K. Pearson 
281 
we used (i) or (//,) there could not have been any difference in the R. M. s. between 
the method of moments and the method of least squares. There is of course a 
distinction between the two methods which it is important to bear in mind, 
I ' 1 1 1 1 — — 1 1— 1 1 1 
-5 -4 -3 .2 -1 0 +1 ,2 -1-3 ^4 4.5 
0^2 345 6 789 10 
Fig. 1. Fitting of Parabolas to Observations. 
namely: the method of least squares takes a curve which passes with the least 
root mean square deviation from 11 observation points; the method of moments 
Biometrika i 28 
