249 
Pollen Mother-cells of Lilium canadense. 
belonging to separate species. I have already referred to his views as to the 
difficulty of adjustment of parental idioplasms derived from different species 
in the heterotypic prophases ; and he suggests that this inability of the 
idioplasms to adjust themselves to one another in some way makes im- 
possible the ‘splitting’ of parental qualities which occurs in the formation 
of the germ-cells of varietal hybrids. I confess that I find it impossible to 
imagine how such a failure of adjustment of the two idioplasms could lead 
to the production of a constant idioplasm, unless it be supposed that it 
necessitates an actual complete fusion of the two parental substances ; and 
such a fusion could hardly occur in such a case unless the possibility of 
fusion is characteristic of the structure of the idioplasms themselves. I am 
inclined to think, therefore, even though at present the known constant 
hybrids be all of specific rank, that their occurrence is evidence of the 
possibility of the perfectly normal occurrence of a more or less complete 
fusion of the idioplasms in the prophases of the heterotypic division. This 
conception of de Vries’ has been adversely criticized also by Correns (’03 c). 
2. There may be in part a fusion such as was described above, while 
certain portions of the two idioplasms remain uncombined and capable of 
distribution to the germ-cells in varying combinations. This case is 
a transitional one, and will, therefore, be considered in connexion with 
the next. 
3. There may be more or less mixture of the idioplasms as a result of 
the contact of the two threads, but no actual fusion. The substance to be 
separated into two portions by the splitting of the thread is then not 
a single idioplasm, and, unless we suppose that the actual number of idio- 
plasmic structures may be increased from generation to generation, there 
cannot be an equation division ; but, for the production of functional germ- 
cells, it seems most likely that there should be a redistribution of the 
parental materials in such a way that each daughter thread receives an 
endowment corresponding to a complete individual. But the two resultant 
idioplasms may contain the materials of the parental idioplasms arranged 
into entirely new and different combinations. This redistribution of the 
materials of the parental idioplasms is a process which can occur only at 
a certain point in ontogeny ; it is here, then, if anywhere, that the element 
of chance may be supposed to enter, especially as its operation will effect 
the greatest possible amount of variation among the offspring. A similar 
operation of the law of chance has been suggested by Strasburger (’04) in 
the separation of the chromatin granules as a result of the division of the 
‘ zygosome.’ As a result of the mixing of the parental idioplasms and 
their redistribution, we should then find, if the cases at hand be numerous 
enough, every possible combination of different portions of the two idio- 
plasms which is capable of giving rise to a complete set of individual 
characters. But this is virtually a restatement of Mendel’s * law of the 
