In the western part of Japan, 1st order triangulations were carried out three different times by GSI. The horizontal displacement in the western part of Japan during the period from 1945-1955 to 1968-1972 is discussed by using the second and third set of geodetic data. In the horizontal displacement obtained under the condition of Συ=0 (Fig. 1), rotational movement is found in the Kui and Shikoku districts which is unusual. First, the accuracy of observation is examined. The distribution of observation error is shown in Fig. 3. The standard deviation of an angle with unit weight is shown in Table I. The accuracy of observation in the third geodetic survey is lower than those of the first and second ones. Second, the Monte Carlo Method is applied to the occurrence of observation error obeying to the Gaussian distribution. The systematic tendency is found here and there in the cumulative error (Fig. 4). It seems likely that the rotational movement appearing in Fig. 1 may be related to the apparent cumulative error. The latest horizontal displacement in the western part of Japan is shown in Fig. 6, assuming 13 fixed points. This assumption is the same as that of Fig. 5, which is con sidered as the co-seismic horizontal distribution of the Nankai Earthquake with M 8.2 of 1946. Fig. 6 clarifies as follows. 1) In the Kui Peninsula the horizontal displacements almost disappear. 2) In the Shikoku district the horizontal movement is about 1 m maximum in the same direction of Plate movement. It is noted that the horizontal displacement of No. 264 at Ashizuri Pt. is in almost the same direction during and after the Earthquake.