K. Pkailson anü A. Lke 361 



It. is Hol t'(ir a uiiuncnt suol;csIi'iI tlial tlic iiislructioiis or sdiodulc form are 

 ideal ; Ihcy ai'c ot coiirse (ipcn \i> criticisin nl a vaiictv dt' kiiids. ISut lln'}' were 

 not Kcttk'd withiuit cohsideralilc tliouj^dit and a dutiiiite reasmi l'oi' cach poiiit 

 stati'd. Thiis fnll gi-owlh is not rcaolied at hS ycars of ao-e, perhaps not tili 25. 

 TIk- growth, howcvcr, fi-oni IS t-o :25 is rclativdy sinall, ailhoiigh sensible, anfl by 

 fixinLf our liniit at i'}, \ve tbnnd a very laioc lunnbor of taniilies wonld be cut off, 

 for b(jtli ])ai-ents would not be survivinL', or, if snrviving, beyond thc age limit 

 fixed for parents. Fnrther, we shoidd liavt; been iinablo to interest College 

 studeuts efi'ectively in the matter, as the bulk of tliem fall between 19 and 22. 

 Again, it wonld bave been bcttei' to take a lower niaxinunn age for the parents, 

 but in doing so we should again have greatly limited cur available material. 

 Better organs uiiglit undoubtedly have been selected tban stature, span and 

 forearm, e.g. hcad and finger measurements, bnt in such cases instruments and 

 greater elaboration are needed, and the diflfictdty of obtaining upwards of 1000 

 families, already very great, wonld have been niiich intensitied. We chose organs 

 easily measured with moderate accnracy and asking for the nearest quarter-inch, 

 only tabulated statnre and span to the nearest inch, and forearm to the nearest 

 half-inch. Thns the slight diurnal variations and the errors of measnrenicnt of the 

 eharaeters will not sensibly atfect the constants calculated from our tables. Only 

 a small pereentage were measured in boots ; we could not insist that ladies and 

 gentlemen in middle lifo must remove their boots, or we might have niet with a 

 far larger number of refusals to be measured. Still the bulk of the measured 

 did remove boots. After some experimenting on the effect of heeks on apparent 

 stature it was found that the subtraction of an inch from the recorded stature 

 fairly represented the avcrage increment due to boots. Hence the small per- 

 eentage of boot entries was reduced before tabling by one inch. 



Of course each family canl diil not provide us with four children, our maximum 

 number allowed. Thus the nnmber of onr parental pairs lies for the difterent 

 tables between 1000 and 1400, wiiile for the fraternal correlations we have results 

 based on oöO to 1400 pairs, according to the nature of the table. Tliis is due to 

 the fact that it was found far more ditficult to get the measurements on iwo 

 adidt brothers, than on two sisters*. It was partly this defect in the number 

 of pairs of brothers which led to the wider system of school measurements on 

 brothers. The latter, however, do not modity but only contirm the results 

 obtained from the smaller series in the Family Records. 



I now [iropose to deal at length with the results obtained from our material. 



(iii) Tlieorij applied. 



The regression in all cases is essentially linear, i.e. very closely linear within the 

 limits of random sampling. It is impossible to give diagrams of all the 2 x 78 



* Probably two adult brothers were far more rarely found botli living at liome, or if at Iiorae declined 

 to be submitted to a measurement, which offered uo immediate advantage to themselves. 



