On Mendel's Laws . 
227 
Law of Ancestral Heredity, lead then directly to a special case of that 
law, for the dominant attribute at least. For the recessive attribute it 
does not hold, the chance of a recessive producing a recessive 
offspring is, on the above hypothesis of random mating, one-half, what¬ 
ever the parentage may he. Nor is it difficult to see why the law ap¬ 
plies in the former case. Exhypothesi pure dominant and hybrid zygotes 
produce dominant forms indistinguishable the one from the other, so 
that the somatic characters of the individual are not an absolute 
guide to the character of his germ cells—they may or may not be of 
the pure dominant type even though his soma be of the dominant 
form. If, however, the parent, grandparent, etc., be of the dominant 
form also, the absence of recessive individuals in the ancestry gives 
a stronger and stronger presumption that the germ cells are of the 
type which the somatic characters of the individual would lead one 
to expect. It is equally easy to see why the law does not hold in the 
case of the recessive attribute. Ex hypotliesi again, recessive forms 
can only be produced by pure recessive germ zygotes; it is therefore 
certain without any further witness that the germ cells produced by 
recessive individuals must be themselves recessive—knowledge of 
the ancestry is useless for predicting the nature of the germ cells of 
such an individual, and therefore equally useless for predicting the 
nature of his offspring. Mendel’s Laws, in implying “ that the cross¬ 
breeding of parents need not diminish the purity of their germ cells 
or consequently the purity of their offspring ” (Mendel’s Principles^ 
p. 114 ), do not assert, as stated by Mr. Bateson, “a proposition 
absolutely at variance with all the laws of ancestral heredity however 
formulated ” (my italics). Purity of germ cells may very well subsist 
for a proportion of the individuals of a race without in any way 
invalidating the principle of the Law of Ancestral Heredity, in the 
sense defined; it is a law applying to aggregates and predicates 
nothing concerning the individual. The value of the work of Mendel 
and his successors lies not in discovering a phenomenon inconsistent 
with that law, but in shewing that a process, consistent with it, 
though neither suggested nor postulated by it, might actually occur. 
The form of the law of ancestral heredity to which Mendel’s 
principles have led us is, however, clearly a special case, and the 
next question to be asked is therefore this :—in what way may the 
special conditions under which Mendel’s Laws hold good be 
broadened so as to permit of a generalisation of the results ? Two 
of these conditions suggest themselves at once as being in all 
probability somewhat exceptional in character, viz : (i.) the necessity 
