View article

The estimation of the most likely magnitude M of an earthquake related to a fault is often based on the regressional analysis of the data on M and associated length of the fault L. Recently, Wyss has cogently argued for the use of the area of the fault A rather than L. Based on published worldwide high quality data, we have carried out straight-line and quadratic regressions of M on A ignoring as well as considering the probable errors in the data. The quadratic regressional analysis with errors is based on an approximate theory. The estimates of the most likely magnitude obtained from the regressional analyses assuming reasonable errors in data and ignoring those errors are not significantly different.