392 



NATURE 



{August 2 2, 1889 



Nor do we ever find that amputation of a limb, or loss of the 

 cortex of the kidney from Bright's disease, is followed by cor- 

 responding anatomical deficiencies in children. In the African 

 transported to northern latitudes, the dark skin persists through 

 indeterminable generations — provided there is no cross-breeding 

 — but the endemic diseases of his lace are not transported with 

 him. 



"Hitherto all reasoning upon the heritableness of diseases has 

 proceeded on the tacit assumption that morbid changes are sub- 

 ject to the same laws of vital action as healthy changes. It has 

 been discovered, hoWever, that the two are dissimilar both in 

 nature and in the circumstances of their genesis. The traits we 

 every day recognize as inherited are the results of an infinity of 

 co-ordinate actions. There may be instanced the bony frame- 

 work of the face, the colour of the iris, the gait, special mental 

 aptitudes. All these, and attributes of the same order, represent 

 a vast integration of forces, groups of organized energies. It is 

 this organization which gives them individuality and makes 

 their hereditary transmission possible. They are, in other 

 words, self-existent, having been independent of the original 

 conditions out of which they grew." 



The conclusion, deduced from evolutionary principles, that 

 non-congenital diseases and injuries are not inheritable, might, I 

 think, be supported inductively from the facts of medical obser- 

 vation, and it is most interesting that the results of Dr. Weis- 

 mann's investigations are confirmatory. But from what has been 

 said it clearly appears, in harmony with the annotation I 

 commenced by quoting, that the non-inheritance of injuries 

 is no evidence of the non-inheritance of acquired useful 

 variations. C. Pitfield Mitchell. 



24 De Vere Gardens, W. 



Classified Cataloguing. 



Whenever a collection has been catalogued anew, and all 

 the numbers are in the museum order of the specimens, the 

 placing of additions at the end, without any sequence but that 

 of acquisition, always seems a melancholy collapse of the order 

 just established. So strongly is this felt that some curators even 

 enter additions with the same numbers as similar specimens, 

 distinguished by letters, as 3247a ; but, as formerly in the British 

 Museum, this system breaks down when such additions far out- 

 number the original series, and we reach figures like 2>'^\']fy. At 

 the same time this is an approach to an entirely different and 

 logical system of cataloguing, which ought to be considered. 

 Another stage of arrangement has been by appropriating so 

 many thousand numbers to each branch, so that the articles of 

 one class may have contiguous numbers. 



The complete system of cataloguing which has been thus felt 

 after, and sought for, is what may be called "fractional cata- 

 loguing," treating all numbers as decimal fractions and arranging 

 them accordingly. Thus 21 '765, 21 77, 21 '8, and 22 might appear 

 as successive numbers in a catalogue : the numbers being ar- 

 ranged solely by their successive order of the left-hand figures, 

 regardless of the length of the number. By this system, there- 

 fore, any quantity of additions can be brought into their right 

 order without disturbance ; fifty new specimens like No. 371, for 

 instance, being numbered 371 "Oi to 371 "50. 



The first two or three places of the number will therefore 

 indicate the nature of the specimen in any given catalogue : and 

 this leads at once to the desirability of all collections having a 

 similar numerical basis for their catalogues, so that if all the 

 parrots, for instance, begin with 56 in one collection they should 

 do so in all other museums. 



The first step therefore in classified cataloguing would be to 

 agree on a set of 100 or 1000 numbers, to subdivide each branch 

 of science, the distribution of the numbers being partly settled 

 by the average number of specimens, partly by natural divisions. 

 Thus in mineralogy, elements might be 001 to 099 ; binary com- 

 pounds 100 to 299 ; silicates 300 to 799 ; non-metallic acid salts 

 800 to 899 ; metallic acid salts 900 to 999. In all museums, then, 

 silicates, say of lime, magnesia, and alumina, would begin with 

 61, the different species being marked 610 to 619, and varieties 

 and individual specimens numbered with additional decimals 

 following these bases, e.g. 61 5 "47. The set of numbers in each 

 science would be best fixed by a committee at some International 

 Congress, so as to insure general acceptance, like the scheme of 

 geological colouring. 



The disadvantages of this system would be — (i) that the 

 catalogue would have to be kept like that of a library, subject to 



additions at any point, and therefore on slips which could be 

 transferred ; and (2) that the total number of specimens would 

 not be known except by counting. These are not serious 

 difficulties, and the following advantages seem to entirely 

 out-weigh them. 



(i) The numbers would indicate to all students the nature of 

 the specimens quoted in any collection. (2) The catalogue 

 would be classified in natural order throughout, so that all 

 similar specimens would be described together. (3) The 

 numbers in the museums would be in order from end to end. 

 (4) Any specimens moved could be rearranged by unskilled 

 assistance, solely by the numbers. (5) Any object in the 

 catalogue or hand-books could be at once found in the museum 

 by its number. (6) A great help would be given to the arrange- 

 ment of minor museums by having a uniform scheme of 

 cataloguing fixed. (7) The numbers being in constant use would 

 soon form technical symbols for species, a short-hand briefer 

 than chemical symbols even, and applied to all sciences ; and 

 also a valuable key to the memory. 



Bromley, Kent. W. M. Flinders Petrie. 



Head Measurements of Students at the University of 

 Cambridge. 



I WAS rather too precipitate when I stated that the figures 

 relied on by Mr. Galton were totally inadequate to support his 

 conclusions ; for, as regards the second of them, viz. that the 

 "high honour" man has a head perceptibly larger than the 

 " poll " man, the evidence is fairly strong ; but with regard to 

 the other three conclusions, referring to the growth of heads, I 

 must repeat what I have said. In the light of the discussion 

 given below of a large number of observations, I cannot even 

 admit that the tables and curves given in Mr. Galton's paper 

 (see Nature, vol. xxxviii. p. 15) give even "an approximately 

 true idea of what we should find, if we had the opportunity of 

 discussing a much larger number of observations. " 



Having heard that all the measurements taken have been in- 

 dexed for reference, I went to the laboratory, and, by the kind 

 permission of the custodian, copied out the head measurements 

 of fifteen individuals, each of whom had been measured at least 

 five times. In one case, measurements had been taken at 

 seventeen times. The average number was 7"i. 



Since the first case quoted in my last letter forms one of these, 

 I had better point out that Mr. Galton's objection to it is un- 

 sound. He notes a grouping of the observations, which makes 

 him suspect that "some peculiarity in the shape of the head 

 caused doubt as to the exact line of maximum height." But the 

 observations of height are 5-2, 5-3, 5-4, 5-5, 5*5, and 5-6; and 

 show no grouping. Mr. Galton must have meant that the 

 calculated products were grouped. This is the case, but could 

 not be due to the cause he suspects, for that would cause 

 grouping in the simple heights. 



The fifteen series of measurements fully bear out the conclu- 

 sions which I drew before from two only. The measurements 

 of width vary o'l, o'2, or o"3 inch, those of length vary to the 

 same extent, and in one instance up to 0*4 inch, while the 

 height in most cases varies o '4 or o "5 inch. In only two cases 

 does it vary so little as o'2 inch, while in one case it varies 0*7 

 inch. The last case (where the figures are 5-4, 5'6, 5-1, 5'8, 

 5'5) is partly accounted for by the fact that the first three obser- 

 vations were taken by one observer, the other two by a second. 

 (The statement in Mr. Galton's original paper that all the 

 measurements were taken by one observer, must have been due 

 to misinformation.) I have calculated the probable error of each 

 observation of the height of head for each series of observations, 

 using the approximate formula 



\'n{n — I ' 



and I find it on the average 0'095 inch. Since the average 

 height is less than 5-5 inches, this error amounts to 17 per cent. 

 If the error in length a> d width were each half of this, the 

 probable error of the product would be about 2 per cent. 



To test whether any of the variation found is due to actual 

 growth, and not to accidental error, I have used the following 

 method. Arrange all the measurements of any one individual 

 in the order of the dates on which they were taken, and separate 

 them into two equal groups. Take the mean measurements of 

 the first set, and put opposite them the mean of the dates ; then 



