Abstract
Damage statistics for the 1923 Kanto earthquake frequently show some inconsistency in amount of damages. Some datasets provide different numbers of damages from another for the same municipality. It has turned out that these differences are caused by; 1) the difference in measure of damages using number of buildings or number of households, 2) the number of non-residential buildings that were occasionally summed up in datasets, and 3) whether buildings burnt down after collapse by shaking were taken into account or not. A newly developed dataset is presented in this paper for specifying uniform numbers of damaged houses due to the Kanto earthquake, in which the damage data have been re-evaluated from the report by Matsuzawa (1925) and Ministry of Home Affairs (1926). A seismic intensity distribution estimated from the collapse rate of houses is also shown covering the entire region of the affected area. It is confirmed that the relationship between the seismic intensity and the ground condition can be clarified precisely from the new dataset.
