The Germ-cells of Lophius. 577 



nearly as could be determined, the cases figured represent average 

 conditions. Usually most of the germ-cells are grouped in a 

 more or less compact mass, with a few scattered ones in front of 

 this group and others distributed for some distance from the main 

 mass in all embryos studied (figs. 17 and 18). It would be of 

 interest to know whether these scattered cells take part in the 

 formation of the gonad or whether they degenerate. 



Between stages 8 and 13 there is a gradual increase in the antero- 

 posterior extent of the germ-cells. This is correlated with the in- 

 crease of length of the embryo, or rather with the growth of the 

 coelomic mesoblast in which they lie. The elongation of the seg- 

 mental structures does not have any influence. In all embryos, the 

 anterior extent is about co-incident with the forward limit of seg- 

 mentation, and the posterior not far forward of the termination 

 of the gut. 



It should be noted in this connection, that the germ-cell group- 

 ings on the two sides of the embryo is not symmetrical. The 

 groups of one side do not necessarily correspond to those on the 

 other. 



3. Number of Germ-cells 



In order to find if there is in Lophius the period of rest from 

 division of the germ-cells common in vertebrates, a careful tabu- 

 lation of the germ-cells in embryos of each age was made. Though 

 their number in Lophius is small, to make an accurate count in 

 any embryo is not easy. In the early stages especially the danger 

 of overlooking some of them is very great. There is also, when 

 the nucleus is cut and appears in two adjacent sections, the dan- 

 ger of counting the same cell twice. To avoid this danger, camera 

 lucida drawings were made of each section in which there were 

 germ-cells. These drawings were made on transparent paper 

 so that drawings of adjacent sections could be superimposed, and 

 thus corresponding halves of nuclei properly related, and the 

 danger of error in counts made by this method is, I believe, small. 

 After making a few counts by this method it appeared that counts 

 made previously by less careful methods were of no value. 



