A STATISTICAL TEST OF CORRELATIONS AND PERIODICITIES IN THE GEOLOGICAL RECORDS

Matsumoto & Kubotani argued that there is a positive and statistically significant correlation between cratering and mass extinction. This a rgument is critically examined by adopting a method of Ertel used by M atsumoto & Kubotani but by applying it more directly to the extinction and cratering records. It is shown that on the null-hypothesis of ran dom distribution of elater ages, the observed correlation has a probab ility of occurrence of 13%. However, when large craters are excluded w hose ages agree with the times of peaks of extinction rate of marine f auna, one obtains a negative correlation. This result strongly indicat es that mass extinction are not due to accumulation of impacts but due to isolated gigantic impacts. Further, correlations between cratering and flood basalts and between mass extinction and flood basalts origi nally recognized by Stothers are further investigated. It is shown tha t for the former, large craters corresponds to 7 of 11 major flood bas alts. Periodicity analyses are carried out for the geomagnetic reversa l events and extinction rate peaks recently given by Sepkoski for the interval up to 540M yr BP. The former exhibits a significant periodici ty at P = 25M yr, but the latter shows no significant period. It seems premature to regard the periodicity in geological records established .