Beer. Studies in Spore Development. II. 719 
plasm of both the species of Crepis and of Matricaria may have a nuclear 
origin. The nucleoli which are cast into the cytoplasm during the divisions 
are no doubt also responsible for some of these granules. 
Somatic Divisions. 
In order to make my account of the nuclear phenomena of the Com- 
positae more complete, and for comparison with the meiotic phase described 
in the foregoing pages, I have also studied the somatic divisions in Crepis 
virens and Tragopogon . As the results were essentially similar in the two 
plants I will restrict my description to Crepis virens alone. This plant is 
particularly well adapted for our purpose on account of the low number of 
its chromosomes, viz. it possesses six somatic and three heterotype chromo- 
somes respectively, as Rosenberg ( 20 ) has already pointed out. 
The somatic divisions were studied partly in the vegetative cells of the 
anther and partly in the other vegetative tissues of the capitulum. The 
resting nucleus contains, besides the nucleolus, a very delicate reticulum. In 
some cases this reticulum is so fine that it appears more as a cloudy pre- 
cipitate of chromatin than as a network (Fig. 61). When such a nucleus is 
about to divide we find that the chromatin tends to aggregate along certain 
lines. The manner of this aggregation is, however, not always the same. 
In what is probably the majority of cases the chromatin gathers together at 
certain spots from two directions, so that a double line of stainable material 
is formed (Fig. 62). In this way a spireme gradually condenses out from 
the indefinite reticulum of the resting nucleus, and this thread is composed 
of two parallel halves throughout its development (Fig. 65). This corre- 
sponds very closely to the descriptions which have been given of this process 
in other plants by Gregoire (10, 11), Miss Digby ( 4 ), and Dr. Fraser and 
Mr. Snell ( 9 ). In other nuclei of this plant (and also in Tragopogon and 
Crepis t ar ax aci folia) the chromatin appears to concentrate along the lines 
of the future spireme as an undivided whole in which no parallel halves are 
to be seen (Fig. 63). As was pointed out in the case of the homotype 
division, this may in some instances be due to the angle from which the 
spireme rudiment is viewed, but I believe that in both divisions there are 
cases in which the chromosomes develop as single undivided structures during 
the early stages. It is more particularly in those nuclei in which the chro- 
matin has the appearance of a cloudy precipitate that this type of develop- 
ment is met with. Roth Gregoire ( 11 ) and Miss Digby ( 4 ) have described 
an uneven concentration of the chromatin to form a spireme which has 
a corkscrew-like appearance. I have observed a similar zigzag arrangement 
of the developing chromosome bands in Crepis virens (Fig. 64). 
Whichever plan of concentration is followed it results in the formation 
of a spireme which is most certainly discontinuous, and which is more or less 
clearly divided longitudinally (Fig. 66, a and b). Dr. Fraser and Mr. Snell ( 9 ) 
