476 PROCEEDINGS OF THE NATIONAL MUSEUM. vol. xxix. 



derm, and in nearly all others of this and the next stage, a considera- 

 ble elongation is noticeable. 



Figs. 8 and 9, Plate V, show variations of this normal type, which 

 are more common than the type itself, but are easily referable to it. 

 Fig. 39, Plate VII, shows a section of fig. 7, Plate V, in the plane c^h. 

 In it one of the two central cells is completely cut out of the proto- 

 plasm, while at the inner end of the cell wall, parti}' cutting out the 

 other cell, there is a little split, which in sections nearer the center will 

 push a short distance to the left, but on the right will extend clea7' 

 across, completely cutting out the cell and extending the segmentation 

 cavity {s. c). The Uwer of protoplasm with 3^olk marked c. p. is the 

 central periblast, and the cavity above it is the segmentation cavity. 

 This, however, is not the first appearance of either, since a section 

 in the plane «-5, in fig. 4, Plate V, would show both. I regret that 

 I have not been able to find such a section. The outer periblast 

 never shows the periblastic ridge figured by Wilson (1891) for 

 Serranus. Fig. 40, Plate VII, is through the plane a-b of fig. 16, 

 Plate VI, a normal sixteen-celled stage, but it will show the state of 

 things in the plane c-d through fig. 7, Plate V. In this part of the 

 normal blastoderm of this stage, the central cells are separated from 

 the periblast by a large segmentation cavity, which extends for a 

 short distance under the peripheral cells, in this case the end cells 

 of tig. 7, Plate V. 



Fig. 41, Plate VII, is a section at right angles to the long axis of a 

 Dlastoderm, similar to fig. 7, Plate V, Here the two cells are separated 

 from each other by a wide segmentation cavity {s. c.) roofed over by 

 a protoplasmic bridge [p. h.) connecting the two blastomeres. A thin 

 split extends for some distance under each cell and partiall}' separates 

 it from the central periblast {e. p.), which is heavily laden with j'olk 

 in its lower parts. Such protoplasmic bridges as the one shown here 

 are not uncommon in this and the next stage. All that can be said of 

 their origin is that they have been left behind when the cells were 

 cut out of the protoplasm. Structures similar to this would be found 

 by making sections at right angles to the long axes of figs. 8 and 9, 

 Plate V. So far as I know, these protoplasmic bridges have not been 

 figured and described before. 



The periblast never comes away freely from the yolk, but is so 

 obscured with fragments of this latter that it has in all cases been 

 drawn semi-diagraramatically, the general course of the break only 

 being followed. 



Fig. 10, Plate V, shows a type of eight-celled blastoderm far more 

 common in the pipefish than the preceding. In this the plane of the 

 third furrow shifts until it becomes equatorial and cuts off four some- 

 what smaller blastomeres from four underlying larger ones. Hen- 

 neguy (1888, fig. 39) shows a blastoderm for the trout which is almost 



