By dividing deviations between observed travel times and those of the Jeffreys-Bullen travel time tables into a station and an epicenter correction, it is possible to calculate corrections to these tables. A significant dependence of station and epicenter corrections upon locality is demonstrated. For the epicenter correction this is especially pronounced in the region of Japan. The correlation between calculated corrections and structures is discussed and the conclusion is that the corrections can be explained by velocity anomalies in the crust and upper part of the mantle. However, systematic errors in the calculated values are possible, especially for the epicenter corrections. No systematic correlation between the observed deviations of P and S waves can be detected, although such a relation should be expected from their respective velocity distributions. An appropriate way to improve the epicenter determinations is proposed, by incorporating the station and epicenter corrections in the solution procedure.