326 
On the Fimdamental Conceptions of Biology 
whatever be the nature of the distribution of the character in the fraternity. The 
coefficient of parental inheritance judged from upwards of 50 cases in insect, 
animal, and plant life is about '-i to "6. We may conclude then that whatever 
character we choose to deal with we shall find "discontinuity" between parent and 
offspring. Such " discontinuity " has nothing specially to do with vital deviations 
or with inheritance, it is a simple fact of the statistical distribution of any two 
quantities not perfectly correlated, — e.g. the number of trumps in two partners' 
hands at whist. 
(ii) In collecting the materials for a study of variation as defined by 
Mr Bateson we must give particulars of hotli parent and offspring. We do not 
know whether a character in the offspring is a variation or not until we have a 
knowledge of the parent. The biometrician's definition of variation involves only 
a knowledge of the distribution of a character in a population ; its relation to the 
distribution in the parent population involves a study of heredity. Mr Bateson 
includes under variation three distinct studies : (a) a change of type between 
parental and filial population, (6) a change in variability (in the biometrician's 
sense), and (c) an investigation of heredity. 
Mr Bateson scarcely mentions heredity throughout the whole of his bulky 
volume. He does not compare parent and offspring, and thus in not one of the 
cases cited by him is there evidence whether or not the instance described is one 
of variation or not according to his own definition of variation ! That is to say he 
tacitly drops the " Individual Variation " as he has defined it, and which he 
suggests is the source of " Specific Differentiation " and goes off to something else. 
In the bulk of cases this consists in comparing two or more members of a 
population, — not parent and offspring, — and treating their difference as a variation. 
This divergence between theory and practice rendei's it impossible to follow 
Mr Bateson when he uses the term " variation " in his criticisms on my 
memoir. 
Discontiiiaity. We have already seen that when correlation is imperfect, 
whatever be the distribution of two characters, then statistical theory shows us 
discontinuity, and measures its average value. If this was all Mr Bateson meant 
by "discontinuity," he would be in the biometric camp. But in the course of his 
Materials he gives several further definitions, to which I must refer : 
" The chief object, then, with which we shall begin the Study of Variation will 
be the determination of the nature of the series by which forms are evolved." — 
" The first questions that we shall seek to answer refer to the manner in which 
differentiation is introduced in these series. All that we know is the last term of 
the series. By the postulate of Common Descent we take it that the first term 
differed widely from the last, which nevertheless is its lineal descendant : how then 
was the transition from the first term to the last term effected ? If the whole 
series were before us, should we find that this transition had been brought about 
by very minute and insensible differences between successive terms in the series. 
