Abstract
To investingate the deviation of Omega observasions, the following six populations were selected and the study was carried out through the Omega phase difference observations at the Faculty of Fisheries of Hokkaido University. A-D pair from February 1, 1974 till January 31, 1975 A-D pair from February 1, 1975 till January 31, 1976 A-C pair from January 1, 1975 till December 31, 1975 C-D pair from January 1, 1975 till December 31, 1975 A-H pair from January 1, 1976 till December 31, 1976 D-H pair from January 1, 1976 till December 31, 1976 Each day was divideb into 48 parts of 30 minutes and each month into two parts, accordingly a year was divided into 1152 parts which were considered to be the same as to conditions. The differences between mean values and measured values at each part show the observation errors, and the distribution around each mean value in the annual population is gotten by accumlating the errors of the 1152 groups. And the results are shown in Fing. 1〜Fing. 6. The solid curves in those figures show the normal curves, but they deviate from the real situations and the reliability of fitness by the X^2-test less than 1%. Now imagine the following experimental equation p(x)=-k/(√(2π)σ)e-k^2(xi-x^^-)^2/2σ^2 then the frequency curves are indicated in equations (1)〜(6). To replace these equations by a normal curve model, we need to use σ' which satisfies the next equation instead of σ σ'=k^<-1>σ and by substituting the relation in equations (1)〜(6), they are transformed into the normal curve equations, that is p(x)=1/(√(2π)σ)e--(xi-x^^-)^2/2σ'
