328   Life and Letters of Francis Galton

thus twelve figures for coordinates and seven for types and we can communicate this in four `words' of five figures with one figure to spare. After the standard points have been put in on tracing paper, Galton suggests that tracings should be taken of the seven selected standard forms on to this paper very faintly; next

"to harmonise the whole tentatively with faint and brushlike strokes; lastly, with a free and firm hand to draw the outline through them." (p. 128.)

Now there is little doubt that Galton's original method of numeralising profiles allowed their reproduction with astonishing accuracy, and that his original six standard points permit of their accurate lexiconising. Only experience could determine whether the loss of exactness in this his final four-word method would not be at the cost of a considerable part of the certainty of recognition. Galton in his paper in Nature gives eight illustrations and says-with which any one examining the results would agreethat they are by no means deficient in resemblance to their originals.

"I think they are considerably more like to them than the sketches, usually printed in the illustrated newspapers, are to the public characters whom they profess to represent. They are, to say the least, of considerable negative value sufficing to eliminate at the rate of about nineteen out of every twenty individuals as not being the person referred to." (p. 129.)

It must be remembered that the resemblance provided is between a profile and a profile, not between the actual person and the four-worded reproduction of his profile. Dance almost in the manner of a caricaturist emphasised individual characters especially the nasal, and this I fancy renders in the illustrations given in Nature identification of the accurate profiles, and their rough reproductions, relatively easy; it would be a harder matter with the living subjects of the profiles. Only some experience could test the utility, but it would be worth testing as the police value would undoubtedly be large.

Galton fully recognised the limitations of these rougher methods, and noted that the next step to an accurate profile is a large one', requiring our four-word formula to be replaced by one of fifty or more words. Galton had numeralised many profiles in this more elaborate way and found that normal sighted persons who examined them at a distance of 12 inches in a somewhat careless way did not distinguish them from the originals. By such profiles it would be-possible to recognise the living. I am far less certain that the rough profiles suggested in the 1910 paper would be adequate, they certainly would not preserve anything in the nature of an artistic characterisation, which 50 to 80 word formulae undoubtedly achieve. Here we must leave the matter as Galton left it, until another scientific worker feels able to spend a like number of years and an equal enthusiasm on the analysis of portraiture.

1 " I do. not find that a general resemblance can be much increased by using one or a few more quintets or words." (p. 130.)