MATURATION PHENOMENA OF GERM CELLS. 14! 



similar chromosomes are contiguous in this spirem, since in the 

 metakinesis like chromosomes usually lie near each other ; and, 

 as Fig. 7 shows, they retain their contiguity even in the daughter 

 cells. 



Therefore we find, what Janssens and Dumez insisted that I 

 should demonstrate, twelve pairs of chromosomes of the same 

 length in the equatorial plate of the spermatogonia. 



Now for the proof that such chromosomes unite into pairs in 

 the spermatocytes, an explanation which, according to the Belgian 

 cytologists, " est absolument fantastique et demanderait a etre 

 rigoureusement demontree." On none of my preparations were 

 there stages between the early anaphase of the last spermatogonic 

 mitosis and the synapsis stage of the growth period. In the 

 synapsis stage in Desmognathus the long and slender chromosomes 

 are so intricately coiled together that it is impossible in sections 

 to determine their exact relations. The cell has a distinct 

 polarity, the nucleus at one end, at the opposite the greatest mass 

 of cytoplasm containing the idiozome body. When the chromo- 

 somes begin to separate sufficiently for their boundaries to be 

 distinguished, each appears as a loop or U with its ends at that 

 part of the nucleus nearest the idiozome body. This is shown in 

 Fig. 8, where only five of the loops are drawn in their entirety. 

 On a transverse section of such a stage (Fig. 9) one finds 24 

 cross-cut portions of loop ; every two of these portions corre- 

 spond to the two arms of one of the U-shaped loop of Fig. 8. 

 Therefore there are U-shaped chromosomes, with a particular 

 arrangement, to the number of 12 ; hence they must be bivalent 

 with regard to the 24 chromosomes of the spermatogonia. At this 

 early stage (Fig. 8) there is no sign that the space enclosed by 

 such a loop has arisen by a longitudinal splitting of a chromosome, 

 for in fact the characteristic shape of the loops is the same now 

 as at later stages (Figs. 10, 11). Now this was the main basis 

 of my argument before an argument that Janssens and Dumez 

 ignore : that if this were a longitudinal split, we should find its 

 commencement in such an early stage. 



The true longitudinal split commences in the stage of Fig. 

 8, is very prominent in that of Figs. 9, 10 ; this is a clear split- 

 ting of each chromatin granule of the arms of each bivalent 



