Basically, the “ideal” sample size depends on the desired biological significant effect size, the variance of your experiment and the desired power.

Some authors indicate that 10 individuals per study variable should be taken into account for factor analyses.

In R, there is the 'power.anova.test'. See the R script in attachment for some examples.

You may check discovering statistics using SPSS by Andy Field. He has some cool stuff in his book.

Thanks Michiel De Ridder .. good question .. worth pursuing .