A survey is given of the various methods of determining the slope and intercept of a straight line when both variables are subject to error. The results for each method, after one iteration, are compared with the best available computer method by analysing 100 sets of data. A procedure suitable for use on a calculating machine is recommended in which the gradient differs from the best value by some 3‐4 per cent of its standard error for the 100 sets of data considered.