A STUDY IN CHROMOSOME REDUCTION. 611 
1911; and others) attribute little importance to this matter, for, as we shall see later, the 
union here is a temporary one. Up to the time of this lateral pairing of the spiremes 
(fig. 15), as I have already suggested, I have been unable to recognise any funda- 
mental difference from the series of changes which occur in the early stages of an 
ordinary somatic mitosis. I do not regard the enlargement of the nuclear cavity as in 
any sense a fundamental difference, for, as I have elsewhere stated, this feature may be 
rationally accounted for on the basis of the high state of nutrition which prevails in 
these cells. This same feature, to some extent at least, is a character of ordinary storage 
cells, which are likewise under a high state of nutrition but which are not merismatie. 
The slight diminution in the volume occupied by the chromatin which may occur in the 
early stages cannot be claimed as a fundamental difference, for this is undoubtedly 
entirely due to the shortening and thickening of the chromatin threads, a feature which 
is common to all somatic mitoses. After this stage (fig. 15), however, the changes that 
occur are no longer comparable to the somatic prophase, for the early pairing of the 
spiremes prevents the separation of the daughter halves of the somatic chromosomes 
and results in the organisation of the bivalent chromosomes. 
As to the manner in which the bivalent chromosomes are developed, it should be 
noted that Farmer (1905) in his work on Laliwm and Osmunda describes this as result- 
ing from the approximation of sides of loops of serially distinct regions of the spireme 
as a whole—that with the looping of the chromatin filament and the approximation of 
the sides of the loop, the spireme segments transversely leaving the two halves of the 
loop as the bivalent chromosome. A similar process has more recently been described 
for Galtonia by Dicgpy (1910). I have made careful search for like conditions in the 
material under present investigation, namely, Smalacina, Kniphofia, and Aloe. In none 
of these plants have I been able to find any evidence in support of the views of these 
writers. It should be remembered that the interpretations of Farmer (1905) and 
his pupil are based on the assumption that the chromatin consists of one continuous 
spireme which segments transversely so as to form definitive chromosomes. The 
recent work of Stomps (1910) has, I think, established it beyond much doubt that in 
Spinacia, at least, the chromosomes are never arranged in a continuous spireme. [ 
shall attempt to prove that this is also true for Smilacina, Kniphofia, and Aloe. Let 
us examine the prophase of these plants in turn, taking Smlacina first. In the reticu- 
lum stages the difficulties of identifying the individual spiremes are of course quite 
obyious. Our conclusion from these very early stages are based upon the number of 
free ends that may be distinguished—and here we cannot be certain that these ends are 
not sometimes due to sectioning. A little later, however, when we have transition 
stages from the reticulum to the spireme conditions, and at the close of the growth 
period, a large numberof independent spiremes may be distinguished with comparative 
ease. In the consecutive series of stages represented in figs. 6, 7, 8, 9, and 10, the 
number of free independent chromatic filaments that are clearly observable is sufficiently 
convincing that in no stage of the prophase in Smilacina is there one continuous 
