458 



MR ROBERT WORTHINGTON. 



and 72 inches would be arranged in twenty-five groups, and in 

 the group labelled 66 inches the actual measurements would all 

 lie between 65f inches and 66^ inches. 



If now such a system of ' grouped measurements ' be plotted 

 on paper, with reference to two axes at right angles, so that the 

 ordinates represent the number of measurements in each group 

 and the abscissa? their typical measurements ; then, on joining 

 the points so obtained, we shall obtain a polygon somewhat of 

 the form shown in fig. 1. 



Fig. 1. 



This is known as a ' frequency polygon,' and defines graphi- 

 cally the distribution of the organ in the sample of the popula- 

 tion measured. 



If the number of organs is large, so that they can be con- 

 veniently divided into a consideraljle number of groups, it will 

 not be difficult to draw a continuous curve through the angles 

 of the polygon. And if wc assume the variation to he continuous, 

 each point on the curve will indicate the frequency in the 

 population of organs of a particular size. Now, it has been 

 found that, in the majority of cases, this curve approximates 

 very closely indeed to what has long been known to mathema- 

 ticians as the curve of ' Probability of Error.' And it is on the 

 assumption of the identity of these curves that exact statistical 

 methods are based. 



