2 74 



SCIENCE. 



[Vol. XX. No 510 



rule are much the same as those enumerated in the above para- 

 graphs. Of the three pronunciations of this termination -ide, ide, 

 and Ide, in varying degrees of usage amongst us, the second ap- 

 peared undoubtedly to be the most preferable; -ide is an uncom- 

 mon, almost unnatural, pronunciation of the vowel in English, 

 although it would bring our usage into unison with that of Euro- 

 pean countries, and simplify phonetic values for tlie ears of for- 

 eigners; -ide leads frequently to confusion with -if e. and is the 

 value of i farthest removed from European usage; -Id approxi- 

 mates closely to the Continental i, into which it is easily length- 

 ened, is readily recognized by the foreign ear, is not confused with 

 the termination -ite, is in line with present phonetic progress, and 

 has the backing of authority and usage. The short sound of i 

 naturally dictates the dropping of the final e. "According to 

 Smart and Cull, chemical terms ending in -ide — as bromide, 

 chloride, etc. — should be pronounced with the ilong; but all 

 other orthDepists are unanimous in making the vowel short; and 

 the propriety of the latter mode of pronunciation is established 

 by the fact that this whole class of words is not unfrequently 

 spelt without the final e, thus bromid, chlorid " (Webster's Dic- 

 lionary, " Principles of Pronunciation," p. xliv.). 



In conclusion, it may be said that the chemical section of the 

 American Association recognizes the fact that there is still room 

 for advancement in the path of phonetic reform, and that ques- 

 tions may still arise with regard to divergent usage or defects in 

 existing rules. The task of collecting and collating such questions 

 and of presenting them at a later date to the Association for 

 action has been assigned to Professor Jas. Lewis Howe of Louis- 

 ville, who will gladly receive all information, suggestions, or 

 propositions pertinent to the subject from those interested in the 

 perfecting of our chemical nomenclature. 



Calculated from Tables by Bowditch of Heights of Boston 

 School-boys. 



BOSTON SCHOOL-BOYS. 



BY FRANCIS GALTON, F.B.S., LOHDON, ENGLAND. 



Numerous results may be shown to flow from the excellently 

 arranged data in the valuable memoir of Professor H. P. Bow- 

 ditch on the Growth of Children (Twenty-Second Annual Report 

 of the State Board of Massachusetts, Boston, 1891). Permit me 

 to draw attention to two of them. 



It is necessary to premise that the method was adopted by him 

 of describing classes by nieaus of eleven percentiles, but, for the 

 present purpose, three are enough, namely, the 10th, 50th, and 

 90th. In other words, it is sufficient now to deal with the statures 

 of the persons who occupy those posts in any class along whose 

 length 100 posts have been marked at equal intervals. It follows 

 that 10 per cent of the whole class are shorter than the lOch per- 

 centile and 90 per cent are taller. These conditions are reversed 

 in respect to the 90th percentile; as for the oOth, it is the median 

 Talue, which one half of the class falls short of and the other half 

 exceeds. The median in most series differs little from the arith- 

 metical mean, and may be used instead of it, as a serviceable 

 standard of comparison. 



The variability of a series may be measured by the difference 

 between any two named percentiles. The wider these are apart 

 the more is the scale magnified ; on the other hand, the less trust- 

 worthy does the measure become. In the present series we can 

 with propriety use the difference between the 10th and the 90th 

 percentiles, but we cannot in all cases, owing to the paucity of 

 data, use that between the 5th and the 95th ; the former will there- 

 fore be here adopted as the measure of variability. 



In order to compare on equal terms the variability in stature of 

 growing boys at different ages we must so reduce their measures 

 that the median shall in all cases be the same. It is customary 

 for this purpose to take the median as 100, but there is more 

 significance in the results when it is taken at a value that repre- 

 sents the average stature, or thereabouts, of male adults. Here 

 it will be taken at 67 inches. In the following table the 10th and 

 90th percentiles for the several ages are those given by Bowditch, 

 after multiplying them by 67, and then dividing the result by the 

 median stature at that age. 



On examining the columns of differences, we find a remarkable 

 increase in the differences between the 10th and 90lh percentiles 

 during the interval between the ages of 11* and 15i years; that 

 is, of boys who at their last birthday were 11 or 15 years old. 

 The period in question is that during some portion of which the 

 growth is apt to be temporarily accelerated, but the precise epoch 

 of acceleration differs; some boys being more precocious than 

 others. Consequently the variability among boys of the same 

 age, between the ages of \\\ and \^\ years, is greater than at other 

 times. The point to which I wish now to direct attention, is the 

 much greater variability during this period of the children of 

 Americans than of those of Irish, for which it seems difl[icult to ac- 

 count. It can hardly be owing to variations of nurture, because 

 its influences would probably be greatest on those classes who 

 were least assured in their habits of life; now it is difBcult to 

 suppose that the Irish in Boston are, as a class, better established 

 and more well-off than the Americans. As regards the effects of 

 race, it is true that the Americans are more mixed in origin than 

 the Irish, but we should have expected purity of race to manifest 

 itself by a reduced variability Tat all ages, and not only at the 

 particular period we are considering. However, it seems to be 

 otherwise, and that the great variability of American children at 

 the time in question may really be due to their mixed ancestry. 

 In confirmation of this variability being a racial effect, we note 

 how much earlier the epoch of its increase sets in among the chil- 

 dren of Americans than among those of Irish, the difference 

 amounting to at least one year. Anyhow, these statistics suggest 

 the possible existence of an hitherto unobserved physiological 

 difference between the children of the Americans and of the Irish, 

 which might repay investigation. 



A considerable agreement will be found in the figures contained 

 in each of the four columns of percentiles in the table; their 

 variations ranging through 1.2, 1.9, 0.7, and 1.3 inches, respec- 

 tively. In other words, they range between limits that are hardly 

 more than one inch on the average apart, while of course the 

 range in other percentiles that are nearest the meiiian is pro- 

 gressively smaller, till at the median itself the range is nil. There 

 is, therefore, a fair approximation towards constancy in the ratio 

 between any given percentile and the corresponding median that 

 holds good for all these ages. It follows that if we are given all 

 the eleven percentiles of stature that are found in Bowditch's 

 memoir, together with the median heights for the several suc- 

 cessive ages, we should have sufficient data to reproduce, in a 

 roughly approximate way, the entire table of distribution of 

 growth. The variability and the median are not such independent 



