PURDIE'S REACTION 199 



crystals. Hydrolysis of the latter gave the syrupy sugar 

 trimethyl Z-arabinose, which, when alkylated in its turn, 

 yielded a mixture of trimethyl a- and /3-methylarabinosides 

 containing a large excess of the latter isomeride. In the case 

 of the methylpentose rhamnose, Purdie and the writer (34) 

 obtained very similar results, with the exception that mixtures 

 of the stereoisomeric a- and ^-rhamnosides were dealt with 

 throughout. The sugar trimethylrhamnose was identified 

 by the formation of a crystalline hydrazone. 



The only ketose to which the silver oxide reaction has so 

 far been applied is fructose, and here much difficulty was 

 encountered (Purdie and Paul (38) ; Irvine and Hynd (46) ). 

 As was the case with rhamnose, the glucosidic derivative used 

 (methyl fructoside) could not be obtained in the crystalline 

 form. The material initially subjected to alkylation was a 

 syrupy mixture of the isomeric a- and j8-methylfructosides 

 possibly contaminated with other substances, and this, after 

 treatment with silver oxide and methyl iodide and purification 

 of the product by vacuum distillation and otherwise, yielded 

 a liquid mixture of tetramethyl a- and ^8-methylfructosides. 

 Considerable difficulty was experienced in isolating the pro- 

 duets ; this may have been due to the susceptibility of the 

 ketoses and their derivatives to oxidation. The sugar ob- 

 tained by hydrolysis of the alkylated fructoside mixture was 

 syrupy, but by realkylating this and again hydrolising the 

 product, a small quantity of crystalline tetramethyl fructose 

 was eventually isolated. It was found impossible to prepare 

 either of the pure tetramethyl a- or /3-methylfructosides ; 

 mixtures of these, however, were obtained, in one of which 

 the a-isomeride, and in the other the /3-variety, predominated. 

 Irvine and Hynd subsequently obtained a definite mono- 

 methyl fructose which is described in another communication 

 to this volume. 



