It is the present usage by the National Physical Laboratory and the Trade to give, on its Certificate, the ‘constant’ (graduation and centring) errors of a sextant only at stated points on the scale. If these errors are of a magnitude which is worth while applying as a correction, this usage is not quite what the user requires. The N.P.L. Class A Test, issued for instruments reading directly to 10″ (or o′·2 in the case of decimal subdivision) allows combined graduation and centring error up to 40″ or 4 scale divisions of the micrometer or vernier; the Class B Test for instruments graduated to 20″ permits such error up to 2′ or 6 scale divisions. Thus, even with instruments that pass these tests, it may be worth the marine navigator's while to apply the error as a correction. What he wants to know, however, are the limits of such errors as are worth his while to apply. In other words, he requires a Critical Table of Corrections, of the type used in the Air Almanac and in certain mathematical tables by L. J. Comrie, by whom, it is understood, the method was introduced.