Eugenics as. a Creed and the Last Decade of Galton's Life 257
graph of the skull that you send is exceedingly good, and is I presume (together with the rest) taken under standard conditions, and selected in some way free from bias, other than what may be clearly stated about them as intended to be conveyed by the word " English." English unless narrowly limited includes so great a diversity of type:-dark and fair, Cornish, Sussex, Midlands, Yorkshire, Welsh, Scotch, Irish, &c. &c.-ill-fed and well-fed, educated and uneducated, etc. etc.-that it is very difficult to deal with English as a whole, except by 'taking homogeneous subgroups*. I found this emphatically the case with my S. Kensington anthropometric data. Out of many thousand cases I failed to form a single homogeneous (quasi-homogeneous) group that satisfied me. If you think that your collection is fairly free from this difficulty, please tell me what you think the cost of printing them would be, and I will see if it be possible for me to afford it. It is most desirable that some standard and unquestionably useful work-obviously useful to biologists-should appear in Biometrika.
About the finger-prints, what I sent was a mere scrap and would require a great deal both of explanation and of collateral conclusions. The lump at the commencement of the series is to me of the greatest interest, for it emphasises the fact that the patterns do not form a continuous series, but a group or order composed of many sub-groups or 'species ; each of these has a curve of frequency of its own. They are in some sense convertible, and they form hybrids, but the arches are far more "pure blooded" (so to speak) than any of the others. They are antipathetic to whorls. An arch on one forefinger is associated with a whorl on the other only once or twice in a hundred cases, and then only imperfectly. Then there is the case of radial and ulnar slopes, and their connection with whorls. We have in fact a menagerie of different creatures, breeding promiscuously, and yet at all times divisible into a limited number of definite types, each with its own law of frequency, whose statistical proportions between themselves seem to be constant. Its study has therefore a very close bearing on the evolution of species (as indeed I pointed out in my first paper on Finger-Prints iu Phil. Trans.). This study has the great advantages (1) that age has no effect on the patterns, when the ridge interval is taken as unit of measurement, and consequently (2) that it would be easy to get and to use family prints to 3 and even 4 generations, (3) that the data when once obtained are free from all error of measurement, for they are themselves the things to be measured f. I send prints of my own fingers, which are a worse example by far than the generality of those one might get, chiefly because the wrinkles of age leave numerous gaps in the form of - white streaks, and also because I have smeared them by manipulations immediately after they were made, but they will serve to explain the dimension measured in the table I sent. The loops are troublesome only in the sense that the very best dimension is hard to define; on the other hand many reasonably alternative dimensions give practically the same result. The measure desired
* [The skulls in question all came from a single 17th century pit in Whitechapel, and were reasonably homogeneous and close to similar series from Liverpool Street and Farringdon Street. The photographs were the first of the series of standardly orientated crania on a large scale which have since then continuously appeared in Biometrika_ Galton's offer was spontaneous like several others, but not accepted. "Of course I could not think of your aiding us further at present. We made up the loss to the reserve fund with your Darwin Medal grant, and it left £30 to the good which might be reasonably expended on illustration if you approved. ...The photographs were all taken the same distance from the objective and in the same manner for each aspect, but different aspects bad to be treated rather differently- a profile on a smaller scale than a frontal view, etc. The difficulty of getting a ' mean' focus on a solid body must cause some variation,' however, even in the, distance. On the whole, I think, photographs of skulls must be taken to represent qualitative characters, which are after all, if . indescribable, realities. I have tried a good deal, but do not believe that cranial photographs will ever serve usefully purposes of measurement .... I hope you will come back fit and well for climbing ' May hill,' which an old medical friend always describes as the great task of the year. I am going to Newbury to meet Weldon to-morrow to talk over Part III, while I hunt for Easter quarters. We want to be near Oxford, Weldon for the mice and I for Weldon." K. P. to F. G., March 20, 1903.
I think Galton must mean here that the stored data are free from error of measurement. Whether we take head measurements or finger-print measurements (and Galton is speaking of quantitative not qualitative classification) the measurement must be taken once.]
T G 111 33