20-Oxo-steroids, in presence of a t-alkoxide in the corresponding alcohol, react with oxygen to give the 17a-hydroperoxy-20-ketone in fair yield; reduction, best with zinc and acetic acid, then gives the 17a-hydroxy-20-ketone. The speed of hydroperoxidation is notably affected by the nature of the substituents at C(,,j and in ring c. In the same way, a 6-oxo-5a-steroid is converted, via the fia-hydroperoxy-derivative, into the 5a-hydroxy-6-ketone. 3-0xo-5a-and 12-oxo-steroids, on the other hand, are oxidised to the 2 3and 11,12-dioxo-compounds respectively.KETONES react rather sluggishly with oxygen, and the products are usually those resulting from cleavage of the carbon chain adjacent to the carbonyl group; the a-hydroperoxyketone has been postulated as an intermediate, and such a compound has been isolated, along with its decomposition products, from the oxidation of di-isopropyl ketone at However, when a ketone forms a stable enol, the latter (but not the ketonic form), in solution in ether, light petroleum, or benzene, reacts readily with oxygen at ordinary temperature to give the a-hydroperoxy-ketone, which, under these mild conditions, is easily isolated ; 4*5 even secondary a-hydroperoxy-ketones have been prepared in this way, in spite of the ease with which they are dehydrated to a-diketones6Most simple ketones exist to a negligible extent in their enolic forms under neutral conditions, but enolisation is favoured by basic or acidic conditions. However, although several workers have shown that, in strongly basic medium, ketones w i l l take up oxygen rapidly at ordinary temperature, they have, again, usually isolated cleavage products ; this is readily explained by the initial formation of the a-hydroperoxy-ketone, for such compounds are known to be cleaved by base.' That autoxidation of ketones can proceed without cleavage of the carbon chain is indicated by the conversion of +unsaturated ketones of the cyperone series into their y-hydroxy-derivatives by oxygenation in presence of aqueous base8 and by the recent observation that certain rotenone derivatives give a-ketols under similar conditions? It seems probable that the hydroperoxy-ketones are intermediates in such oxidations, since there are analogies for the " reduction " of such compounds to the ketols by dilute alkali.u*c More important, for the present purpose, was the observation that, even under strongly basic conditions, the reaction can leave the carbon chain intact. Thus, cyclic ketones of type (I), in the limonin series, react with oxygen in presence of potassium t-butoxide in