Key points are not available for this paper at this time.
Abstract A set of observations may be partitioned into two clusters to maximize the ratio of between sum of squares to within sum of squares. This maximum ratio is the likelihood ratio test statistic for the observations to come from two normal populations, with different means but the same variance. A table of percentage points of the statistic is given under the null hypothesis, that the observations are a sample from a single normal population.
Engelman et al. (Mon,) studied this question.