Editorial 
281 
■ .j'-'^'Qr^ ' 
Accordingly we see that when samples N are taken from normal material the 
array 7i^, of varying size in these samples will not ditfer very greatly from normality. 
For example if n., — 25 and the sample N be 1000, wc shall have = '024 and 
-S2 = 3'12, showing no great deviation from normality, although more than in 
the case of a sample of constant size. It is probable that the deviation from 
normality will be somewhat greater when the sampled population itself is not 
normal. Still it is important to note that the distribution of cr,,,^ for all cases is 
likely to be far closer to the normal, and the " probable er'ror " of o-„^ therefore more 
intelligible, than is the case with In the same way it is extremely probable 
that the distributions of {np/^s)'' Jind („p/u-4)"' are more nearly normal than those of 
np^^■J and „pf^i- 
(V) Mathematical Contributions to the Theory of Evolution, xix. Second Siqj- 
plement to a Memoir on Shew Variation. Phil. Trans. Series A — Vol. 216, 
pp. 429—457. 
There are one or two corrections to be made in this paper by Pearson : 
(a) p. 439, 1. 18. The printer has drawn the solidus and the 3ySi — 2/3. + (3, 
which followed it. Thus 
4 - Hi = 2 (^, + 3) 
should be read as 
4 - m = 2 + 3)/(3/3, - 2yg, + 6) ; 
(b) p. 441, 1. 18, about the middle of the page the ecpiation 
7 = 12 (sec 0 — oosec 6) 
is given. It should be 
7 = 3 X 12 (sec ^ - cosec ^)-/sec f^, 
but no use has been made of the equation in the paper. 
Biometrika xn 
19 
