stronger than the material in the first test, and its ductility, as measured by the contraction of area, fell from 60 to 6 per cent. This indicates that discrepant fatigue limits may be due to discrepant form. There is also another point to be noted. If, during a fatigue test of the usual type, the machine is stopped when slips have begun to develop, they will heal themselves during the period of rest and so confuse the final result. There are therefore several causes which together act to

more »

... y the results of a test on the fatigue limit of a material by the method of repeated oscillations. I think that a fatigue limit of practical value can be determined from a looping test of the type described in my recent paper. That is to say, the load is applied and removed, and then reapplied and removed through a wider range, until at last a range is found at which looping appears. When this loop appears the fatigue range has just been passed. I have shown in my recent paper that an iron or mild-steel bar, when tested in this way, and when loops corresponding to considerable overstrain have been described, will show an elastic line again when re-tested again after a rest. (See Curve 5, sheet 1, " Phil. Trans. " (1920).) B ut this process does not alter the real fatigue limit ■of the material, as can be seen from the curves. The investigation described in this communication had for its primary object the production of formaldehyde or methyl alcohol. In this aim we have been so far unsuccessful, but our results have led to the discovery of an action, which apparently has hitherto escaped notice, by which methane is formed. Apart from theoretical interest the method would appear to have important technical possibilities, either as a means of producing pure methane or of enriching of town's gas in methane, because by its use more methane is formed from water-gas than by other methods. In order to make clear the course of the action about to be described it will be well briefly to enumerate some of the known interactions of carbon