2i8 A GENERAL REVIEW OF 



excess of the alkylating mixture in every case is therefore 

 evident. 



Certain of the higher alkyl iodides react more readily 

 with silver oxide to form alkyl ethers than do methyl and ethyl 

 iodides. It is possibly for this reason that the latter give better 

 results in the alkylation process than, for instance, isopropyl 

 iodide (Lander (4) ). 



There is little direct experimental evidence bearing on 

 the mechanism of the silver oxide reaction, so that any con- 

 ception of the course of the reaction must, for the present, be 

 largely speculative. It is highly improbable that, in the 

 alkylation of hydroxy-compounds, the silver oxide acts simply 

 by removing hydrogen iodide, since no reaction occurs if 

 litharge, zinc oxide, cupric oxide, or magnesium oxide are sub- 

 stituted for silver oxide. McKenzie (3) suggests that the most 

 plausible hypothesis is that, by replacement of alcoholic 

 H by Ag, an unstable silver derivative is formed and subse- 

 quently undergoes double decomposition with the alkyl halide. 



The suggestion is endorsed by Lander (4 and 17) and by 

 Purdie and Irvine (8). In accordance with this view, cuprous 

 oxide might be expected to behave similarly to silver oxide. 

 Alkylation of methyl tartrate by means of cuprous oxide and 

 methyl or ethyl iodides has not been effected, but in the case 

 of isopropyl iodide there is reason to believe that alkylation 

 does take place, but imperfectly. (Private communication 

 from Professors Purdie and Irvine.) The remarkable series 

 of colour changes which are sometimes noticed during alkyla- 

 tion by this method might possibly be advanced as a further 

 argument in support of the silver derivative hypothesis. It 

 must, however, be admitted that, as yet, there is no positive 

 evidence of the formation of a definite derivative of this kind 

 during alkylation. Apparently the only attempt that has 

 been made to isolate such an intermediate compound is that 

 of Irvine and Moodie (36), already mentioned in connection 

 with tetramethyl glucose. It is unlikely that alkylation of 



