134 REPORT OF THE CONFERENCE ON GENETICS. 
“Yes,’’ was my reply. “Then one or both the parents of these families 
were really dominants,’’ was the answer to me. 
I did not think that this theory was true, because it seemed to me that 
to say, that, unless a quarter of the family produced by an animal 
consisted of recessives it was not a hybrid, was a very easy way of estab- 
lishing the fact that hybrids always produce recessives in the proportion 
of one in every four. It seemed that an argument of that kind was not 
likely to be based on anything having any existence in nature, but it is 
a strong warning not to be led away by appearances that when I tested 
this theory I found it to be true. The animals falling into the category 
CC in F, are sharply distinguished into two kinds: (i) hybrids that will 
produce albinos in the proportion of one in every four, and (ii) dominants 
which when mated inter se produce no albinos atall, and when mated with 
albinos are dominant over them. I have proved these two kinds to exist. 
But the existence of the second of them is fatal to the suggested 
Mendelian theory outlined above. For how could individuals whose 
germ-cells appear to bear a new unit g’g arise, from a hybrid in which 
the germ-cells, either carried an element representing g’ or g, and never 
both ? The suggestion that neighbouring ova, or spermatozoa, fuse is not 
likely to meet with general approval, yet it is the only one which will 
account for their appearance if our Mendelian theory is true. We have 
to choose between the two improbabilities: (i) that neighbouring germ- 
cells fuse, and (ii) that none of the germ-cells of the hybrids bear 
elements representing animals like them ; and two probabilities: (i) that 
neighbouring germ-cells do not fuse, and (ii) that some of the germ-cells 
of the hybrid contain elements representing animals like them. I 
choose the probabilities. But, in claiming to have demonstrated the 
falsity of the Mendelian theory described above, I do not wish to be 
credited with having “discovered an exception to Mendel’s Law.’’ On the 
contrary, the best measure of the progress which Mendelian inquiry has 
made in these two years is the fact that while at the beginning of them 
the existence of hybrids that breed true would have been regarded as a 
difficulty, to-day a reasonable explanation of their occurrence has been 
given. 
Progress in knowledge is made by the suggestion of hypotheses, and their 
rejection when found to be false; by this means the Mendelian has been 
able to account for some very complicated cases of segregation, and for 
reversion in some cases as well. 
It is natural to inquire how much experiments of this kind tell us 
about heredity. We are told that Mendel’s law only applies to a very 
limited class of facts, that there is only a certain very limited and definite 
set of characters to which it applies, or that it only deals with the 
phenomena of hybridisation. 
Let us consider these objections one by one. With regard to the first, 
I would say, what I have said before, that the service which Mendelian 
theory has done to progress in the study of heredity lies partly in the 
facts which it has accounted for, and partly in the method which it has. 
introduced ; and that even if Mendel’s law has a limited application, his 
method has a great future. 
