712 PROFESSOR W. C. M‘INTOSH AND MR E. E. PRINCE ON 
endodermic cells. When at its maximum it is a slightly flattened dome-like cavity 
(Pl. II. fig. 155, ge); but with the extension of the blastoderm its roof is depressed, 
and it thus appears subsequently as a mere fissure. Now LrREBOULLET figures his 
cavity as a narrow fissure extending almost from margin to margin of the blastoderm ; 
whereas BAMBEKE’S is a compact, but loftier and more spacious chamber.* It is 
noteworthy that BamBeKE was struck by this dissimilarity, and after examining the 
segmentation-cavity in the roach was prompted to seek for a germinal cavity underneath 
the blastoderm, and found one, as he indicates in his figs. 4 and 6 (vide No. 20a); 
but he adds that “a comparative examination of preparations forces me to regard it as a 
simple accident and artificial, for the prominences and depressions of roof and floor 
coincide.” There is much reason to suppose, therefore, from the shape and nature of 
the floor, that LEREBOULLET’s cavity is not a segmentation-cavity, such as BAMBEKE 
supposes, and, if this be so, then LeresoutLer likewise discovered this flattened 
germinal cavity, as E. van BrNEpEN says (No. 25, p. 47), though this author is wrong 
in according the discovery also to VAN BamBexe. If, on the other hand, LerEBouLLEt’s 
be really Von Barr’s (and Van Bampexkr’s) cavity, then H. Raruxe first signalised the 
germinal cavity in Zoarces ; and he was followed by SrrickEr. It is therefore not correct 
to speak of a cavity of LerEBOULLET with Van BeNEDEN but rather of a (sub-blastodermic) 
germinal cavity, which is persistent through all embryonic life, as distinct from the (intra- 
blastodermic) segmentation-cavity which wholly disappears.t 
What then is the significance of the germinal cavity thus distinguished? By the 
fact that its floor is formed of yolk, or rather the protoplasmic cortical film (or inter- 
mediary layer), and that it is roofed over by endoderm (lower layer) and epiblast-cells, it 
is comparable to the ‘ Keimhéhle” in the fowl’s ovum.{ At a later stage the hypoblast- 
cells which intrude from the periphery to form the blastodermie rim (bv) and shield 
(Pl. IL. figs. 15, a-e, and 17) do not pass across the floor of the cavity, but creep 
up the sides and partially arch it over, forming in fact a gastrula which would open ex- 
ternally by the blastopore, were not this aperture plugged up by the mass of yolk (really 
Ecker’s plug), which is so large that the invaginated lip is compelled to pass round, and 
epibolically envelop it. The germinal cavity, arched over as it is by the thick blastodermie 
roof, bdm (Pl. IL. fig. 15, a-e), is never truly open in the sense indicated; but potentially it 
is so, the removal of the concentrated trophic matter (y) which does not segment would leave 
the blastoderm a simple gastrula—indeed, as RypER remarks in regard to Alosa, that 
“the yelk might be removed at any stage without taking away any essential part of the 
embryo except the floor of the cavity” (No. 141, p. 569). Van BaMBEKe does not hesitate 
to regard his chamber as a gastrula-cavity, and finds in it therefore great phylogenetic 
* A glance at LEREBOULLET's figure (No. 93, pl. iii. fig. 3) and BamprKke’s (No. 20a) sufliciently shows this. 
t+ See a paper “On the Significance of the Yolk in the Eggs of Osseous Fishes,” by E. E. Prrxce, Ann. Nat. Hist., 
July 1887. 
+ It is interesting to observe that, with the appearance of the germinal cavity, the thick periblast-floor in some forms 
becomes thinner, The Keimhdhle or germinal cavity is often called the segmentation-cavity in the fowl’s ovum. 
