189 5.J Remarks on Variation in Animals and Plants. 381 



2. The questions raised by the Darwinian hypothesis are purely 

 statistical, and the statistical method is the only one at present 

 obvious by which that hypothesis can be experimentally checked. 



In order to estimate the effect of small variations upon the chance 

 of survival, in a given species, it is necessary to measure first, the 

 percentage of young animals exhibiting this variation ; secondly, the 

 percentage of adults in which it is present. If the percentage of 

 adults exhibiting the variation is less than the percentage of young, 

 then a certain percentage of young animals has either lost the 

 character during growth or has been destroyed. The law of growth 

 having been ascertained, the rate of destruction may be measured ; 

 and in this way an estimate of the advantage or disadvantage of a 

 variation may be obtained. In order to estimate the effect of devia- 

 tions of one organ upon the rest of the body, it is necessary to 

 measure the average character of the rest of the body in individuals 

 with varying magnitude of the given organ ; and by the application 

 of Mr. Galton's method of measuring correlation, a simple estimate 

 of this effect maty be obtained. In the same way a numerical measure! 

 of the effect df parental abnormality upon abnormality of offspring 

 may be obtained by the use of Galton's correlation function, and such 

 measurements have been made, in the case of human stature, by 

 Mr. Galton himself; 



It is to be observed that numerical data, of the kind here indicated, 

 contain all the information necessary for a knowledge of the direction 

 and rate of evolution. Knowing that a given deviation from the 

 mean character is associated with a greater or less percentage death- 

 rate in the animals possessing it, the importance of such a deviation 

 can be estimated without the necessity of inquiring how that increase 

 or decrease in the death-rate is brought about, so that all ideas of 

 "functional adaptation" become unnecessary. In the same way, a 

 theory of the mechanism of heredity is not necessary in order to 

 measure the abnormality of offspring associated with a given parental 

 abnormality. The importance of such numerical statements, by 

 which the current theories of adaptation, &c., may be tested, is 

 strongly urged. 



3. The report itself describes an attempt to furnish some of the 

 numerical data referred to for two dimensions of the shore crab. The 

 data collected give an approximation to the law of frequency with 

 which deviations from the average character occur at various ages. 

 The conclusions drawn are (a) that there is a period of growth 

 during which the frequency of deviations increases, illustrating 

 Darwin's statement that variations frequently appear late in life ; 

 (6) that in one case the preliminary increase is followed by a decrease 

 in the frequency of deviations of given magnitude, in the other case 

 it is not; and that (c), assuming a particular law of growth (which 



