372 Darwinism, Biometry ami Some Recent Biology. I. 
how all these pure lines came into existence in a single local race on any philo- 
sophical scheme of evolution ; but it would not affect his treatment of inheritance 
in populations. Why then should he criticise the work of Johannsen, Hanel or 
Jennings on " pure lines " ? Simply because the experiments made by these 
biologists are in his opinion insufficient to provide a basis for the sweeping 
inferences drawn from them, and further because they contradict other biological 
v^'ork, which he thinks just as important and valid. The assumption made by 
the pure linist is that in reproduction from a single individual the same gametic 
constitution is transferred to every offspring. The biometrician sees no sufficient 
justification of such a view in our present state of knowledge; two seeds from 
the same self- fertilised plant may differ in gametic constitution, and this differen- 
tiation may appear in the correlated somatic characters. Two Aphides from 
the same mother, according to Warren, do show such differentiation. Does or does 
not differentiation follow the division in Paramecial We do not know, because 
no heredity of the "pure lines" has yet been shown to exist for the character 
selected by Jennings in the only way that would satisfy a statistician, i.e. a corre- 
lation table of the pure line parents with their offspring. On the other hand, 
having convinced himself that size of litter, for example, is not inherited to any 
appreciable extent in mice, but that colour and area of colouring are, he is not 
likely to accept the conclusion of Pearl, based on the non-inheritance of fertility 
in hens, that " the chief if not the entire function of selection in breeding is to 
isolate pure strains from a mixed population. It is found in actual experience 
impossible to bring about by selection improvement beyond a point already existing 
in the pure (isolated) strain at the beginning" {loc. cit. p. 79). It is difficult 
to understand where in the pure strain at the beginning there existed the chief 
characteristics of any of our domestic animals of to-day. Such statements appear 
too sweeping in the light of any evidence at present available ; they discard the 
Darwinian spirit of slow and patient inquiry, at the same time that they destroy 
the Darwinian theory of selective evolution*. 
That selection in a pa^^ticular direction cannot go beyond certain limits is of 
course a commonplace of breeders, indeed it was illustrated at a very early stage 
by one of the first of biometricians. Sir Francis Galton himself, in his discussion on 
organic stability. The selection of any one character changes all correlated 
characters, and a condition is rapidly reached at which the organism is either 
incompatible with its environment, or possibly its germinal determinants become 
unstable and assume a new form of equilibrium. 
(5) While Jennings himself, as I have said, provides us with no data upon 
which we can test (i) whether the character he is dealing with is really inherited 
in Paramecia, (ii) whether in an unselected population the degree of resemblance 
of parent to offspring is no greater than that of grandparent — while in fact both 
Pearl and he appear to confuse non-heredity of a character with pure line heredity 
* Pearl, R. and Surface, F. M. : "Is there a cumulative effect of Selection?" Zeitsehrift filr 
induktive Abstammungs- und Vererbungslehre, Bd. ii., S. 257, 1909, answer a very big question from the 
same egg data. 
