K. Pkak.son AM) A. Lkk ."571 



Now tliis 'L'ahlo ciiiituiiis a iimubLT i>t' niost iufcorestiiig poiiit.s. 



In llu' lirst place tlie probahle cri'ors show us that tni' all three characters in 

 biitli stwis die youiiger ovnoratimi is disliiictly largcr llian ihe older «^'C'iieratioii, 

 soll thaii tatliiT, ilaiinlilcr iJian iiiDtrhi.T. Is Ulis a i'i'al pro^Tcss in type:' Taking 

 Mr Powys' iliai^raiii f'nr sliriiikaj;i' in slaturc*, \vc sh()\il(| cxpect oiir mcii to reach 

 a inaxiiinuii at ahout 2S and onr wiinicii at 2'>. llciici', sinoe tlic averacfc age of 

 our youniicr i^cncraiiim is not iiioi\' llian 22 yoars, thc yoiingci' s^cncration cannotr 

 have reached its inaxiiniim. <)n t.lio other band, oui' avurage agc of pareiits iiiust 

 be aboiit. ")0. Lct ns supposc tlioin to bo 55 cvon. Tlio diffoicnce in age of 

 parents and oll's|ii-ing wonld iJius ni.uk a slirinkagc of abont '5" at a inaxinmin. 

 lUit tilie ditfercucu bi'tworn lalluTs and s(]iis is abunt an incli for stature and 

 Kjiaii, and for iiiothers and danglitcis abont: an indi and a. half. It seeins inipos- 

 sible thcrefore to attribntc tlic wliolo {-liangc bctwocii llic two generations to old- 

 age shriiikage. In tho ncxt ])lac(', eaii it lic dtu' to periodic selection, i.e. only 

 a portion of tbc yonnger goiioration becoinu fatlicr.s and mothers? If so, \ve 

 slionld expect not only a cliange in type, bat a chaiigu in variability between 

 tbe two generations. Coniparing tlie standai<l doviatioiis of fatluTs and soiis, \ve 

 See tbat fatheis and sons aio witliin tlie liniits of landoin sanipling equalh' 

 variable. On the other band danghters' Standard deviations arc in every caso 

 sensibly larger tlian tliose of their mothers. It «cmld thns seeni iiighly probable 

 that the causes at work in tbe cases of the two sexes are not entirely tlie sanie. 

 Mothers of adult children are a more stringently selected portion of the population 

 than fathers appear to be. Of conrse some change in type between mothers and 

 adult danghters is iindoubtedly due to the f;\ct of child-bearing, independent of 

 any selection in childbed. Bnt it is difificult to see how a physiological eti'ect of 

 this kind could change variability as well as type. I have sbown that there is 

 a slight correlatioii between size and fertility in woinenf, and this may be partially 

 the source of the observed effect. Whether, however, the result be due to natural 

 or reproductive selection, the change in the variability of the two generations 

 of women seenis to nie to indieate that there is a selective change going on in the 

 women of the niiddle classes in this country. The differeuce in type between 

 fathers and sons, — since there is no change of variability, — niight be niore likely 

 to be due to iinproved 2ih\sical exercise. Of course a jiortion of the change in 

 the woinen must also be attributed to this, but the change in variability forbids, 

 I tbink, its being entirely attributed to this source. 



However we judge the matter, whether we consider it due to selection, or to 

 better environment, nourishment, or exercise, there seems no reason to supi>ose 

 that the population, as far as the middle classes are concerned, is degeuerating. 

 In span, stature and forearni the yonnger generation is sensibly better thau its 

 parents. 



* lliometrika., Vol. i. p. 47. 



t li. S. l'roc. Vol. 5'J, p. 303. See a.lso Vol. Ijti, p. 28 et seq. 



