192 Variation and Correlation of the ffnman Skull 
J. Barnard Davis' Thesaurus Graniorum, a catalogue of the human skulls 
in his own collection, published in 1867, deals with the skulls of a great variety 
of races, but contains no important series of any one race. He divides the races 
of the British Islands into "Ancient British " and " Modern British"; the former 
he subdivides into Aboriginal, Ancient Roman, and so on, and of these ancient 
races he catalogues 85 skulls; the latter he subdivides into English, Welsh, 
Scottish, and Irish. Of his modern British races he has 135 skulls in all, of 
which 81 are English, male and female, but for some of these no measurements 
are given, no doubt owing to the impei'fection of the material ; and while some 
are really modern, others are obviously hundreds of years old. Davis, when 
possible, took 18 measurements on each of his skulls, and if he had possessed 
a larger and more homogeneous series of English skulls he would have placed 
extremely valuable material at our disposal for purposes of comparison. 
Twelve years after the publication of Davis' Thesaurus, Sir William Flower 
issued Part I. of the Osteological Catalogue of the Royal College of Surgeons, 
dealing with "Homo Sapiens"; in this work 1018 skulls are catalogued, of which 
only 278 are European, ancient and modern, and of these 278 only 63 are from 
the British Isles, and range from "Ancient British" to "Modern." Here again 
there is practically no material for comparison. One of the most important series 
in the catalogue appears to be that of 114 modern Italians, mostly males, and as 
11 measurements were taken, and 5 indices calculated, on each skull when possible, 
there is in this case a considerable mass of first-class material ready for reduction *. 
No further work of importance on British craniology appears to have been 
issued in this country until 1903, when Sir William Turner published, in the 
Transactions of the Royal Society of Edinburgh (Vol. XL. Part ill. pp. 547 — 613), 
a paper on Scottish skulls. This paper contains the measurements of 176 male 
and female skulls, and is illustrated by photographs : the number of measure- 
ments taken on each skull is no less than 33, and 8 indices are calculated, and 
the mean and maximum and minimum values of each character are given. The 
great value of the paper as a contribution to craniological literature is obvious, 
but unfortunately for our purposes, the skulls which it deals with were obtained 
from different parts of Scotland, and range from quite unknown antiquity to 
modern specimens from the dissecting rooms; they cannot, therefore, be considered 
a homogeneous series, and this conclusion is confirmed by an examination of 
their variability. For example, I find that the standard deviation of the glabello- 
occipital length in these Scottish skulls is 7 418 for males and 7-151 for females, 
which shows a much higher variability than is found in fairly contemporary 
individuals of homogeneous races "}*. I shall give later in summarising my results 
the chief biometric constants of Turner's material. 
* See, however, Pearson: Chances of Death, Vol. i. pp. 332 .and 353. 
t See postea, p. 206, and Biometrika, Vol. ii. p. 346. 
