It has been known for many years that, when muscles are placed in hypertonic solutions, their loss in weight is generally less than would be expected if the muscle were a simple osmotic sac. Even in the strongest solutions muscles lose only about 20 % of their initial weight (see Loeb, 1897;Cooke, 1898). Overton (1902) showed that even when allowance was made for dry matter and for an extracellular space of about 20 % (corrections not adequately appreciated by his predecessors), there was still a discrepancy between the calculated and the measured changes in the weight of the muscle. He accounted for this by proposing that about a fifth of the water was bound within the cell in the form of 'Quellungswasser', which could not be displaced except by very large gradients of vapour pressure (see his pp. 139-142, 155). Overton's conclusion was disputed by Hill (1930), who showed that almost all the fibre water could dissolve solutes in a normal manner. The assumption that all the water in muscle fibres participates in osmotic effects was later shown to lead to a satisfactory explanation of the weight changes in mildly hypotonic and hypertonic solutions . It seemed to us that the discrepancy in strongly hypertonic solutions might be explained by Overton's own qualitative observation, that in such solutions the extracellular space appeared to be increased.In this paper we have confirmed and extended this observation of Overton's by actual measurement of the extracellular (sucrose) space in solutions with various osmotic strengths. This permits, by subtraction, a direct calculation of the amount of fibre water as a function of osmotic strength. It then turns out that in solutions up to four or five times as concentrated as normal Ringer's solution the muscle fibres behave as simple semipermeable bags containing a fixed amount of solute: there is no need to invoke binding of any appreciable part of the fibre water.