OF THE GEKM CELLS OP METAZOA. 225 



But in Ascaris (Braucr, 1893 b), in Salamandra (Meves, 1896), in the Rat (Len- 

 hossek, 1898), in Selach'd (Moore, 1895), and in Amphiuma (McGregor, 1899) the au- 

 thorities cited agree that both maturation divisions are equational. Now it does not seem 

 a priori probable that in some Metazoa a reduction division should occur, and in others 

 not. The case of Ascaris would seem to show no sign of a reduction division, for Brauer's 

 careful study apparently shows that each bivalent chromosome becomes split longitudi- 

 nally twice ; yet Sabaschnikoff 's more recent study (1898) would show that another inter- 

 pretation of the phenomena is possible (but not proved), namely, that the chromatin 

 microsomes may become rearranged into fours in such a way that one of the maturation 

 divisions may be reductional. In the Salamander, Flemming (1887) showed that the 

 mitosis of the first generation of spermatocytes has remarkable peculiarities, so that he 

 named it a " heterotypic " mitosis. The most remarkable of its characteristics is that the 

 chromosomes are longitudinally split in shape like horseshoes, and that they open up into 

 the forms of rings, the ends of the daughter horseshoes retaining their mutual connection. 

 Such a heterotypic mitosis was corroborated by Meves in his description of the spermato- 

 genesis of Salamandra ; and it has been shown to be characteristic of the first maturation 

 division in Selachii, the Rat, and Amphiuma by Moore, Lenhossek and McGregor, respec- 

 tively.* All these writers show that the heterotypic mitosis results in a longitudinal 

 division of the chromosome, and there can be no reasonable doubt of the correctness of 

 their descriptions. But I think they are mistaken in concluding that because the hetero- 

 typic division is a longitudinal division of the chromosomes, that therefore it is an equa- 

 tion division. For the chromosomes of these spermatocytes are bivalent there are just 

 half as many in the spermatocytes as in the sperruatogonia. Since there is no loss of 

 chromosomes in the spermatocytes, there must take place a union of univalent chromo- 

 somes into pairs during some part of the growth period i. e., in the synapsis stage. I 

 venture the view that in the Vertebrates either (1) the bivalent chromosomes are formed 

 by every two univalent chromosomes becoming apposed to one another side to side i. e., 

 along their whole length, so that the two would compose a double horseshoe or (2) by the 

 two ends of one univalent chromosome becoming closely connected with the two ends of the 

 other, so that the whole would have the form of a ring. From what has been described 

 for these bivalent chromosomes, we know that the longitudinal split does not divide their 

 ends, but the ends are unsplit. Accordingly, it would appear probable that the essential 

 process in the formation of these bivalent chromosomes is that the two ends of one univa- 

 lent chromosome become united with the two ends of another, while it would be of secon- 

 dary importance whether the two chromosomes might be apposed along their whole lengths 

 or not. 



* Moore states that the second maturation division is also heterotypic, but his figures do not prove his point, 

 which needs ree'xamination. 



