IF the horizontal layers of air were isothermal (the upper layers having the lower temperature), then the gradient winds at different heights would be proportional to the temperatures (absolute) at those heights. Consequently the wind would decrease with height, and although a higher temperature at a given altitude over the higher pressure is a necessary corollary of an increase of wind with height, the converse is not necessarily true, it is clear on reflection that with such a temperature distribution as that described above, the temperature at any point in BD of Mr. Dines's diagram in NATURE of March 8 (p. 24) would be below the temperature at the corresponding point in AC, so that BD would be less than AC, and consequently v less than V; but the actual relation runs some risk of being obscured by Mr. Dines's use of isobaric surfaces which in other respects gives an admirably simple exposition of a theorem in atmospheric dynamics, and shows also that if the isobaric and isothermal surfaces coincide there is no variation of wind with height. Incidentally, during the past winter months the mean isotherms have run from N.W. to S.E., and have given at 6000 ft. a N.W. thermal wind of about 1½ metres per second superposed on the wind between 1000 and 1500 ft.