Earlier work1 has shown that the simple physical adsorption theory of boundary lubrication is inadequate; fatty acids are not really effective in lubricating metals unless they are able to react chemically with the surface to form the metallic soap. Gregory's measurements have shown, for example, that platinum is not effectively lubricated by a dilute solution of a fatty acid, whereas cadmium is lubricated up to the temperature at which the appropriate cadmium soap softens. Using a variety of metals he found a general relation between lubrication and chemical attack of the metal. With many metals, the chemical reaction of the fatty acid to form the soap occurs via the surface oxide film or in the presence of oxygen2. If this view is correct, we should expect that fatty acids would not prove effective lubricants even on metals such as iron or cadmium if oxygen is excluded.