II. Variation avd CorrcJatioii in Lesscr Celandine 



140 



These rosiilts f'onii ;i fnrther proof, wevo aiiy iicedf'ul, thal mcu caiinot bo 

 defined by correlatioii*. Plauts aro, of coiirse, niore subject to cnvironmont than 

 aiiiiiial lit'r, aiid IIk- incans and variabilities change with changc of" ciiviromnciit 

 and üf seasDii vcry ri'adily. Man, jjorhaps, of all aninial lifo seeniw icast affcuted 

 by cnvinmineiit, but u table like thu abovc showing how different the moan, varia- 

 bility and correlation ai-e when there is slight change in locality or seasoii, ought 

 to render us cautious of cataloguing cven local ractjs of man by the nieans alone 

 of sniall craniologieal series. Niirtare and environment during growth may 

 sensibly influence even liunian niean characters. The biometrician, far from being 

 discouraged by the fact that the simpler biometric qnantities — means, variabilities, 

 correlations — are not constant for local races, ought rather to recognise that if 

 natural selection be a reality these aro the very qnantities which he would expect 

 to change from one local race to another, and from one environment to a second. 

 Indeed it is exactly thcsc difFerencos which form the foundation upon vvliich \ve 

 hope to build up in the future the evolutionary history of local races; they are 

 the mateiial from which wo nuist determine not only the characters which in 

 each case have been selected, but also the absolute constants which define the 

 species itsolf+. 



(4) Comparing the first four columns — the new data — with tho last foiir 

 colunins — the old data for central Europe — one is at once Struck with tho high 

 degree of variability in sepals and petals of tho latter as compared with the 

 former. The sepals are more than four times, the petals are nearly twice, as 

 variable in tho seccmil group, i.o. the mean values are -ß.544 and r0946 as agaiust 

 •1.598 and ■ö!)44 rospectively. Now -we see in the case of the Dorset group that 

 the petal variability can rise to double the value it hail in Bordighera, Guernsey 

 or Surrey. But an examination of the tables below shows that much of the 

 difference arises from the absence of individuals with less than 3 sepals or 7 petals 

 in the first group, and as our data are drawn from very diverse districts we must 

 await with interest the result of bud countings from Germany. 



TABLE II. 



* Phih Trans. Vol. 187, A, pp. 266, 280. 



t " Mathematical Contiilmtions to tlie Tlifiny of Kvolutioii. XI. On the IiiI1iii'th-c.' of Niitiiral 

 Selection ou Variation and Correlation." Phil. Trans. Vol. 200, A, pp. 18—21. 



