LECTURE II. 21 



globe. Traditions, historical facts, physical peculiarities, point 

 to extensive intermixtures, which either affected the purity of 

 the original stock, or perhaps gave rise to a new race. How 

 are we to get out of this maze ? Can we possibly discover a 

 method which may diminish the sources of error, and lead to 

 more certain results ? 



Natural philosophy, and its allied sciences, have long since 

 solved this problem, and it only remains to apply the same 

 method to our investigations. Where inquiries necessarily in- 

 volve many sources of errors, these can only be reduced to their 

 minimum by frequently repeating the observations and measure- 

 ments, so that we obtain from the mass of experiments an aver- 

 age representing a law. The greater the number of individual 

 facts accumulated, and the more strictly they are defined, by 

 selecting, for instance, cases of the same sex, age, and condi- 

 tion, the more exact will be the results. Let us illustrate this 

 by an example. In countries where conscription is in force, aU 

 males, excepting cripples, are measured during their twenty-first 

 year, and those are excluded who do not possess the prescribed 

 military height. We can thus determine the average height 

 of the males of twenty-one years of age in certain countries. It 

 is clear that great errors would arise if only a hundred recruits 

 were measured ; for these may be, as for example, in France, 

 either from Alsace, Brittany, or Provence, which are inhabited 

 by three different stocks vaiying in stature. But after the mea- 

 surement of a thousand recruits from different regions, the 

 calculation of the mean height will be less liable to error ; and 

 by further measurement of all the conscript^ of a certain 

 year the result will be singularly near the truth. Still, even 

 such a proceeding may prove somewhat fallacious, as a parti- 

 cular year may be distinguished by special peculiarities. Thus, 

 it is a fact that during a famin* fewer children are born, and 

 these are, on the average, weaker and less developed than those 

 born during other periods. But by extending the measure- 

 ments to a number of years, even this source of error will be 

 greatly diminished, and the result very nearly approach the 

 truth. 



I have purposely selected this example to show what striking 



