IMPROVEMENT OF THE HUMAN BREED. 



525 



in using this law in the way I am about to. 1 say that //"certain quali- 

 ties vary normally, such and such will be the results; that these qualities 

 are of a class that are found, whenever they have been tested, to vary 

 normally to a fair degree of approximation, and consequently we may 

 infer that our results are trustworthy indications of real facts. 



A talent is a sum wdiose exact value few of us care to know, although 

 we all appreciate the inner senseof the beautiful parable. 1 will, there- 

 fore, venture to adapt the phraseology of the allegory to my present 

 pur})ose by substituting for "•talent" the words "normal talent." The 

 value of this normal talent in respect to each and any specified quality 

 or faculty is such that one-quarter of the people receive for their 

 respective shares more than one normal talent over and ahove the aver- 

 age of all the shares. Our normal talent is therefore identical with 

 what is technically known as the "probable error," Therefrom the 

 whole of the following table starts into life, evolved from that of the 

 ' "jrrohahHify in tegral. " 



Table I. — XoniKil (!lstrihiill(»i (to the nearest per ten thotisaiid and to the nearest }>er 



hundred). 



It expresses the distribution of any normal quality, or any group 

 of normal qualities, among 10,000 persons in terms of the normal tal- 

 ent. The M in the upper line occupies the position of mediocrity, or 

 that of the average of what all have received; the +1'', -H2^, etc., and 

 the —1°, —2°, etc., refer to normal talents. These numerals stand as 

 graduations at the heads of the vertical lines by which the table is 

 divided. The entries })etween the divisions are the numbers per 

 10,000 of those who receive sums between the amounts specified by 

 those divisions. Thus, by the hypothesis, 2,500 receive more than 

 M but less than M + 1 ., 1,H13 receive more than M+1^ but less than 

 M+2 , and so on. The terminals have only an inner limit; thus, 35 

 receive more than -t, some to perhaps a very large but indefinite 

 amount. The divisions might have been carried much further, but the 

 numbers in the classes between them would become less and less trust- 

 worthy. The left half of the s(>ries exactly reflects the right half. 

 As it will be useful henceforth to distinguish these classes, I have used 

 the capital or large letters, R, S, T, U, V, for those above mediocrity 

 and corresponding italic or small letters, r, 6f. t, v, t\ for those below 

 mediocrity, r being the counterpart of R, s of S, and so on. 



