, however, 

 are extended on each side into a lateral cord of mesoderm cells, which lies in the blastocoele at the 

 side of the blastopore (fig. is/'). Some of the cells of the dorsal wall of the sac send out pseu- 

 dopodia, which attach themselves to an ectodermal thickening, and this thickening will become 

 the ganglion of the Actinotrocha (tig. L9). 



The above condition continues until the oval blastopore becomes smaller and round in outline 

 (figs. 20, 20< ). which change is also accompanied by further growth of the enteron in a posterior 

 direction until it almost touches the end of the larva. The cells of the two lateral cords of 

 mesoderm have now increased in number, have arranged themselves so a- to inclose a cavity, 

 continuous with the cavity of the anterior one described above, and have become attached both 

 to the lateral ectodermal wall and the lateral endodermal wall (tig. 205). Anteriorly this sac, 

 which is now horseshoe shape (tig. 20e), is still only attached to the ganglionic thickening and 

 the ventral ectodermal wall (tig. 20). The conditions just described are not due to the shrinkage 

 of the mesodermal lining away from the wall of the larva, for the transparency of the living 

 larva makes it possible to see the formation of the mesodermal sac. We have followed this 

 formation step by step many times in the living gastrula and larva, as well as in sections and 

 surface mounts. 



As tlie anterior end of the larva bends farther ventrally and becomes a definite preoral lobe, 

 the round blastopore assumes the shape of an oval with its major axis transverse to the long axis 

 of the larva (tig. -_'l). The posterior part of the larva increases in length and the enteron sends 

 out a posterior diverticulum, the beginning of the intestinal canal, whose blind end fuses with 

 the ectoderm of the posterior end of the larva. The walls of the mesodermal sac become applied 

 to the walls of the preoral lobe more generally, thus forming a definite mesodermal epithelium 

 lor the cavity of the preoral lobe digs. _':;. 21, 22). Posteriorly, as Masterman ( 1."-) ha- described 

 for the fully developed Actinotrocha, the cavity is produced "into two horn- running back 

 laterally" (rig. 22a), but as yet there is no complete mesodermal lining in the cavity back of this 

 (figs. _'_' I. c). The posterior wall of the mesodermal lining of the preoral lobe forms a definite 

 septum (figs. 21, 21a), but it is not as yet, at least, composed of two layer-, a- Masterman finds 

 in the older Actinotrocha. 



