RECENT ADVANCES IN ANIMAL BREEDING. 158 
chance union of g’ and g, borne by different parents. A hybrid in this 
generation—I',—is, therefore, never produced by the union of similar, but 
always by the union of dissimilar gametes; and if an I’, hybrid could 
multiply parthenogenetically, none of its offspring would be like itself. 
This theory as to the nature of the character-representing elements borne 
by the hybrids is so remarkable that one requires very strong evidence for 
it, to believe it. The only evidence so far adduced is that the proportions 
in which the various kinds of young occur are those demanded by the 
theory ; but this does not prove the theory to be true. 
The question we have to ask ourselves in considering the value of the 
evidence for an hypothesis is not ‘‘ How many cases are there which are 
consonant with its truth ?”’ but “Is there a single case which is not?” 
The list of cases in which the proportion 1:2:1 obtains in the F, 
generation is lengthening every day, and it is imagined that the value of 
the evidence for this particular theory becomes greater as the list becomes 
longer. The simple truth, that I have stated in the form of two questions 
above, is often forgotten. 
We are too apt to think that it is sufficient to rest content with the 
many that are with us; and too ready to forget that we ought to be up 
and seeking out one that may be against us. 
Are there any facts which render the above-outlined Mendelian theory 
untenable? I have at my disposal two, to only one of which will I refer 
now. I may say by way of preface that I do not wish my remarks to be 
construed as antagonistic to Mendelian theory as a whole, but merely 
critical of a particular hypothesis bearing that name, which was put forward 
two years ago: an hypothesis in which Mr. Bateson ceased to believe before 
I did. 
At a time when I still thought that it was a useful subject for investi- 
gation to try to find out which of the two theories, Galton’s or Mendel’s, 
fitted the result of my experiment best, I obtained a result that was 
apparently conclusive in favour of the former. 
The result, which I have described before, but which I may briefly 
recapitulate here, was obtained by tabulating the difference between the 
results of making two kinds of hybrids differing not in any visible 
character but in their pedigree. The two kinds of hybrids that were 
used were (i) a hybrid produced by the union of two hybrids, and 
(ii) a hybrid produced by crossing a hybrid with an albino; we may call 
the former HH and the latter HA. Three kinds of matings can be made 
with this material; namely, HH x HH, HH x HA, and HA x HA. 
In each of these types of uniona hybrid is mated with a hybrid. So that 
T argued that according to the Mendelian theory a quarter of the population 
produced in each of the three cases should be albino; but that according 
to the commonly accepted view of heredity, known as the law of contribu- 
tion, one would expect the proportion of albinos to be greater in proportion 
as the number of albinos in the pedigree of the hybrid parents was 
greater. This was found to be the case. But it was pointed out to me 
that this result was not evidence against Mendel’s theory, unless I had 
established the hybrid nature of every individual used in the experiment. 
“Have you done this?’ I was asked. -“ Are there any cases of 
families in unions of type (i) where no albinos have been produced ?”’ 
