A Beeck-type calorimeter has been used to measure the integral heat of adsorption of oxygen on evaporated films of titanium, chromium, manganese, iron, cobalt, nickel, niobium, molybdenum, rhodium, palladium, tantalum, tungsten, platinum and aluminium. The variation of the heat with the extent of the adsorption has also been determined, except with palladium, platinum and aluminium, for which the amounts of oxygen adsorption were too small to allow this to be done. The heats of adsorption reported are based on a careful determination of the heat capacity of the calorimeter and the overall error is not more than $\pm$5 kcal/mole. Some investigations of calorimetric behaviour have been made. The surface areas of the films have been measured before and after an adsorption by the B.E.T. method; krypton isotherms at -196 $^\circ$C were used for this purpose. From these data, it has been possible to estimate the number of atoms of oxygen adsorbed per surface metal atom. In most cases, this quantity, together with the heat of adsorption, can be related to the oxygen/metal ratios and the heats of formation of certain oxides; this observation is remarkable in view of the limited extent of the adsorption. In the cases of rhodium, palladium and platinum the adsorption stops short of a monolayer and is probably true chemisorption rather than oxide formation. For those metals on which oxide formation occurs, the decrease in heat of adsorption with increasing coverage has been interpreted, in general, in terms of an increase in the oxidation number of surface metal ions by electron transfer within the adsorbed layer. Some theoretical correlations between the heat of adsorption and the atomic and the bulk properties of the species concerned have been examined and found of little value; an empirical relation with the metallic radius is given.