Recognized HTML document

316   Life and Letters of Francis Galton

These other points involve merely suggestions of slight changes: (i) Surely you have inverted the order of our Huxley Lectures. My lecture was in 1903, but I think yours was two years earlier and not the year after. In fact you put the right date on the top of p. 7. So here you will see, you, not I, led the way ! (ii) Will you think me ungrateful, if I ask you not to praise me quite so much? It is natural that I should feel and speak strongly about your work, because I owe so much to it for method and suggestion, but if you praise me 'tis as you branded your own herring as of peculiar virtue. Please re-read in this sense pp. 2-3 and 9. I know you will grasp how much I appreciate all your praise, but others possibly will not see it from the same standpoint. (iii) Would it not be well to free yourself on p. 21 from your unit by measuring your A and B in terms of their standard deviations? You thus avoid the difficulty which occurs to the mind coming fresh to the subject of the index of correlation* depending on the units used-lbs. weight, inches of stature, etc.-and thus providing no comparable ratio, but one varying with the units. If you agree to measuring in terms of your standard deviations as units, all values of the index of correlation are comparable and lie between - 1 and + 1. All this is, of course, very familiar to you [see, indeed, our pp. 5, 51, and Vol. ii, p. 393, but it passed from Galton's mind when preparing his manuscript].

You would bring it home to your hearer and save him solve difficulty, if you gave a hint that the coefficient of correlation lies arithmetically between 0 and 1, and has only a numerical value, being independent of scales, such as those of weight, length or units of pigment intensity.

I wish I could come to Oxford to hear and possibly help you. I would if it were July, but I am under rather high pressure, and one of my ears is giving me much trouble and exciting the neck in some way. Affectionately, KARL PEARSON.

I shall hear how the lecture goes, I have no doubt; but I should like to hear when you have a chance how the lecturer gets through the exertion, which is another matter.

Galton was not fit to speak at Oxford, one of the reasons being the accident referred to in the following letter. The lecture 'was read by Mr Arthur Galton.

42, RUTLAND GATE, S.W. May 27, 1907.

MY DEAR KARL PEARSON, I have now a bout of ill fortune. Feeling particularly well I went on Friday to Bushey Park and returned a bit tired but nothing more. However a horrid bout of bronchitis came on and on Saturday night 12.30 on getting out of bed I rested in the dark on an insecure table with crockery and tumbled on the floor with such a clatter and bound with the bed-clothing dragged after me. I had not the strength to free myself so, there I lay till 6.30 when the household stirred and the united strength of three maids got me into bed with a very sharp sciatica. It is possible that I may be fit to go to Oxford on June.5 but I feel practically sure that my lecture must be read for me.

* I use here the term employed by Galton in his lecture ; by 1907 the name "coef coefficient of correlation" was in general use.