8 THE MEASUREMENT OF VARIATION. 



we see in the diagram that the most frequently occur- 

 ring measurement is one of 10.5 inches. Wing lengths 

 smaller or greater than the mean occur less and less fre- 

 quently, in rough proportion to their degree of devia- 

 tion from it, so that finally, beyond the extreme devia- 

 tions of 9.6 and 11.7 inches, no measurements were ob- 

 served at all. 



The number of observations here plotted out is ob- 

 viously much too small to yield at all a regular 

 series, but it is quite sufficient to show that the 



9-5 10-0 10-5 11-0 H-5 



FIG. 2. Wing of Sterna Mrundo. 



measurements are by no means evenly distributed 

 through the whole range of their variation. There is 

 a most conspicuous collection of them, or heaping up, 

 in the region of the mean measurement. Supposing 

 the number of observations were increased, then 

 one would expect as a general rule to get a more and 

 more even series; in fact, to get a fairly accurate idea 

 as to the kind of series obtainable, supposing an in- 

 finite number of observations were made. In Fig. 3 

 is plotted out a curve representing the distribution of 

 1923 measurements made by Warren * on a certain 

 dimension, viz., the carapace breadth of the crab Por- 



*Proc. Roy. Soc., vol. Ix. p. 225. 



