THE FOEMATION OF THE LAYEES IN AJIPHIOXUS. 285 
tions I liave just summarised. In it I denied that there was 
any histological difference between the cells forming the 
dorsal and those forming the ventral wall of the archenteron. 
I denied, also, that tliere was any inflection of ectodermal 
cells round the dorsal lip of the blastopore, but I asserted 
that the gastrulation was due, in the first instance, to a rapid 
increase of endoderm cells in this neighbourhood, which pro- 
duced a lateral strain on the endoderm to which it yielded by 
invagination, that the blastopore was directed posteriorly, not 
dorsal ly, and that the closing of the blastopore, or, to speak 
more correctly, its reduction to a small pore, was due to the 
appearance of a new centre of rapid growth in the ventral lip 
of the blastopore. I asserted further that the first pair of 
somites were independent formations from the gut-wall, which 
I compared to the collar-cavities of Balanoglossus, and 
that the other somites owed their origin to the segmentation 
of a hinder pair of outgrowths from the gut-wall, which 1 
compared to the trunk-cavities of Balanoglossus, and 
w’hich grew in len gth pari passu with the growth of the 
animal, and that only the hindermost as yet unsegmented 
portions of these grooves retained openings into the gut. I 
further asserted that whilst the other somites divided into 
myotonies above and ventral portions below, which fused to 
form a continuous splanchiiocele, this was not true of the first 
pair (the collar-cavities J. In these the dorsal myotomic por- 
tion remained in open communication with the lower thin- 
walled portion of the somite. This latter did not fuse with 
its successors to aid in forming the splanchnocele, but on the 
contrary extended posteriorly outside the splanchnocele form- 
ing the basis of the future atrial fold. In a subsequent paper 
( 27 ) I corrected a hasty and ill-founded statement made in 
the first paper, that in the spherical blastula the blastomeres 
were all equal in size, and I showed that the backward 
extensions of the atrial cavity became solid, and apparently 
furnished the material for the transverse muscle of the atrial 
floor. 
In the same year that this second paper Avas published, a 
