520 On the Distribution of the Means of Samples 
TABLE IX. 
Chances of Obtaining Certain Deviations in Index Values. 
Samples of 50 
Opsonic Index in 
terms of the 
Fraction of the Total 
Odds against 
Mean 
Area bounded by the 
the occurrence of 
corresponding ordi- 
such a Deviation 
nate (Total Area = 1) 
or a Greater 
•7 
■0112 
88^3 to 1 
■8 
•0618 
15^2 to 1 
•9 
•2283 
3^4 to 1 
1-1 
•2160 
3^6 to 1 
1-2 
•0770 
12 to 1 
1-3 
•0248 ' 
39^3 to 1 
Beyond the limits I'S — '7 
•0360 
26^8 to 1 
1-2— -8 
•1388 
6^2 to 1 
1-1— -9 
•4443 
1^3 to 1 
TABLE X. 
Chances of Obtaining Ce^-tain Deviations in Index Values. 
Samples 
OF 100 
Opsonic Index in 
terms of the 
Fraction of the Total 
Odds against 
Mean 
Area bounded by the 
the occurrence of 
corresponding ordi- 
such a Deviation 
nate (Total Area = 1) 
or a Greater 
■0006 
1666 to 1 
■8 
•0216 
45^4 to 1 
■9 
■1661 
5^0 to 1 
1-1 
•1618 
5^2 to 1 
1-2 
•0420 
22-8 to 1 
1-3 
•0109 
90-8 to 1 
Beyond the limits IS — '7 
•0115 
86 ■e to 1 
1-2— -8 
•0636 
14 •? to 1 
1-1— -9 
•3279 
2^1 to 1 
If we adopt about a ten to one chance as the limiting value for evidence of 
differentiation, then the limits, for single determinations of the index, are in the 
case of 25's, roughly 1^3— "7 ; for 50's, 1-25— -75 ; for lOO's, 1-2— -85. Everyone 
has his own standard of accuracy or reliability ; all we mean is that unless a given 
