The stability of the ring structure in kojic acid toward various reagents was investigated.Cleavage to an open-chain, dienol derivative in i V s o d i~~m hydroxide a t 25", or fragmentation of the structure, was very slow. The benzoyl group ill the fifth (phenolic) position of dibe~~zoylkojic acid was removed by hydroxy1-amine hydrochloride in pyridine so selectively that the method was of value in synthesizing certain new derivatibes. Although catalytic hydrogenation readily reduced the pyrone ring in kojic acid to undefined substances, zinc dust in glacial acetic acid reduced the hydroxyrnethyl group in the dibenzoate to a methyl group and yielded benzoylallomaltol. The ring in dibe~lzoylkojic acid was apparently opened, with retention of both benzoyl ~~O L I P S , by semicarbazide hydrochloride 2nd pyridine to yield two isomeric "dise~nicarbazones", C?~Hz?Ns07, decomp., 215" and 172-173", respectively. The higher-melting isomer when boiled with dilute acid gave a compoulld C21HljN3Ojr m.p. 241", which was apparently cyclic; the same isomer with nitrous acid yielded two isomeric, apparently open-chain "monosemicarbazones", C2LHleN30,, decomp., 215" and 178-179", respectively. The structures of these five compounds were not determined. The following substances were thought to be new: sodium lcojate, a white powder, and its crystalline tetrahydrate; 5-hydroxy-2-(methoxymethy1)-y-pyrone, m.p. 75-76', and its crystalline aluminum salt, decomp. 270-271"; molecular addition compound of dibenzoyllcojic acid and benzoic acid, m.p. 120-121"; 2-(acetoxyn~ethyl)-5-benzoxy-y-pyrone, m.p. 144"; 2-(benzoxymethy1)--5-hydroxy-y-pyrone, m.p. 180-181"; and 2-(benzoxy1nethyl)-5-rnetho.uy-~-pyrone, 1n.p. 110-111". A more convenient synthesis was discovered for 5-benzoxy-2-(hydroxymethy1)-y-pyrone, and the published melting point was revised from 136" to 1T3-144"; the melting point of 5-benzoxy-2-(triphenylmethoxyrnethy1)-y-pyrone was revised from 206-208" to 213-214O.
