Walter Frank Raphael Weldon. 18G0— 1906. 
25 
up such a heterogeneity even into Gaussian components was a problem not tlien 
solved, and one which has not since been solved. It was cruel fate that thrust such 
a problem on Weldon, and kept him over it for weeks. He was struggling with 
most highly complex mathematical difficulties, and actually beginning with a 
problem which a more highly trained mathematician would certainly have put 
on one side in the then state of statistical analysis*. 
The next portion of the Committee's work was far more successful — the 
" Attempt to Measure the Death-rate due to the Selective Destruction of Carcinus 
moenas, with respect to a Particular Dimension " (17). This formed the first report 
of the Committee, and was presented to the Royal Society in November, 1894. 
Weldon's general project in this case was, I believe, absolutely novel at the time, 
and embraces, I consider, the best manner still of testing the truth of the Darwinian 
theory. It consists in determining whether the death-rate is correlated with 
measurable characters of the organism, or, as he himself put it, " in comparing the 
frequency of abnormalities in young individuals at various stages of growth with 
the frequency of the same abnormalities in adult life, so as to determine whether 
any evidence of selective destruction during growth could be discovered or not." 
Thus stated the problem might appear an easy one, but it is the very reverse. 
How is the 'abnormality,' i.e. what we should now term the deviation from type, to 
be measured at each stage of growth ? What is to determine ' adult ' life ? What 
measure is there of the time during which the individual adult life has been 
exposed to the selective destruction ? Weldon undoubtedly chose the crab because 
of the facilities it offers for measurement. But its age then becomes an apprecia- 
tion based merely on the obviously close, but probably imperfect correlation 
between age and size. Further, the law of growth, complicated rather than simpli- 
fied by the moults, and the question as to how far the variability of the characters 
dealt with is aftected by growth combine, in the case of crabs, to form an exceedingly 
difficult problem. It is practically impossible to keep a sufficiently large series 
of crabs through the whole period of adolescence, and if it were possible, it is 
far from certain that the claustral environment necessary would not sensibly affect 
their law of growth. 
Looking back now on Weldon's paper of 1894, one realises its great merits ; it 
formulates the whole range of problems which must be dealt with bioraetrically 
before the principle of selection can be raised from hypothesis to law. Almost each 
step of it suggests a mathematical problem of vital importance in evolution, which 
has since been developed at length, or still awaits the labour of the ardent 
biometrician. On the whole, I think, Weldon came very near to demonstrating 
his point, but whether he did or did not scarcely affects the suggestiveness of 
the paper f. 
* We now know that some of the most skew distributions are given by the parts of flowers, and the 
problem propounded to Weldon was to resolve into a number, probably five or six, of such skew com- 
ponents a strikingly irregular frequency distribution for ray florets ! 
t Beading through the criticisms I communicated to him at the time, criticisms written purely from 
Biometrika v 4 
