In comparing two averages of ten observations each, the difference should not exceed the following
critical differences in 95 out of 100 cases when both sets of observations are taken by the same well-trained operator using the same piece of test equipment and specimens randomly drawn from the same sample of material.