Early in the course of an investigation of the structure of penicillin2 it was observed that treatment of the latter with cold, aqueous alkali resulted in the production of a new substance, benzylpenicilloic acid. The release of 1 mole of acid during the course of the reaction, the difference between the empirical formula of the product, C16H20N2O5S, and that of benzylpenicillin, CmHmNsCXS, and the differences between the infrared spectra of the two, indicated that the treatment with alkali had effected hydrolysis of an inner anhydride. The isolated product, however, proved refractory to crystallization, and was of indefinite melting point. Furthermore, the specific rotation of repeated preparations differed greatly, ranging from 73°to 120°( sodium bicarbonate solution).A number of attempts were made to prepare a crystalline derivative of definite physical properties. The sodium, potassium, ammonium, and S-(p-bromobenzyl)thiuronium salts could not be obtained in crystalline form. Attempted acylation with ketene in acetone solution, and with benzoyl chloride in alkali or in pyridine, led to intractable oils. Crystallization of the oily picrate was not effected. The initial, partially crystalline precipitates produced by treatment with ammonium Reineckate and with ammonium rhodanilate were of inconstant composition and decomposed on attempted recrystallization.* Analyses were performed by the Microanalytical Laboratory, Technical Service Division, these Laboratories, under the direction of Dr. J. A. Kuck, except as otherwise noted.8 These experiments were begun on a crystalline sample (1650 units/mg.) prepared by the authors (3) from crude sodium salt (150 units/mg.) furnished by the Lederle Laboratories Division, American Cyanamid Company. For later samples of crystalline sodium benzylpenicillin we are indebted to Mr. E. F. Williams of the Physics Division, these Laboratories. 7 We are indebted to Dr. A. F. Kirkpatrick of the Physics Division, these Laboratories, for all microscopic examinations reported. 8 We are grateful to Misses Nydia Ananenko and Marion Cook for the assays, which were performed by the paper-disc method of Vincent and Vincent (7).