1897 ] LIFE HISTORY OF LILIUM PHILADELPHICUM 435 
lying on the spindle have often been mistaken for centrosomes, 
which would explain the instances where many centrospheres 
have been reported at the ends of the spindle threads ; for it is 
just at this stage that the micronucleoli would have such a 
position. 
The formation of the spindle was not traced. In the mother 
star stage one centrosome appears very definitely at each pole 
( fig. 26). During metakinesis the centrosome divides into two 
( fig. 36 a). In the daughter skein stage two large centro- 
spheres are sometimes seen at the poles (fg. 38). No special 
effort was made to bring out the centrospheres, and they were 
not often seen, but wherever they appeared they showed their 
normal structure and position. 
During the formation of the chromosomes from the chroma- 
tin band the twisted loops begin to shorten and thicken, giving 
the appearance of a single twisted linin thread with an irregular 
double row of chromatin granules. The linin thread also, espe- 
cially at this stage, stains a very dark purple or black with Dela- 
field’s haematoxylin, exactly like the chromatin granules them- 
selves. At this stage also there is a deposit formed around the 
chromatin loop which gradually becomes thicker as the chromo- 
some reaches maturity (figs. 27-23 6). With Delafield’s 
haematoxylin and erythrosin this deposit stains a light pinkish 
red, while the chromatin band stains a very dark purple. At a 
later “stage, just before the formation of the mother star, the 
whole chromosome begins to stain deeper, until it finally has a 
homogeneous appearance when treated with this double stain, 
and shows no structure whatever, the whole chromosome appear- 
ing like a huge mass of chromatin matter (figs. 2g-27), and 
it is necessary to employ other stains to differentiate the chro- 
matin band. 
When the chromosomes become arranged on the spindle 
_ threads in the equatorial plane they are so situated that the end 
_ having the two free ends of the chromatin band are attached to 
the spindle threads, the loop being turned outward and project- 
_ ing freely beyond the spindle ( jigs. 27-28). There is no 
