AbstractWinds and temperatures in the boundary layer measured during two Australian expeditions are analysed according to the similarity scheme, with the use of four stability classifications.Under conditions of deep convection there is a minimum potential temperature and a maximum velocity component in the direction of the surface wind, at a height of about 0.12 u*/f. In the very stable cases, the temperature gradient follows rather closely a z−2law for a considerable height range from 0.08 u*/f upwards. An expression closely fitting the mean data in stable conditions is suggested for the vertical temperature structure at all levels in the boundary layer.Wind data processed in this way show, for all four stability classes, a rudimentary Ekman spiral. With deep convection the spiral is found to be reversed in sense, but the flow in the convecting layer is down the gradient of pressure. If the convective limit lies broadly within the Ekman layer, a spiral of the expected sense is found. The upper limit of the Ekman layer (as defined by the ‘spiral’) is found to lie at a height of 0.17 to 0.24 (increasing with stability) in units of u*/f. Stress and heat‐flux are apparently considerable above this level, with sub‐geostrophic wind, when deep convection is occurring.Approximations to the universal distributions of stress, eddy coefficients, mixing length and rate of degradation of mean flow kinetic energy are computed for the various stabilities. The mixing length in unstable conditions increases almost as height up to a level of about 0.08 u*/f, and then decreases, but in general appears not to vanish in stable layers above the boundary layer. In the unstable boundary layer with deep convection, the eddy transfer coefficient for heat exceeds that for momentum up to 0.12 u*/f, where it becomes infinite, and is negative at higher levels. In stable conditions the transfer coefficients for a small sample of soundings were estimated to be closely similar.The universal functions of stability, A, B and C, which enable one to compute free atmosphere wind vector and temperature, given surface conditions, have been evaluated with moderate success, although B, which essentially describes the change of wind direction with height, exhibits excessive variability. A method is suggested for computing horizontal advection in the boundary layer when this is to be ‘parameterized’ in mathematical models.The drag coefficient, in terms of free atmosphere wind, has almost a 50‐fold range, due to stability variation only. Most of the variation occurs relatively close to neutral, so warning against too ready an assumption of neutrality in practical applications.It is suggested that, for modelling purposes, it is preferable to adopt boundary layer formulations which are not too sensitive to departures from ideal conditions, and eddy coefficients, perhaps based on mixing lengths, may well provide the best approach curr