between these “variability characteristi¢s" (computed for two test Maite Series as well as for different geographical areas) are subjected to . a standard statistical test. Table 6.1 shows the results of that test in terms of probability. The probabilities shown reflect the likelihood that the "variability characteristics" of gummed groups were computed from a single homogeneous population, i.e., a small | probability infers that the groups were NOT draw from a homogeneous . population and are, therefore, different, while a probability in the vicinity of 0.5 infers that tne apparent differences could have occurred by chance. TABLE 6.1 ‘Probability that "Variability Characteristic" for Compared Groups Was Computed from a Single Homogeneous Population Duplicates Compared , IVY - thited States vs. Probability’ : ‘remainder of world | UPSIDT-KIOTHOLE - East ee 0.30 a ‘of Mississippi River vs. 0.55. west. of Mississippi River . IVY + United States vs. UPSH)T-KNOTHULE, United States IVY - all date vs. 0.10 . | UPSHOT-KNOTDLE, United States a 0.01 .and Canada Table 6.1 indicates that the tendency of adjacent. gummed papers to be different from one another depends upon the test ‘series rather than the area in which they were exposed. This con- clusion is interesting because it suggests possible causes. For example, the relative. levels or radioactivity in the area comparisons of IVY data was quite different from the relative levels of activity involved in the UFSHDT-KICTHOLE data yet the "variability characteristic" remained unchanged from area to area. This suggests - that the level of activity has a minor effect on this statistic. On the other hand, one obvious difference between collections of “Probabilities are resultof "t-test" for significance of difference in means (5, Table 12, p. 138). - bh - . a . . . Ce