Saliva was entirely diverted from the rumen of sheep via an oesophageal fistula and usually replaced by infusion of a synthetic saliva of known composition. Water movements in the rumen were measured by the use of soluble markers, and sodium and potassium movements were calculated. Water and various solutions were added to the rumen to vary its osmolality over the range 180-550 m-osmole/kg. When saliva was not replaced, recovery of the normal rumen osmolality following water loading was very slow indeed. When saliva was replaced by a synthetic solution the rate of water movement across the rumen wall (y, [1./hr]) was a linear function of the osmotic pressure (x, [m-osmole/kg]), such that y = 0 404-0 001206x (r =-0 90). This suggests that the filtration permeability of the rumen wall was similar to that of frogskin or toad bladder. Potassium absorption across the rumen wall was also slow but positively related to the potassium concentration. Sodium absorption appeared to depend on the potassium concentration and the osmotic pressure as well as on the sodium concentration. While moderate levels of potassium increased the rate of sodium absorption, grossly high levels, sufficient to produce a substantial entry of water across the rumen wall, decreased it, sometimes markedly. The rate of potasium and perhaps sodium absorption seemed to depend on the immediate state of nutrition of the animal. There is some conflict of evidence and opinion on the rate of movement of water between the rumen and the blood across the rumen wall [see Warner and Stacy, 1968b; Engelhardt, 1970]. In our early work calculations of trans-epithelial water movement were usually based on assumptions about the rate of salivary secretion; sheep fitted with oesophageal fistulae, allowing actual measurement of salivation rate, were used in only a few experiments. In the present work, saliva was entirely diverted via an oesophageal fistula and replaced by a synthetic saliva solution of known composition and entry rate. The composition of the rumen contents was altered by the addition of water or solutions and the effects of these treatments on the rates of movement of water, sodium and potassium across the rumen wall were measured. METHODS Sheep and Diets. Merino ewes weighing about 35 kg and fitted with rumen and oeso-phageal cannulae were kept in metabolism crates. They were accustomed to the experimental routine. They were usually given 700 g of ground lucerne hay every morning but occasionally individual sheep refused this for a period and other rough-age diets were substituted. There was no evidence that such differences in diet had any effects on the experiment findings. Water was freely available. 103