AbstractWhen the atmospheric turbulent flux of a minor constituent such as CO2 (or of water vapour as a special case) is measured by either the eddy covariance or the mean gradient technique, account may need to be taken of variations of the constituent's density due to the presence of a flux of heat and/or water vapour. In this paper the basic relationships are discussed in the context of vertical transfer in the lower atmosphere, and the required corrections to the measured flux are derived.If the measurement involves sensing of the fluctuations or mean gradient of the constituent'smixing ratiorelative to the dry air component, then no correction is required; while with sensing of the constituent'sspecific mass contentrelative to the total moist air, a correction arising from the water vapour flux only is required. Correspondingly, if in mean gradient measurements the constituent'sdensityis measured in air from different heights which has been pre‐dried and brought to a common temperature, then again no correction is required; while if the original (moist) air itself is brought to a common temperature, then only a correction arising from the water vapour flux is required.If the constituent'sdensityfluctuations or mean gradients are measured directly in the airin situ, then corrections arising from both heat and water vapour fluxes are required.These corrections will often be very important. That due to the heat flux is about five times as great as that due to an equal latent heat (water vapour) flux. In CO2 flux measurements the magnitude of the correction will commonly exceed that of the flux itself. The correction to measurements of water vapour flux will often be only a few per cent but will sometimes exceed 10 per cen