Published online by Cambridge University Press: 15 May 2009
The meaning of the footnote on p. 162 is somewhat ambiguous and the statement comprised in it is not true for all values of n, the number of differences in N.F.T. of which the significance of the mean is tested.
The statement that, by the use of a certain approximate distribution, significance of an observation is underestimated means that the estimated value of P in the neighbourhood of the significance level is greater than the true value.
1 “On tests of the significance of differences in degrees of pollution by coliform bacteria and on the estimation of such differences,” Vol. 41, pp. 139–68.