24 
Walter Frank Raphael Weldon. 1860—1906. 
demanded. Trial and experiment were peculiarly needful in 1893 ; the statistical 
calculus itself was not then even partially completed ; biometric computations were 
not reduced to routine methods, and the mere work of collecting, observing, ex- 
perimenting, and measuring was more than enough for one man. Weldon with his 
"volume of life" was eager to do all these things, and run a laboratory with perhaps 
sixty students as well. He was impatient because the probable errors of biometric 
constants, on which tests of significant differences depend, were not at once forth- 
coming; he wanted the whole mathematical theory of selection, the due allowances 
for time and growth, the treatment of selective death-i'ates and the tests of 
heterogeneity and dimorphism settled in an afternoon's sitting. The Committee 
did not possess a mathematician to put on the brake, and Weldon attempted too 
much in too short a time. Each week Weldon had new and exciting problems, he 
thrust them upon his friends, demanded solutions, propounded solutions, and was 
never discouraged when difficulties were pointed out and time asked for. 
One of the first subjects to be taken up by the new Committee was to test 
whether the method of resolution into two Gaussian curves, which suggested 
dimorphism in the Naples ci'abs, would be helpful in confirming a similar 
dimorphism said to exist in the herring. Several thousand herrings arrived at 
University College, a measurer was trained to deal with them, and the variability 
of a wide series of characters determined. The distributions came out skew, and 
Weldon was intensely hopeful that statistical evidence of dimorphism would be 
forthcoming. Instead of this, the analysis showed dimorphic Gaussian components 
to be impossible. This result was a great disappointment to him, and, I believe, 
to the Committee. I could never understand why. A most extensive and 
valuable series of measurements had been made, which in themselves were well 
worth publishing. It had been shown that simple dimorphism of a Gaussian kind 
certainly did not hold for these herrings; in all probability it was a typical 
case of skew frequency, which would have been most valuable as adding to the 
known instances, and aiding statisticians eventually to classify such occurrences. 
But Weldon, and, I presume, the Committee were disheartened, they had been 
searching for dimorphism and had not found it. Tlie herring data were put on one 
side by Weldon, and as far as I know have never been published. It is much 
to be hoped that they may some day be resuscitated from the archives of the 
Committee (16). 
The next point that I personally became aware of in relation to the Evolution 
Committee was Weldon's attempt to solve the problem of subraces in the case of 
the ray florets of ox-eyed daisies. I am unaware who brought the material before 
the Committee, but it was obviously heterogeneous in the highest degree. There 
was no evidence at all that any attempt had been made to allow for seasonal and 
environmental effects, and whatever truth there may be in a tendency of the 
modes to fall into Fibonacci groups, we now know that varying season and period 
will produce within a certain range almost any mode in this flower*. To break 
* Biometrika, Vol. i. pp. 305, 309 et seq. 
