202 Farmer and Digby.—On the Cytological Features exhibited by 
These peculiarities persist throughout their existence, and are very strongly 
marked in the later stages of diakinesis. As regards numbers they exceed 
those of the corresponding nuclei of either parent, though careful counting 
has convinced us that there is no very great constancy in the numbers 
actually present in the nuclei of this plant. We have already alluded 
to this circumstance in connexion with the other species, but it is far more 
obvious here. The chromosomes, so far as we could estimate them, vary 
from about 95 to 125, though about 95 to 105 is the more common range. 
The striking difference in individual size (see Figs. 30, 31, 32) is a point of 
some interest, inasmuch as it might be taken to indicate that the larger 
ones represented bivalents whilst the smaller ones constituted the univalent 
remainders which had failed to pair. This hypothesis would fit the fact 
that the chromosomes contributed by the two parents respectively are 
numerically different (about 34 in P. aureum and 90 or more in P. virig. 
elegant.). But in any case such unions to form bivalents must at best be 
irregular, and we entirely failed to find anything which would justify us in 
asserting that there were 34 bivalents or indeed any other constant number 
of larger chromosomes which could be really identified as such. Certainly 
there is no such regularity in this instance as exists, for example, in the 
hybrid Drosera according to Rosenberg. We would, however, point out 
that this seems to us to be in no way surprising. Indeed it is difficult, on 
the assumption that the chromosomes ordinarily contributed by the male 
and female parent respectively are to be regarded as homologous, and 
therefore destined to pair at meiosis, to see how any such homology can 
exist at all when the numbers are different in the two parents. For it is, 
in the first place, extremely improbable that any single one of those of 
the more numerous lot can be equivalent to any single one of the less 
numerous group, and it is obviously impossible that there can be any 
close resemblance between the whole of the lot comprised in the smaller 
number and an eqjial number (less than the whole) of the larger set 
furnished by the other parent. In many of these cases there are, as in 
the Drosera plants under consideration, twice as many chromosomes 
contributed by one parent as by the other. It seems not unlikely that the 
double number of the one has arisen by a division of the individual chromo¬ 
somes in the ancestor of D. longifolia which have thenceforth remained 
distinct. If this has occurred by a transverse fission, then we might expect 
one chromosome derived from the (10) parent ( D . rotundifolia) to unite at 
meiosis with two chromosomes from the (20) parent (D. longifolia ), so that 
there would be practically ten bivalents produced. If, however, it be 
argued that the twenty chromosomes of D. longifolia have arisen not by 
transverse, but by longitudinal fission, and that they have, in the absence of 
hypothetical ‘ regulative ’ processes, persisted unchanged, and so have come 
to be characteristic of the species in question, we are thereby driven to the 
