80 HEREDITY AND VARIATION 



relations between them in an equally compendious manner. They 

 commence by marshalling the values in order of magnitude from 

 the smallest up to the largest, thereby converting a mob into an 

 orderly array, which, like a regiment, thenceforth becomes a 

 tactical unit. Conceive each value to be represented by an ex- 

 tremely slender rod of proportionate length, and the rods to be 

 erected side by side, touching one another, upon a horizontal base. 

 The array of closely-packed rods will then form a plane area, 

 bounded by straight lines at its sides and along its base, but by 

 a flowing curve above, which takes note of every one of the values 

 on which it is founded, however immense their multitude may be. 

 The shape of the curve is characteristic of the particular group of 

 values to which it refers, but all arrays have a family resemblance 

 due to similarity of origin ; they all drop steeply at one end, rise 

 steeply at the other, and have a sloping back. An array that has 

 been drilled into some such formation as this, is the tactical unit 

 of the new statistics " (Biometrika, vol. i., 1901, p. 7). 



Theory of Evolution by Selection of Fluctuating Varia- 

 tions. It is certain that most offspring differ from their parents 

 in many quantitative details. It is certain that when measure- 

 ments are taken of a large number of individuals of the same 

 species in reference to a particular character, the results, when 

 plotted out, conform approximately to the normal curve of 

 frequency. If measurements be taken in a subsequent genera- 

 tion there is a similar result, but the curve need not be precisely 

 the same. The mode of the curve i.e. the most frequently 

 occurring dimension of the measured character, may change 

 from one generation to another. It is usually believed that one 

 of the ways in which this change can be effected is by natural 

 selection. But to think of new species arising by slow changes 

 of this sort is in many ways difficult, apart altogether from the 

 fact that definite demonstration of the operation of selection 

 has been rarely attempted. 



(i) Such a character as a Roman nose is certainly heritable, 

 though it is not always inherited. But we cannot speak so 



