. 
1903] STUDIES IN SPINDLE FORMATION 95 
Very careful search was made for the radial condition described 
by Osterhout for the second spindle in Agave, but nothing like 
this was found. The fibrils are never radial at this stage. 
Having reached a certain thickness the weft no longer 
increases uniformly, but grows out at irregular intervals from 
the primary cones of the spindle in identically the same fashion 
as it does in the formation of the first spindle (fig. 2¢). 
SUMMARY. 
I. In Iris the formation of the spindle is initiated by the 
transformation of the cytoplasmic reticulum close to the nuclear 
membrane into a weft of kinoplasmic fibrils, which forms a 
complete zone about the nucleus. 
After increasing to a certain thickness, the zone projects out- 
wards at irregular intervals, forming a series of cones which ter- 
minate in sharp points. 
The cones apparently develop at the expense of the cyto- 
‘plasmic reticulum into which they project, and as they grow the 
fibrils composing them lengthen and converge at the apex. 
During the complete formation of the primary cones the 
nuclear wall persists. 
Upon the breaking down of the nuclear membrane the cones 
fuse until there are two groups of them pointing in opposite 
directions. 
The points at which the cones forming these groups meet at 
their apices become the poles of the bipolar spindle. 
2. In Disporum the first indication of the spindle is the for- 
mation of a weft of kinoplasmic fibrils which partially surrounds 
the nucleus. As in Iris, the fibrils composing the weft are 
formed by the transformation of the cytoplasmic reticulum. 
Unlike Iris, the meshes of the weft do not run parallel to the 
nuclear membrane. 
The weft increases irregularly, forming several projections 
which become the primary cones of the spindle. As they grow 
outward the cones become sharp-pointed and their fibrils are 
sharply defined. It is very clear that the kinoplasmic weft is of 
cytoplasmic origin. ; 
