VIVIPAROUS FISHES OF THE PACIFIC COAST. 
439 
these are based on actual observations differing from theirs, are probably all due to 
a modification of the process in different fish. The present function of the periblast 
nuclei is the appropriation of the yolk for the blastoderm — a view well borne out by 
the reduction of these nuclei with the reduction of the yolk to a mere vestige in 
Cymatogaster. 
At the closing of the blastopore the periblast consists of about 12 (10 to 18) large 
irregular nuclei embedded in the cortical layer of the yolk. There is no layer of proto- 
plasm connected with these nuclei. They are usually grouped in pairs and sometimes 
are almost entirely confined to that portion of the yolk which originally was not cov- 
ered by the blastoderm. Its origin is similar to that in Ctenolabrus as described by 
Agassiz and Whitman. It becomes, however, an independent layer much earlier than 
in Ctenolabrus where the nuclei do not become independent till close of segmentation. 
Not nearly all the marginal cells of the blastoderm take part in its formation. “At the 
end of the seventh segmentation (fig. 44) some of the marginal cells have the lower 
margin ill-defined and continuous with the yolk. This is never the case in more cen- 
tral cells. The nucleus is also slightly more refringent than in the neighboring cells. 
At the end of the eighth segmentation (fig. 45) there is still a distinct dorsal wall to 
the periblast cell, but its nucleus is now much larger than that of the surrounding 
cells.” In shape it does not differ greatly from the surrounding cells. It is thus seen 
that before the nuclei are well separated from the blastoderm they begin the change to 
the characteristic of later stages of other teleosts. “At the end of the ninth segmenta- 
tion (figs. 46 to 51) the periblast is an independent structure consisting of a few large, 
refringent nuclei embedded in protoplasm which is restricted to the immediate neigh- 
borhood of the marginal cells of the blastoderm.” Exteriorly these cells are partly 
covered by the epidermal cells, dorsally they are covered by the cells just within the 
epidermal cell. The plasmodium of the periblast does not extend beneath the central 
cells of the blastoderm and the nuclei do not reach this region till much later, if at>all. 
There is never a continuous layer of protoplasm at the base of the blastoderm. In 
fig. 51 , of an egg sectioned horizontally, there are 10 periblast nuclei, one of them 
dividing and all of them lying just below the marginal cells of the blastoderm. This 
egg is near the end of the ninth segmentation, containing about 450 nuclei. Up to this 
stage the nuclei retain their spherical outline, but from this on they increase in volume 
and lose their regular contours and the surrounding protoplasm is greatly reduced or 
disappears entirely (figs. 38, 40, 41, etc.). 
The fact that the nuclei are very frequently in pairs would indicate that they 
undergo division, and indeed some sections show this to be the case. It is, however, 
doubtful if more than two generations of nuclei are produced in this manner. 
At the end of cleavage (fig. 29) there are several series of flat cells covering the 
greater part of the yolk. Beneath them the periblast nuclei lie. In Ctenolabrus , at 
the end of cleavage, the periblast forms a wreath of flattened cells. 
The exact number of nuclei arising directly by a change of the cells containing 
them I was unable to determine, but their ultimate number rarely, if ever, reaches 20. 
Reduced to such simple conditions and differing as the nuclei do from the nuclei of 
the blastoderm even before they are entirely free, I can state with the greatest confi- 
dence that these nuclei have no share whatever in forming the embryo or in giving 
rise to even a single cell of the embryo. The number of periblast nuclei probably 
remains the same from the closing of the blastopore to the final disappearance of the 
