734 EIGHTH PACIFIC SCIENCE CONGRESS 



The dashed curve extremely rises at Osaka by the local land sub- 

 sidence in the western part of the City. The total amount of variation 

 in 19 years (1926-44) is nearly 110 cm., or 6 cm. per year. At other 

 stations, the rates at which the sea-level changes are within 1 cm/year 

 or so, except for the sudden or rapid changes accompanied with the 

 great earthquakes (Aburatsubo, Shimizu, Shimotsu). 



3. Periodic Variations 



At first, we eliminate the progressive variations from the observed 

 series of the mean sea-level. These residuals really make the signifi- 

 cant variations or not. The check is easily made by plotting the cor- 

 relograms (Fig. 3), and we find that the variations are existent at the 

 significance level of 5%. Figure 3 also suggests the existence of the 

 periods 11-12 year and 18-19 year. 



The exact periods will be determined by the usual method of 

 periodgram analysis. According to the computed results, the most 

 probable periods are 



11.6 and 18.6 years for Kushimoto, and 

 11.4 and 18.6 years for Tonoura. 

 Amplitudes of these variations are significant (significance level 5%) 

 at these stations. Besides, the amplitude of 9.3 year variation is larger 

 than that of 18.6 year variation at Tonoura, which means the existence 

 of bitide for the 18.6 year tide. 



The amplitudes and phases (T = at 1940) of these periodic va- 

 riations are computed for several stations, and shown in Table I. These 

 coefficients are coiTected and the effects of other periods are eliminated. 

 The statistical tests are applied according to Ogawara's method (5), and 

 insignificant variations are omitted there. 



These results are not in good agieement -with those computed by 

 Kawakami (6) chiefly because of his neglect of 11,5 year variation. 



4, Effects of the Land Deformation 

 The land deformation is considered to be the effective origin of 

 the sea-level variations. The results of the precise levelling carried out 

 by the Geographical Survey Institute will be available to ascertain 

 the relation between them. 



(i) Aburatsubo— The precise levelling is frequently carried out 

 along the route Tokyo-Aburatsubo [cf. Yamaguchi (7)]. These results 

 fairly resemble the secular trends shown in Figure 2. However, the 

 correlation coefficient between the leveled height of the datum line 

 and the corresponding annual mean sea-level is 

 r = -0.59 (n = 6) 



