Comparisons of successive frequency distributions may be made in two distinct ways that are frequently confused. One way is to make comparison against a given set of class intervals covering the relevant range of the data. The result is a fixed frame of reference, a single classification grid. The other way is to compare "corresponding relative parts" of the frequency distributions-the quartiles, deciles, or similar scale orderings. Such scale orderings, in effect, adjust the class intervals for each distribution so that they are keyed to the central tendency of that distribution. This analysis uses data on stature of fathers and their sons to illustrate the confusion that can occur.