Consider 3 independent normal population with a commonvariance σ2 = 100.

Suppose that we would like to detect a difference in the means with a prob-ability of 90%, if the range in the means is ∆ = 20. Assume that σ2 = 100.Determine the appropriate sample size of each treatment group.

Im not sure I know the formula of where to start? Im assuming we have to use the F distribution with non-central parameters?

Any help is appreciated