Statistical InvextiyatiiniH 335 



Gallon contributed the hundredth section entitletl "StatisticB." He opeos 

 with the characteristic sentence: 



"Tho topics Huitablo to HtAtiHticH aro too numerous to Kpocify; thoy inc! vthinR to 



whicli such phrases lis 'udually,' 'wliloin,' ' very often ' «nfl th« likv aru n|>i>i .liich vnx 



tho int(>llip-iit ri'ador hy their vague-nogs and make liini impatient at the absence of more 

 prcoiHo data." (p. 143.) 



He then refers to the necessity of homogeneity, the breaking up even of 

 homogeneous groups when there is a variation largely governed by a dominant 

 influence, e.g. age, and the need for a truly riiiidoiu selection. He says that 

 precision varies as the s(juuro root of the numljer of observations, but that 

 number nuist not be reached at the expense of accurate reporting. He then 

 turns to the "law of deviations" and suggests the "ranking" of characters in 

 individuals, and the mejusurement of the mid (500th), the •25Uth and the 

 750th individuals in ranks of a thousand, or what we now term the median 

 and (piartiles'. The ranking gives him his so-calletl ogive curve, and his 

 whole appeal to theory consists in the statement that when individual 

 dirterences in a homogeneous population are due to manv small and inde- 

 pendent variable influences then the excess of the {m-\-t)t\\ individual if m 

 be the mid number will equal the defect of the (»i — <)th individual from the 

 mill individual. Galton does not enter into the mathematics of the matter. 

 He says this 



"law of devirttioiis liolds for the mUiIiiio of men iind animals, and apparently in a useful 

 degree for every homogeneous group of qualities or comijouiid qualities, mental or Iwdily, that 

 can be named." 



Galton gives no proof of the "normal curve of deviations," but suggests that 

 it is mathematically deducible on making certain rather forced suppositions 

 to render calculation feasible. Comparing fact, however, with theory 



" wherever comparison is possible, it is fuund that they agree very fairly and in many cases 

 surprisingly well." (p. 14 t.) 



He concludes with the statement that a good book on these matters has yet 

 to be written. 



"Quetelet's Letters on the Theory of Proliabilities is perhaps the most suitable to the non- 

 mathematical reader." (p. 14(5.) 



It will be clear that Galton was proceeding gradually, and the dose was a 

 very small and simple one'. 



In the second edition we find Galton contributing some further sections. 



' Galton then termed the 500th individual in a thousand the "average." Tlie middle man 

 is practically the 500th, but not so theoretical ly. The diagrnm in later editions disappeared. 



" Other contributions l)y (!alton to the first edition were No. xciv on "Population," which 

 begins characteristically with "Count wherever you can," No. LVin on "Communications," remi- 

 niscent of the Art of Travel, No. Liv "Causes that limit Population," No. x.\i "Astronomy," 

 with special reference to the seasons, and to steering by sun and stars. There is also (pp. 21-2) 

 a note on heredity, giving a list of her«>ditary characters which admit of preci.se testing; 

 "those who confuse the eflects of nature and nurtun? give information that is of very little 

 use." The first e<lition also contains a section (No. i.\) on "Physiognomy," by Charles Darwin, 

 who was collecting material for his work on Kxpres.sion of the Emotions, a section which 

 Dr Garson had the temerity to revise in later editions. 



