and are appropriate for testing the null hypothesis: ,8i = 8j (Bolch and Huang, 1974, p. 128). We have carried out this calculation for only the 90-day rates in order to save space. Tables 1 and 2 taken together make a strong case for seasonality. Not only are many coefficients in table 1 significantly different from the intercept for the 90-day rate, but also table 2 shows that they are significantly different from each other at the 0.05 level. A glance at the R2 coefficients in table 1, along with the other material presented, clearly indicates that seasonality is present, but that it is not powerful enough to be of much use in prediction.