WILCOX : SPERMATOGENESIS. 11 



The dumb-bell figures become associated in pairs (Plate V. Figs. 229, 

 2-12, 243). Each of the six groups (Diagram G) thus formed has the 

 value of four chromosomes, each dumb-bell beiug equal to two chromo- 

 somes (Plate IV. Fig. 178, Plate V. Figs. 242, 243). These quadri- 

 valent groups may be formed by the approximation of the pairs of dumb- 

 bells in one or the other of two ways. Either they become arranged 

 quite irregularly (Plate III. Fig. 116, Plate V. Fig. 229), or the pairs 

 may at first lie across each other at right angles, and later come to be 

 parallel (Plate V. Figs. 229, 242, 243). A comparison of the draw- 

 ings last mentioned will show how by the fusion of the ends of the two 

 parallel dumb-bells a ring results, such as is shown in Plate IV. Figs. 

 178, 179, 181. 



By a slight variation in the time at which the massing of the chroma- 

 tin granules takes place the process up to this point may pursue a course 

 apparently quite different from that described. The chromatin granules 

 of the original chromatin thread do not become massed into definite 

 chromosomes as early as in the method just outlined. Consequently 

 the transverse divisions result in the formation of twelve segments 

 (Diagram 8) with very irregularly serrated edges. These segmenta asso- 

 ciate themselves (Diagram 9) in pairs (Plate V. Figs. 201, 225, 227). 

 They are either so closely applied to each other as to appear like single 

 rods, or else show two rows of granules (Fig. 227), and thus give the 

 same appearance that would have resulted from a longitudinal splitting 

 of a single segment. The component halves of these six segments sep- 

 arate from each other except at their ends, and thus form rings, as in 

 Plate IV. Fig. 174. The granules scattered along these rings then 

 collect into four chromosomes (Diagram 10). The result is, therefore, 

 the same as by the process first mentioned. 



This account of the formation of rings varies somewhat from those 

 of vom Rath ('93) and Hacker ('93), and is entirely different from 

 Brauer's ('93) account. These differences, as well as the points of agree- 

 ment, will be discussed under the literature of the subject. 



The position of the chromatic rings at the equator of the spindle is 

 shown in Plate V. Figs. 192-195 and 237-241. The rings are always 

 complete at this stage, and the first step in the metakinesis of the sper- 

 matocytes consists in a separation of the rings into half-rings. ^With 

 the iron-hsematoxylin method the majority of the spindles present the 

 appearance of Figures 194, 195. The planes of the rings all pass through 

 the axis of the spindle. Hence it is impossible to see that the chromo- 

 somes are arranged in rings, except when the rings are turned broadside 



