454 The Botanical Gazette. [November, 
plasm by a well defined granular layer; and in cases where 
karyokinesis had advanced to any considerable extent, two 
centrosomes could be distinguished at each pole of the spindle. 
Sometimes there appeared to be but one at each pole, but 
careful focusing generally demonstrated the fact that one was 
lying below the other. Inthe root tips of Allium, where the 
division is tranverse to the axis of the root when one goes a 
little distance from the apex, the attraction-spheres always 
appear at the upper or lower end of the nucleus as seen in 
longitudinal section. In the resting cells, they generally lie 
quite close to the nucleus in a little indentation. 
In the epidermis of the onion scales I observed these bodies 
_ in a sufficient number cf cases to convince me that they were 
true attraction-spheres, since they had the same appearance 
and took the same stain as those which I saw by the side of 
the close skein of the daughter nucleus. Now in these epi- 
dermal cells of the onion scales the nuclei are all resting, and 
therefore the objection that the centrosomes may have just 
come out of the nucleus in the beginning of division cannot 
be made; and so I hold that the attraction-spheres with their 
centrosomes do not enter the nucleus during its resting stage 
but remain permanently outside of the nuclear membrane. 
Moreover, these cells of the epidermis of the bulb scales of 
Allium were all definitive resting cells; yet with the iron-tannin- 
Safranin stain it was demonstrated that the centrosomes and 
attraction-spheres were still present beside the nucleus, and 
that they retained their usual structure. 
When division of the nucleus takes place, I found the 
attraction-spheres in the onion roots at the very beginning of 
the close mother skein stage, one at the upper and one at the 
lower pole of the future spindle, still close to or in contact 
with the nucleus (fig. 2); and though I did not find any 
stage where one of the bodies had gone only part of the way | 
around, yet there can be no doubt that one or both had 
traveled around from their original position to the poles. In 
the following stages the spheres elongate, and generally by 
the time when the nucleus has reached the loose mother skein 
(fig. 3) the centrosomes and _ their spheres have divided, 
though they still lie closer together than in the later stage>- 
During metakinesis and the daughter star stage (figs. 5-7); 
they can be seen very distinctly at each pole; and they keep 
this position in relation with the nucleus through all the suc 
