The fact that many of the colored organo-alkali compounds conduct the electric current in ether and in liquid ammonia solution,3 and behave in other ways as salts, implies that the corresponding hydrides should exhibit an acidic character, albeit an exceedingly weak one. In fact, it has been shown that they may react with the ammono-base, potassium amide, in liquid ammonia solution to form the potassium salts. Thus triphenylmethane forms potassium triphenylmethide4and potassium benzhydrolate yields dipotassium benzophenone6 (CeHshCHOK + KNH2 -> (C6H5)3CKOK + NH"(2)This behavior indicates that these hydrides are ionized to a greater degree than the solvent ammonia, for otherwise reaction would be expected to proceed in the opposite direction corresponding to an ammonolysis of the salts. In the case of the alkali alkyls and alkali phenyls8 such an ammonolysis does occur, demonstrating that the corresponding hydrocarbons are ionized to a less degree than ammonia.7 Accordingly it is impossible to obtain organo-alkali compounds by the action of potassium amide upon these hydrocarbons.

When this investigation was undertaken no colored organo-alkali compounds had been reported except those in which one or more aromatic rings were attached to the same carbon atom as was the alkali metal.3 A formula containing a quinonoid ring may be readily constructed for these compounds and the existence of tautomers so constituted had been suggested as the source of their color.4 It appeared to be of interest, therefore, to investigate the influence upon the properties of the carbon-hydrogen and carbon-sodium bonds, in hydrocarbons and organo-alkali compounds, when aromatic rings are present only in the /3-position with respect to the carbon atom in question. If organo-alkali compounds of this type were colored, the hypothesis just mentioned would require the assumption of a tautomeric quinonoid structure containing a trimethylene ring.It has been shown that a, a, -triphenylethane does not react with potassium amide in liquid ammonia,6 and attempts to prepare a nonaphenyltert.-butane were unsuccessful. However, since it had been observed6 that sodium benzylide could be prepared from sodium and benzyl bromide in liquid ammonia, although potassium amide is also without action on toluene, it was suspected that an organo-alkali compound might be obtained when /3-chloro-a, a, -triphenylethane was treated with sodium. This proved to be correct but when the red compound which resulted was decomposed with ammonium chloride (which replaces the alkali metal with hydrogen) the product was , ,/3-triphenylethane. It is obvious that a rearrangement occurred6 and a preliminary study which has been made of this rearrangement is the subject of this paper.