374 BOTANICAL GAZETTE [JUNE 
never so prominent asin Lilium Philadelphicum (figs. 25-25a). In 
L. Philadelphicum the chromatin granules divide and the chroma- 
tin band undergoes longitudinal splitting before much shorten- 
ing and looping take place, but in Erythronium the division of 
the granules seems to be somewhat later, and they do not 
appear double until the band has twisted into the twelve loops 
(jigs. 26, 27, 28,37). The granules are large and more or less 
irregular in shape. A little later the chromosomes appear 
homogeneous throughout. This would certainly give support 
to the belief that the division of the chromatin granules is a 
mechanical contrivance for bringing about the longitudinal 
division of the linin band, although this might not at all interfere 
with their supposed function as bearers of hereditary tendencies. 
The breaking apart of the twelve loops to form the twelve 
individual chromosomes ( fig. 29) appears to be accomplished by 
the twisting and contracting movements of the band. The 
chromosomes are usually of various sizes, some being much 
larger than others. They often appear as single coiled loops 
(figs. 37, 38), but usually their double nature can be readily 
observed. The chromatin loops are not so closely coiled as in 
Lilium Philadelphicum, and when they are arranged in the mother 
star the twisted condition can still be distinguished. The 
chromosomes appear to be attached to the spindle threads near 
the two free ends of the loop, the closed end extending outward 
(see figs. 39-48, chromosomes before division, and figs. 49-57: 
chromosomes after division). During metakinesis the loops are 
uncoiled, and the two free ends are gradually pulled apart until 
each chromosome breaks in the middle, thus accomplishing a 
transverse division, one end of the original chromatin loop 
going to the one pole, and the other to the opposite one. This 
would be a true reducing division. It is exceedingly difficult 
to follow out the course of events at this point, and there is 
always room for doubt as to correct interpretation. There is a 
possibility that the loops are attached at the closed ends. 
However, some of the examples are quite convincing, and 
another than a transverse division seems out of the question (see 



