[thras]

Yes, it is fairly pointless. However, they do mean something exact. You quantify the "within-group" variation not with standard deviations, but by measuring all individuals in a group and finding the endpoints.

So you find the tallest white person in the world and the shortest, and note that they differ in height by a number of feet. Then you measure the difference between the white height average and the Asian height average, and note that the difference is only a few centimeters. Wallah! Race doesn't matter. Most variation is within group.

Like I said, I chuckle a bit.