DAVIS: SPERMATOGENESIS. 97 
cases it is usually possible to recognize three autosomes of the rod type, 
two of which are much smaller than the other chromosomes, while 
the third is medium sized; three autosomes of the cross type, of 
which two are medium sized and the third is one of the larger chromo- 
somes; and five autosomes of the ring and crossed-loop type, of which 
four are large and one medium sized. 
As the chromosomes divide, the rod-shaped forms first become 
dumbbell-shaped by thé formation of a transverse constriction, which 
continues to deepen until the two components are entirely separated. 
The cross-shaped chromosomes first become converted into rods by 
an elongation in the direction of the spindle axis and a corresponding 
shortening of the transverse arms. Later they become dumbbell- 
shaped and finally divide as in the former case. In the rings and 
crossed Joops the process is more complicated. ‘The ends of the univ- 
alent components are attached to mantle fibers connected with the 
more distant pole. Consequently as the chromosome divides its 
univalent components are pulled past each other (Plate 4, Figs. 63, 
65, Plate 7, Figs. 179, 180). In all cases as the univalent components 
separate the longitudinal split, which has been temporarily hidden, 
again appears so that the chromosomes as they move toward the pole 
are composed of two rods, which may lie parallel or may diverge at 
the ends nearest the equator of the spindle, giving rise to the well known 
V-shaped chromosomes (Fig. 65, Fig. W) so characteristic of the het- 
erotypical mitosis. As the ring- or loop-shaped autosomes (Figs. 181, 
182) are pulled out during division and the longitudinal split reap- 
pears, it can be plainly seen in favorable cases that the rods resulting 
from the longitudinal split do not lie parallel but are crossed at the 
middle of the dividing autosome. Later the autosomes divide at this 
point, which, as the slight enlargement indicates, is the place where 
the univalent components are joined. I regard this crossed condition 
of the two longitudinal halves of the autosomes as strong confirmatory 
evidence of my interpretation of their structure. 
If the reader will take two flexible, parallel wires, or stiff cords, and 
bend them until the ends are crossed (Fig. Oa), thus forming a loop, 
as in the case of the tetrads, he can get a much clearer idea of the 
method of separation than by any description. Now, by pulling the 
two ends in opposite directions the loop will diminish in size until a 
kink (Fig. Ob) develops in the middle, and when the wires are fully 
straightened out (Fig. O«) they will be crossed just as the autosomes 
are in Figures 181 and 182. 
As a result of their arrangement on the spindle, the univalent com- 
