EVOLUTION 



385 



of I and 3 capsules respectively. A good measure of varia- 

 tion must be sensible to changes in the frequency, especially 

 of extreme variations, even if the range remain constant. 

 It is just these extreme variations, and their relative 

 numbers, which are all-important for the problem of evolu- 

 tion. Now we shall define a deviation to be the amount 

 by which any individual differs in a given character from 

 the type. Here the type may be measured by the mode 

 or the mean, usually we take deviations from the 

 mean. Thus in the poppy capsules individuals with 

 a deviation of -H 3 from the mode {i.e. with thirteen 

 stigmatic bands) occur in 54 out of 2268 cases ; i.e. about 

 2.4 per cent of cases, or again, individuals with a deviation 

 of — 1.84 from the mean (9.84) occur in 295 out of 

 2268, or in about 13 per cent of cases. Deviations must 

 thus be looked upon as negative and positive, according 

 as the individual has less or more than the type value of 

 the character. We now require some method of appreciat- 

 ing how deviations are distributed along the range. 

 Graphically we can represent the distribution by a polygon 

 of frequency, obtained by scaling uniform lengths along 

 the horizontal, 5, 6, 7, 8 . . . 14, 15, 16 to represent the 

 number of stigmatic bands, and " plotting up " to these 

 the vertical lengths which represent on some scale the 



Diagram of Percentage Frequencies of a Character in two 

 Individuals and in a Local Race. 



(/) = First Poppy. 

 (ii) = Second Poppy. 

 : Local Race. 



7 r, e 7 s 9 10 II IS 13 II- 15 16 17 IS 



Number of Stigmatic Bands on Seed Capsules of wild Poppy (Paj>avei- R/uras). 



Fig. 26. 



frequencies of these individual types, conveniently reduced 

 to percentages. In the accompanying diagram, we see such 



25 



