in the Pollen- Mother- Cells of Larix . 285 
into polar caps, and from these caps fibres grow out into 
a liquid which appears between the fibrous cap and the 
nuclear membrane. But there is no sharp line of distinction 
between the two methods of spindle-formation ; for in the 
pollen-mother-cell of Iris the origin of the first spindle is 
multipolar polyarch, that of the second multipolar diarch ; 
while in the pollen-mother-cell of Nymphaea the origin of the 
first spindle is similar to that of the second in Iris . Duggar 
(’00) finds, too, that in the division of the microspore nucleus 
of Symplocarpus and Peltandra the spindle is originally 
multipolar diarch, the fibres being arranged perpendicularly 
to the wall near which the nucleus lies. Mottier (’98) describes, 
in the vegetative cells of Lilium , a multipolar stage preceded 
by an extra-nuclear felt, just as in pollen-mother-cells ; and 
a multipolar spindle is figured by Ikeno (’98) in the embryo 
of Cycas. 
Multipolar spindle-figures have an important bearing on 
the question as to the presence or absence of central bodies 
(attraction-spheres, centrospheres, centrosomes, &c.). We 
have seen that the evidence is conclusive for the general 
existence of a multipolar stage in the history of the spindle 
in the cells of the flowering plants ; and such a stage seems to 
negative the possibility of the formation of the spindle in 
these plants through the agency of centrosomes which station 
themselves at opposite points in the cell, so initiating mitosis 
and determining the position of the spindle-poles. But 
Guignard, who (’91 a , ’91 b, *91 c) first described attraction- 
spheres in Phanerogams, has more recently (’97 a, ’97 b , ’98) 
maintained that the poles of the multipolar figure are often 
occupied by centrosome-like granules, which are, he holds, 
true kinetic centres, and that the bipolar stage results from 
their fusion into two typical centrospheres. A very similar 
process has been described by Moore (’94) in certain animal 
cells. But even Guignard admits that the spheres may dis- 
appear in the resting stage and be formed de novo during 
mitosis, sometimes not until after the appearance of the 
multipolar spindle ; in the latter case, the cones arise by the 
