C. S. Durst has tried to discover the frequency distribution of aircraft navigational errors from D. C. Willis's data. But how reliable are Willis's data for this purpose?

D.R. errors and track-keeping errors are differences between assumed and true positions. Evidence of assumed position is ayailable, but how can true position over the Atlantic ever be discovered accurately? Willis estimated true position by post flight analysis of navigational and meteorological data, but any such analysis must be based on some doubtful assumptions, for instance that compass error or wind was constant between fixes or that the true position was at the ‘middle’ of a ‘reliable’ fix. The analysis must also comprise many delicate judgments, for instance that the true position was at A, not B, because if it were at B an abnormal wind change must have occurred for which there is only slight evidence.