Simulate observations like those of Figure 12.1 under the null hypothesis of no treatmen...

Simulate observations like those of Figure 12.1 under the null hypothesis of no treatment effects. That is, simulate seven batches of ten normally distributed random numbers with mean 4 and variance .0037. Make parallel boxplots of these seven batches like those of Figure 12.1. Do this several times. Your figures display the kind of variability that random fluctuations can cause; do you see any pairs of labs that appear quite different in either mean level or dispersion?

Reference