38 
THE VOYAGE OF H.M.S. CHALLENGER. 
( B) If we consider each series of analyses by itself, we have — 
For m — s. 
d — 8. 
I. Series, 
= 00135 
0-0148 
II. Series, 
= 0-0115 
0-0128 
III. Series, 
= 00146 
0-0149 
IV. Series, 
= 00104 
00107 
. t he probable error r in the individual value of 100 (L 0 -^x), as we saw, is =t0’0046, 
whence 2 x r= 0*0092. Our smallest difference is the m — s in series IV., which amounts 
t<> 00104 = 113 x 0'0092 ; and the probability that even this difference owes its existence 
to i' . idental errors is 0'4f), or rather considerably less, because we started with the impro- 
bability that the two r’s are of opposite signs. 
I have to thank Mr. John M'Arthur for the scrupulous care with which he executed 
all the analyses quoted in this appendix. 
