Abstract
The Dunnett procedure for comparing several treatment means with a control is applied to the problem of comparing several treatment variances with the variance of a control. Appropriate critical values are specified and an example is provided.