1905.] NATURAL SCIENCES OF PHILADELPHIA. 167 



maturation mitosis. The other chromosomes have become long and 

 thread-hke, and an irregular nucleolus (A''.) has appeared. Following the 

 stages of figs. 14 and 15 is reached a complete rest stage (fig. 16), with 

 the chromatin globules finely distributed along the linin threads — the 

 nucleus very similar in appearance to that of spermatogonia in the 

 rest stage, except for the presence of the large heterochromosome. A 

 rest stage preceding the synapsis I have never before found in any 

 object, but it has been described for Ascaris and certain other* forms. 

 The heterochromosome is still nearly straight, and when viewed from 

 the proper angle shows not only a transverse constriction, marking the 

 point of junction of the two univalent ones, but also a longitudinal 

 split in each of the latter (figs. 15, 16). In later stages of the spermato- 

 cyte these characteristics of the heterochromosome cannot be distin- 

 guished, and from a study of the later stages alone one might easily be 

 misled to the conclusion that the heterochromosome of the spermato- 

 cyte were a univalent element. 



Next the chromatin reticulum segregates into short loops, very much 

 convoluted and occasionally simulating longitudinal splittings (Plate 

 IX, figs. 17, 18). But a long study of cells in this period shows that the 

 space between two mutually wound loops is not a longitudinal split, 

 and that the latter, i.e., a splitting into two of each chromatin globule, 

 along the length of a loop, rarely ever commences so early. On the con- 

 trary the double loops represent pairs of univalent and correspondent 

 (homologous) chromosomes, so that this stage is the commencement of 

 the conjugation into pairs of the eighteen ordinary chromosomes; this 

 becomes the more obvious on comparison with subsequent conditions. 

 Now, also, the heterochromosome commences to bend at an angle at 

 its middle point, on its path from the earlier straight form to its later 

 one of a nearly closed V. 



This leads, the chromosomes becoming much longer (Plate IX, fig. 

 19), to the synapsis stage (figs. 20-22) ; throughout this stage the nuclear 

 membrane is almost or quite imperceptible, and the chromatin loops in 

 the form of irregular U's and V's, crowded most densely at that nuclear 

 pole (the distal) next the greatest amount of cytoplasm. In all the 

 drawings only a few of these loops are shown, mainly those seen dis- 

 tinctly for their entire length. Their relative lengths differ greatly in 

 size, as is to be seen particularly in fig. 21 . Each such loop is a bivalent 

 chromosome, for they are nine in number, corresponding to the nine 

 pairs of ordinary chromosomes of the spermatogonia, and therefore 

 each arm of one is a univalent chromosome. Two univalent chromo- 

 somes are usually united only by one end, that marked x in fig. 21 ; but 



