1900] MULTIPOLAR SPINDLE IN GLADIOLUS 149 
until a much later stage, as can be seen readily in figs. z-8. This 
fact is significant, because it is quite clear that the method of 
spindle development depends much upon when the nuclear mem- 
brane breaks down. This is very well illustrated by the two 
types which we have before us. In Cobaea, where the nuclear 
wall breaks down at a very early stage, we have a central net- 
work formed which occupies the space of the nuclear cavity. In 
Gladiolus, where the nuclear wall persists, the network is not 
formed in this central position, but is formed outside of and 
immediately surrounding the membrane. This difference also 
has its effect on the character of the network. When the 
nuclear wall breaks down at an early stage the nuclear cavity 
affords a large space for the formation of the network. Its 
meshes are consequently quite large. On the other hand, when 
the nuclear wall persists, as it does in Gladiolus, the network 
appears in the form of a close weft or felted zone. 
Very careful observations were made at this stage to see if 
etic kinoplasmic threads penetrate the nuclear membrane, 
as recently observed by Wiegand in Potamogeton. Although 
hundreds of cells were examined no such penetration was 
detected. : 
When this weft or felted zone commences to form, one might 
st frst think that it arose by the fraying out of the nuclear mem- 
rane. But as it grows to such a width ( fig. 7), and the nuclear 
mm apparently loses none of its distinctness ( jigs. 1-8), 
“sms much more probable that it grows at the expense of 
Perikaryoplasm as the central network does in Cobaea. 
a ihe ce oe of its rst appearance the felted zone stains 
meshes es with gentian-violet, and as it increases in size , 
which it Se larger and the sharp kinoplasmic threads : 
weft oo wey readily distinguished. As soon #8 ; e 
but pushes hen size ( fig. I iF it ceases to grow uniform a 
fat ak the oe into several projections as shown in /ig. Ee i 
Number of rh ite make out, there appears to be no de . 
Was the lar €s€ projections. fig. 2 shows at least five. Six 
est number observed in cross-section. Just how 
