The heights deduced by means of multiplexes were examined in actual field by leveles and the deviations in heights were worked out. The data of deviations in heights thus worked out number 576, and they are subjected to statistical study. The results obtained by the statistical study are: (i). The frequency distribution of the deviations in heights is of Gaussian type, and the standard deviation δis±5, 6 m when the whole data are taken into consideration. If, however, the deviations larger than 10 m, which might have resulted from some mistake or from local distorsion in the photographic films, were excluded, δis±3.9 rn., the data taken in this case being 540 in numberr i. e., 94% of whole data. This lead to the conclusion that the deviation in heighh will be less than 4 m at 5% level of significance, that is, about 30% of the height deduced by means of multiplexes shall be revised by the field examination survey. (ii). The median value of the deviations in heights shifts by about 2 m. in positive side. If this shift of median value is really existing, but not attributable to the personal property of operators, it may caused by the errors in the lens systems of optial appara tuses through which the topographical features are projected on the table of multiplexes, and/or by the asymmetry in the intensities of lights projected through red and blue screens, and soon. This latter fact may deserve further development of statistical and experimental study.
View full abstract