When the posture was changed from horizontal to vertical, or the reverse, the alteration in plasma volume and in the levels of haematocrit, haemoglobin, and plasma protein was much greater in patients with oedema or low plasma protein or albumin concentrations ("the pathological group") than in patients without these abnormalities ("the control group"). This larger fluid shift, and the larger concentration changes dependent on it, are explained in terms of Starling's hypothesis. In the control group there was scarcely any change in the total mass of intravascular plasma protein with change in posture, but in the pathological group this was often considerable. This was interpreted as indicating increased permeability of the capillaries to protein. Attention is drawn to the influence of uncontrolled alterations in posture on biochemical and haematological measurements and a recommendation is made in order to standardize conditions before taking blood for the analysis of non-diffusible constituents.