Recognized HTML document

398   Life and Letters of Francis -Galton

anthropometric laboratory, and this arrangement lasted till 1890. He insists largely on the value of such anthropometric records for the identification of individuals, and cites the claims made for them by Bertillon, Topinard and Herbette. He describes further the index-system of Bertillon

"Whether all that was claimed for the power of M. Bertillon's system, on purely theoretical grounds and in his earlier publications, can be sustained, may fairly be questioned; but there can be no doubt that a series of measurements must be of considerable service as supplementary evidence, either that a person is really the man he professes to be, or negatively that he is not the man for whom he is taken. In speaking of these matters it is impossible not to allude to the Tichborne trial, and the enormous waste of money, effort, and anxiety which might have been spared, had Roger Tichborne passed through an anthropometric laboratory before he went abroad. It would be a reasonable precaution for every person about to leave his country for a long time, having regard to the various accidents of good or ill fortune, to be properly measured, and to leave a copy of his measurements in the safe keeping of an anthropometric laboratory." (p. 252.)

"Another and very important question is as to the degree in which the several bodily proportions that are measured may be looked upon as independent variables. The stature is related with the length of the foot, and with that of the forearm, and we should expect a still closer relation to exist between any two of these taken together, and the third. We have yet to learn the proportion between the number of the elements measured and their value for purposes of identification. The supposition that they may be treated as independent variables, which lies at the bottom of some of the earlier estimates, such as that on page 22 of the Conference at Rome' headed 'TEtendue infinie de la Classification,' cannot be accepted as correct.

The whole subject of 'Personal Identification and Description' forms an important chapter of anthropological research, and it is one on which I hope before long to be in a position to offer some views of my own." (p. 354.)

The careful reader of this passage will note how Galton is beginning to realise the problems of multiple regression, and to see that with a large number of correlated variables, there is a limit to the intensity of the multiple correlation coefficient, which cannot be indefinitely increased by increasing the number. We also see how he is studying the correlation of bodily characters and gradually advancing beyond the indexing by such characters to a method of his own-the identification by finger-prints.

Galton's fourth presidential address, that of 1889, we have already discussed (see our p. 383), and the reader who turns back to our account will have some appreciation of how the ideas of the 1888 address had been developed in the interval. Besides working out the fundamental ideas concerning correlation and heredity to which Galton was led by his "Family Records" and the data from his Anthropometric Laboratory, he contributed numerous short papers on anthropometric and statistical topics to various journals during this period and later at intervals.

H. MISCELLANEOUS STATISTICAL PAPERS

(i) The Horse. The horse had always been a favourite animal with him, notwithstanding his experiences with the camel in Syria and the ox in Damaraland, and no less than four of his papers treat of the horse under various conditions in addition to his paper on standard photographs of horses

1 Louis Herbette and Alphonse Bertillon: "Les Signalements Anthropometriques," Conference faite au Congres Penitentiaire de Rome, Masson, Paris, 1886.