20 BOTANICAL GAZETTE [july 



strand connecting the cells (figs. 45, 46). When the cytoplasm has 

 shrunken slightly away from the cell wall they are particularly clearly 

 observable. These connections appear to be in all cases between 

 mother cells, and in no case have they been observed between the 

 mother cells and the tapetum. Generally one such strand is seen 

 connecting two cells, but not infrequently there are two or three or 

 occasionally even more. There is no constriction or change in the 

 nature of the connective as it passes through the cell wall. These 

 connections are even larger and more conspicuous in O. gigas, where 

 the mother cells are also much larger. They have not been observed 

 in O. Lamarckiana or the other forms, but they doubtless occur in all, 

 being probably smaller and more inconspicuous in some. 



Discussion 



The method of reduction described in this paper at once raises a 

 number of questions of prime importance from the cytological stand- 

 point, as well as from that of the relation subsisting between heredi- 

 tary and cytological phenomena. A discussion of all these features 

 will not be attempted at this time, the intention of the writer being 

 merely to indicate the general directions in which the facts point and 

 the possible bearing which these data may be found to have on the 

 problems connected with the phenomena of mutation in Oenothera. 

 A fuller discussion of these subjects is reserved for a future time, after 

 the presentation of further data. In the present communication 

 reference will be made only to the most recent papers on reduction in 

 plants, the purpose not being a review of the literature, or a dis- 

 cussion of present views, except in so far as they bear directly on the 

 matter in hand. 



The recent accounts of reduction in plants, given by Berghs 

 (3 » 4> 5> 6), Gregoire (16), Strasburger (31), Allen (i, 2), 

 Miyake (18), Overton (22), Rosenberg (25), Yamanouchi (33), 

 and others, have agreed in so far as the following general course of 

 events is concerned : In synapsis a pairing of homologous maternal 

 and paternal elements occurs either in the form of gamosomes (Stras- 

 burger and Miyake), prochromosomes (Overton), or parallel 

 threads (Allen, Rosenberg, Gregoire, Berghs, Cardiff 7, and 

 Yamanouchi). In every case two parallel threads result, which unite 



