54 Univcrsitjf of ('(ilifornia Pii])Iirat{o)i.^ in Botcnn/ [Vol. 7 



to that which Atkinson (1899, p. 10) found in T. grandifionim, where 

 "the period during which tlic division of the pollen mother-cells takes 

 place extends over seven or cigiit months." As in T. grandifionim. 

 Ilic time of division of the pollen mother-cells vai-ies with the season, 

 the location, and the degree of development of the other parts in the 

 individual. As noted above in the case of large i)lants wliidi produce 

 a numl)er of buds during the active season below ground, any two buds 

 may be strikingly different in the amount of development wliicli they 

 exhibit during the month of May. By October oi- November this 

 discrepancy in development appears to have been made up and the 

 reduction division takes place much more nearly simultaneously in the 

 various flowers of a rootstock than inii>'ht have been anticipated. It 

 seems clear that approximately six weeks elapse in average cases be- 

 tween the synaptic condition and the cutting off of the generative 

 nuclei. Of this period of six weeks, the first three include the stages 

 intervening between synapsis and the homotypic division and the 

 remainder are occupied by the stages between the homotypic divisions 

 and the cutting off of the generative nuclei. 



Mention has already been made of the rapid growth of the pollen 

 mother-cells and their nuclei previous to synapsis. The locule also 

 may often increase so rapidly in size that the pollen mother-cells come 

 to lie in a mass in the middle of a large cavity. During synapsis the 

 chromatin often becomes so completely contracted that all semblance 

 of structure disappears. It is during this stage apparently that the 

 nuclei of many pollen mother-cells degenerate, cytoplasm and cell- 

 walls disappearing shortly thereafter. During the later prophases of 

 the heterotypic division the nuclear membrane seems to disappear and 

 the pollen mother-cells round up. Atkinson (1899, p. 12) states that 

 the double nature of the broad chromatin band is indicated by the 

 presence of a double row of dense bodies. In the material at hand 

 it was not possible to distinguish many granules in the bivalent chro- 

 mosomes because of thickness and density of the latter, but evidence 

 of their double nature was furnished by the occasional appearance of 

 a longitudinal split while they were in the e(|uatorial plate stage. In 

 the metaphase the chromosomes are arranged in a nearly flat equa- 

 torial plate as Vs whose free ends stand out around the spindle (pi. 9. 

 fig. 1), as Ernst (1902) reports for the megaspore mother-cell in Paris 

 quadrifolia and Trillium grandiflorvm. The s])in(lle fibers are at- 

 tached to the chromosomes at the apices of the \'s, the chromatin 

 becoming drawn out thin at the point of attachment, as Atkinson 



