328 
On the Fundamental Conceptions of Biology 
size of the cephalic horn. He states that " he is not acquainted with evidence 
as to the course of inheritance in these cases, and I do not know whether ' high ' 
and ' low ' males may be produced bj one mother " (p. 40)*. In other words 
Mr Bateson admits that he is not considering either variation or discontinuity in 
the light of his own definitions. Discontinuous variation means now for him a 
dimorphic distribution of a character in one generation, even when we are quite 
ignorant of whether the immediate or the evolutionary ancestry proceeded by 
continuous or discontinuous series in the sense of the earlier definition. Seekinof 
for further light, we find Mr Bateson suggesting " that the separation of the males 
into two groups was a case of characters which do not readily blend, and are thus 
exempt from what Galton has called the Law of Regression." Since eye-colours 
do not readily blend, and as I have shown in 20 to 30 sets of relationships, 
undoubtedly obey the law of regression, this does not throw more light on 
Mr Bateson's second definition. It is, however, an illustration of what I have 
above referred to, Mr Bateson's confused state of mind as to regression. On p. 42, 
however, Mr Bateson tells us that the existence of intermediate links between the 
types of dimorphic forms " does not touch the fact that the Variation may be 
Discontinuous, for we are concerned not with the question whether or no all 
intermediate gradations are possible, or have ever existed, but with the wholly 
different question whether or no the normal form has passed through each of 
these intermediate conditions. To employ the metaphor which Galton has used 
so well — and which may prove hereafter to be more than a metaphor — we are 
concerned with the question of the position of Organic Stability ; and in so far as 
the intermediate forms are not or have not been positions of Organic Stability, in 
so far is the variation discontinuous." 
This is the third definition of discontinuity implying a new definition of 
variation given within fifty pages ! 
The first depends on variation defined as a deviation between parent and 
offspring being finite. This is true for many, and possibly for all living forms 
whatever their distribution. It is simply a statistical result of a correlation 
coefficient less than unity. 
The second definition refers to one generation alone, and depends upon a 
recognition of statistical heterogeneity in the distribution of the population. 
Mr Bateson apparently supposes such heterogeneity is associated with bimodal 
polygons. 
The third definition has nothing whatever to do with variation or heredity 
as far as I can understand. If m-^ be the mean for one generation and the 
mean for the next, the variation is to be treated as continuous or discontinuous 
according as 111.2 — "h is sensible or insensible. Now I should consider — nii as a 
measure of the change in type produced by environment, natural selection, or other 
* What is equally or more important, we do not know if they may be due to one father, or indeed 
produced at one mating. 
