Error Distributions in Navigation

Extract

Within the University of Wales Institute of Science and Technology, a general study of navigation has led to a review of error distributions which has now reached the stage at which comment on a wide front would be particularly welcome. The work has been undertaken by Wing Commander E. W. Anderson who, with the generous permission of Smiths Industries Limited, is writing a thesis for the Department of Maritime Studies, and by D. M. Ellis, Lecturer in Statistics in the Department of Mathematics. In the first half of this paper, Anderson emphasizes the distinction between distributions involving one item or identical items and those involving a number of items which, though nominally the same, are in fact diverse. In the second half of the paper (from Section 8 on), Ellis shows how the apparent variation of results that arise in practice may be unified into a simple inclusive pattern.

The first step was to investigate the distributions that arise in practice. It was decided that the pages of this Journal would make ideal witnesses because they provide a broad sample and are impartial. From these pages every distribution was extracted which involved more than fifty observations and whose pattern was not markedly skew. The distributions were transferred to log-linear graph paper so that the horizontal error scale was linear and the probability of each error was plotted on the vertical logarithmic scale.