.") 1 2 Miscellanea 



iinderstiind In- "type" aml by "race," an«! tlicii wo .sli.ill lio jvble to see wliether a quantitative 

 (iefinitioii c<in lie fouiid for tho ideas wliich ai)i)p,al to tli(Mn. To tlie hioiiictrioiaii tlieir ide;i.s 

 appear to l« confusod imtioiis, for tlicv turn citlier on "api)rcciatioii," wliicli is out of liis plane 

 altogether, or on Statistical niethods wliirh in nine cjises out of ton are unsound. 



In conclusion I wonld only reply to Professor von Török's requcst tliat I will subinit to einer 

 strengen (luxfii lirlichen Kritik- liis iiiotliod of doiiling with correlation, that I liave alrcady statcd 

 my opiniou of his niethod*. It niay be suninied up as follows : In the present state of 

 stjitistical scicnce it is an extreniely cumbersomo niethod of dealing with multiple correlation. 

 Four correlation coefficients and a multiple i-egression equation (such !is those given in Biometrita, 

 Vol. II., p. 349) should replaco soine 70 pages of l'rofc.s.sor von Török's nieiuoir. Tliis replace- 

 ment would not only give increiised brevity, but increa.sed accuraoy, as all such cranial data 

 deiuand graduatioii. Further in the present state of our craniological knowledge, it is fallacious 

 to cxtend any result of intraracial correlation from one racc to a second +, lience I do not think 

 any conclusion from Hungarian data to the Neanderthal skull or the Anthropopithecus Dubois, 

 can amount to aught but the vagucst Suggestion. 



The great value of Professor von Török's paper lies in the fact that a craniologist von Fadte 

 empha-sises the importance of dealing with correlation, when cranial results are considered. 

 This is an immen.se gain, and will have its influencc on future continental craniological publica- 

 tions. At the same time the iiiathoniatical theory of (Mirrelation is (juite simple, and one can 

 only hope that in future continental writcrs will adopt tlic practice of Publishing coefficients 

 of correlation (.such as have becn used in bioinetrio cranial mcmoirs since 1895) to indicate at a 

 glance the relative degree of aasociation between characters. The "multiple regression equa- 

 tions" and "partial correlation coefficients" then give concisely that relationship between the 

 arrays, which Professor von Török is rightly but most laboriously seeking to e.xprcss. 



K. P. 



* Biometriku, Vol. ii. p. 3.')3, footnote. 

 t lUometrika, Vol. ii. p. 3.5.S, text. 



CAMnniDOE: printkd iiv j. an-d c. f. ci.ay, at the ünivkbsitt pbeb8. 



