375 
Pollen Development in Lactuca . 
in relative length from one nucleus to another, as shown by the rows of 
chromosomes in Figs. 37 a to 55 a , are probably not so great as they 
appear. 
It is also clear that, as would be expected, it is the longest pairs which 
invariably show the most marked twisting in the early stages. Examination 
of the later stages of diakinesis frequently shows the longest or intermediate 
pairs with a straight longitudinal split. Two 'possible explanations of this 
are admissible. Either the loops which were twisted must untwist again, 
or the pairs must split apart with ‘ crossing over ’ of certain segments. 
While we have no demonstrable evidence that such a rearrangement of 
segments rather than untwisting occurs, yet figures such as 33, and 
particularly Figs. 34-36, show that the condition of torsion persists even 
after the condensation of the chromosomes has progressed far. This gives 
reason to believe that the subsequent longitudinal split of the bivalent 
chromosomes will take place across the twists. Moreover, a twist which 
always untwisted again along the original lines would appear to have no 
raison d'etre , whatever are the physical or physiological processes which 
bring about the torsion. 
Occasionally ten chromosomes instead of nine have been counted in 
diakinesis. This has occurred a number of times in the large number of 
nuclei at this stage whose chromosomes were counted. The explanation, 
as indicated by Figs. 37 and 40, is evidently that the two halves of 
a bivalent chromosome have become separated. In Fig. 40 the two halves 
of one bivalent are lying parallel and almost separated from each other. 
The fact that the two bodies marked a in Fig. 37 are narrower than the 
other chromosomes, and are of equal length, makes it clear that they 
represent the two halves of a bivalent chromosome. Other cases of 
precocious splitting are indicated in Figs. 35, 39, 42, and 47, in which 
several of the bivalents are shaped like the V’s or X’s which are so 'usual 
at this stage in many plants. Their relative infrequence here, as well as 
other facts, indicates that in Lactuca the members of the chromosome 
bivalents are held together exceptionally closely. This is in strong contrast 
with Oenothera , where the attraction which leads to pairing of the chromo- 
somes is exceptionally weak and is a source of irregularities in the 
distribution of the meiotic chromosomes. 
Another irregularity occasionally met with is the presence of only 
eight pairs (Figs. 46-48). Figs. 45 and 46 indicate how this condition 
has probably arisen. In Fig. 45 two pairs of bivalents are united by 
strands of deeply staining material, while Fig. 46 shows seven bivalents of 
normal length and one extremely long one, formed probably by the union 
of two long bivalents end to end. 
If this structure is composed of two bivalents united end to end it 
could scarcely have been formed by failure of the spireme to segment 
