A crystalline by-product, formed in the reaction of p-chlorophenacyl chloride (I) with ethylmagnesium chloride, has been identified as 1,3-dichloro-2,4-bis(p-ch1orophenyl)-2,4-hexanediol (IV). Its formation has been attributed t o aldolization of I followed by reaction with Grignard reagent a t the carbonyl function. The dehydrochlorination of IV yielded one of two possible compounds to which the diepoxide structure VI was assigned on the basis of n.m.r. spectra. The dehydrochlorinated product (VI) was synthesized from 1-chloro-2,3-epoxy-2,4-bis(p-chloropheny1)-4-butanone (VIII) by treatment with ethylmagnesium chloride followed by reaction of the resulting epoxy alcohol IX with alcoholic alkali. I t is suggested that the epoxy alcohol I X did not dehydrochlorinate directly to the 3,4-epoxytetrahydrofuran derivative V but instead rearranged first by epoxide alcohol transposition t o X and then dehydrochlorinated to 1,2,3,4-diepoxy-2,4-bis(p-ch1orophenyl)hexane (VI).The condensation of p-chlorophenacyl chloride (I) with ethylmagnesium chloride produced the normal compound, l-chloro-2-~-chlorophenyl-2-butanol (11) (1) as well as a crystalline by-product. The latter was formed in higher yield when the order of addition was reversed and the Grignard reagent was added to I. This investigation describes the properties and identification of this crystalline by-product and provides evidence for the mechanism of its formation.The infrared spectrum of this con~pound showed the presence of hydroxyl groups and the absence of carbonyl groups. The OH band appeared a t 3.0 p (3 333 cm-I), a shift from the norlnal which indicated hydrogen bonding. The reluctance of the conlpound to react with phenyl isocyanate or with acetic anhydride suggested the presence of tertiary hydroxyl groups. Of the possible modes of formation of the hydroxyl group from the carbonyl group, aldolization appeared to be the most likely. Analysis of the unlino~vn compound provided the formula (CgI-IgC120), with half the chlorine not removable and the other half easily removable by dehydrochlorination with cold methanolic or ethanolic potassium hydroxide. The dehydrochlorination was accompanied by loss of the hydroxyl groups, as evidenced by infrared spectra, and this indicated epoxide formation froin a chlorohydrin.The molecular weight of the unlinown compound could not be determined by the Rast method because it liberated hydrogen chloride on heating. The dehydrochlorinated compound, on the other hand, though sensitive to acids and bases, was more stable to heat and could be distilled in vacuo. Analysis and n~olecular weight determinations of the dehydrochlorinated compound gave the empirical formula (CgHBC10)2 from ~vhich the formula (CgHgC120)Z could be deduced for the chlorohydrin. On the basis of the above evidence the structure IV was assigned to the by-product of the Grignard reaction and the structure V or VI to the dehydrochlorinated product. Degradation studies showed that IV could be cleaved by phosphorus pentoxide in boiling benzene t...