NO. 1431. BREEDING HABITS AND EGG OF PIPEFISH— GIWGER. 475 



188(i), and others, which do cut all the way through. Agassiz and 

 Whitman (1880) show that in Ofe/iolahn/s the first furrow may or may 

 not penetrate to the yolk. There is never any such under furrow as 

 the bass and Ctenolahnix .show in the tirst division. 



The eggs are laid at night, as early as 10 o'clock, and probal)ly at 

 any hour thereafter. At any rate, by 7 o'clock the next morning, 

 they are to be found in stages of from two to sixteen cells. Probably 

 from four to six hours elapse before they l)egin to segment, since it 

 takes this long for the germ disk to form on eggs in water, in com- 

 parison with six and one-fourth hours for the herring (Brook, 1887) 

 and twelve to thirteen for the salmon (Hoffmann, 1888). 



FOl'R-CELLEI) STAGE. 



In tig. -t, Plate V, is shown a normal four-celled blastoderm. The 

 second furrow is horizontal and crosses the tirst approximately^ at right 

 angles. Thus there is formed a four-celled symmetrical blastoderm. 

 Sections of this would in no wise differ from those for two-celled stages, 

 save in the plane a-h, where the l^eginnings of the segmentation cavity 

 and the central periblast would be found. Such a section is not at 

 hand, unfortunately. 



Fig. 5, Plate V, a more common form, shows slight inequalities in 

 the size of its blastomeres. Such irregularities become more pro- 

 nounced until they result in reniform blastoderms, as fig. 6, Plate V. 

 Fig. 38, Plate VII, is a nearly horizontal section through the base of 

 such a form as tig. -1, Plate V. The wide separation of two of the cells 

 is an artefact. Of special interest are the segmentation cavitv in the 

 center and the remnants of protoplasmic bridges which connected the 

 blastomeres. 



EIGHT-CELLED STAGE. 



Into the blastoderms of the pipetish Qgg of this stage, many very 

 great and seemingly irreconcilable irregularities enter and greatly 

 confuse the investigator. These were tirst noted on living eggs with 

 four and eight cells below, two, three, and four above. Hardened 

 eggs showed the same irregularities. Surface views of a great many 

 of these eight- to sixteen-celled blastoderms were drawn. When a 

 comparison of these drawings was made, they were found to conform 

 to four general types. This was conffrmed by an examination of all 

 the eggs of this stage which had been preserved. At the close of this 

 section, there is appended a table showing the relative numbers of 

 these various types. 



In tig. 7, Plate V, is shown the normal type of 8-celled teleost 

 blastoderm. It is formed by two furrows nearly parallel to the first 

 and perpendicular to the second plane of segmentation, dividing such 

 a form as fig. 4, Plate V, into eight blastomeres. In this blasto- 



