A SEMI-ANNUAL variation in upper-atmosphere density, with maxima in April and October, and minima in January and July, was discovered in 1960 and has been generally accepted as a feature of the upper atmosphere at heights of 250–600 km. In a recent article, Anderson1 suggested that the variation is a disguised effect of latitude. Anderson's hypothesis can be tested by analysing the orbit of a satellite of a near-circular polar orbit which samples all latitudes in a nearly equal degree. If the semi-annual variation is recognizable in the orbit of such a satellite, it cannot be due to changes in latitude.