A DEPARTMENTAL paper lately published by the Survey of Egypt contains an interesting investigation of a systematic error which has been found to occur in the levelling carried out in Egypt and in the Sudan. The effect of this error, which has the same sign over all kinds of ground, acts in the direction of making the backstaff reading systematically too small and the forestaff reading too great. Movement of the staves or level and other sources of error having been eliminated, the author draws the conclusion that the effect of refraction is not wholly removed by keeping the distances between the level and the staves equal in the conditions under which the work is done. Precise levelling in Egypt is carried out in the winter months and during about three hours after sunrise and three hours before sunset. Experiments have shown that at sunrise a temperature lapse-rate of the order of 1° C. to 2° C. per metre often exists, the air being colder near the ground than higher up. In two or three hours this lapse-rate has disappeared, and a little later becomes reversed, so that with hotter air near the ground setting up convection currents, unsteadiness of the staff-image sets in, preventing further work. In the afternoon the ground cools very slowly, so that the change in the temperature lapse-rate, and consequently in the refraction, is then very gradual.