Published online by Cambridge University Press: 18 January 2010
There are two approaches to the study of errors in astronomical observations at sea. One is to collect as much information as one can from practising navigators; the other, described here, is to carry out controlled experiments with the object of breaking down the various causes of error, a virtually impossible task if the first approach is adopted. In both cases a large mass of data must be carefully studied and analysed, and this is particularly important where one aims to resolve the errors into separate components (those due to dip uncertainties, horizon, body, degree of twilight, and sea state).