An investigation into the errors in measurements by the Tardy and Sénarmont methods shows: (i) that the errors due to inaccurate quarter-wave plates are less in the Sénarmont method than in the Tardy, (ii) that errors due to inaccurate setting of the polaroids are of the same order in both methods, and (iii) that in general the error in the final setting of the analyser is the only one which need be considered. A full account is given of the less well-known Sénarmont method.