K. Pearson 
81 
In starting his investigations Weldon proceeded on very general lines, which 
it may be well to indicate. He was not inclined to accept the theory of unit 
characters, of allelomorphs and of pure gametes as capable of fully describing even 
the inheritance of the simplest characteristics. At the same time he recognised 
the importance of the segregation first pointed out by Mendel in the offspring 
of hybrids, though even here he was not prepared to make the segregating classes 
so distinct and so wanting in continuous variation, as some Mendelians have held 
them to be. He was convinced that in a sufficiently general theory of inheritance 
some place must be left for a normally arising percentage, however small, of variants, 
specially related to the distant ancestry. He was thus seeking for a mechanical 
explanation of latent characters, or in other aspects of reversions and even mutations. 
In starting his work Weldon was perfectly catholic as to the possibility or 
not of paternal and maternal chromosomes retaining their individuality from the 
moment of fertilisation up to the reducing division of the germ cells. He did not 
consider this point as at present absolutely settled by the cytologists. While the 
retention of individuality immensely simplifies the analysis, this did not deter 
Weldon from investigating at great length the numerical results flowing from 
supposing an interchange of chromomeres at every mitosis. 
The following rough notes of January 2, 1905, express the ideas then in his 
mind ; it must be remembered that they were not written for publication, but 
as suggestions for his own work. 
1. It is, I think, evident, from the facts of regeneration, that the theoi'y of a nucleus as 
composed of specific organic determinants is hopeless. It is also evident from the behaviour 
of eggs, such as those of Ctenophora, that a special structure leading to determinate fate of 
special portions, may exist in the cell body. 
2. The above facts do not invalidate conception of nuclear elements as a series of sti'rps, 
in Galton's sense, each containing something capable of exciting the development of any of 
the somatic characters, according to its position in the organism. 
3. It seems necessary to regard a stirp as capable of exciting, not only somatic characters 
like those of its parents, but characters like those of its more remote ancestors under certain 
circumstances. 
4. It is evident, from the facts of growth and regeneration, that the characters of any one 
stirp which become active in any one generation arc determined by the position of that stirp 
with reference to the rest, — i.e. by a process of the same nature as Mendelian "dominance." 
5. In an individual of pure race, the stirps will each contain the present and ancestral 
characters of that race. Assuming equality of numbers in germ cells which can fertilise each 
other (about which there is no real knowledge) the hybrid zygote should contain two sorts of 
stirps in equal numbers, each representing the race-characters of one parent. It seems possible 
on this assumption to develop a theory of nuclear division, which may give Mendel's results 
without eliminating ancestral influence, — i.e. without a theoi'y of the "pure" gamete. 
6. Such a theory would start by taking "chromomeres'' as units. A chromosome of n chromo- 
meres becomes entangled in the nuclear network : say there are in the zygote 2m chromosomes, 
there will be 2mn chromomeres in the resting nucleus. Before division these will be gathered 
into 2m groups for an ordinary mitosis, into «i groups for a maturation mitosis, — the maturation 
division being afterwards heterotypic. 
Biometrika vi 11 
