95 
Fossombronia longiseta , Aust . 
nucleus at this stage failed to reveal anything that might be interpreted 
as a centrosphere or a centrosome. In this respect Fossombronia certainly 
appears to differ from Marchantia polymorpha , for which Ikeno has figured 
a well-defined centrosome appearing first among the linin threads, after- 
wards moving to the surface of the nucleus from which it is discharged. 
It was thought most likely to appear, if present at all, at the time of 
elongation of the nucleus just prior to the development of the spindle, 
but here again the protoplasm surrounding the nucleus disclosed the 
presence of nothing that could be considered a centrosome. The 
protoplasm about the nucleus presents a rather homogeneous granular 
appearance, and it is barely possible that this may account for the failure to 
observe centrosomes were they present. 
The various stages of mitosis are passed through quite rapidly, and 
intermediate stages between the spirem and mature spindle were not 
observed, consequently it is not known what structures, if any, are instru- 
mental in bringing about the polarity of the spindle. In Marchantia 
polymorpha Ikeno figures the centrosome as active in spindle-formation, 
the two exerting an opposed pull upon the cytoplasm after the nuclear 
membrane has extended in the direction of the centrosomes and has made 
contact with them. 
However, since no centrosome or homologous structure was seen 
in either of the above-named or antecedent stages, it is probable that the 
spindle develops in much the same way as in Antlioceros , where centrosomes 
are not known to exist. 
Successive divisions of the spermatogenous tissue occur, in which the 
cells seem, so far as observed, to divide simultaneously, though in one 
antheridium all the cells in one-half of the section were in early prophase, 
while those in the remaining half were in metaphase. Up to the division 
resulting in the formation of the spermatid cells the divisions take place 
so as to form nearly cubical cells which are normally quite regular, 
following, as can be seen in many cases almost to the last division, 
the original lines of the primary and early divisions. Fig. 14, Plate V 
represents mature cells of an antheridium about two-thirds of the way 
along in its development. Fig. 16, Plate V represents similar cells 
of an antheridium just preceding the first of the last two-divisions, at 
which time the antheridium has attained nearly its mature size. Here 
we notice a considerable decrease in size as the cells divide, and it would 
seem that the growth of the antheridium does not parallel the increase 
in the number of cells 1 . 
After a number of successive and regular divisions in the spermato- 
genous tissue, the antheridium is divided into a greatly increased number 
of cells which undergo a final division to form the spermatid cells. 
1 Peirce (’03), pp. 165 - 6 . 
