EXPERIMENTAL METHODS. 37 



open to serious error and are not reliable in a quantitative sense. Of special 

 importance is the fact that 12 per cent hydrochloric acid is not without 

 effect upon the furfural formed from the sugars. The amount of furfural 

 also varies with the proportion of pentose in the original mixture; it is 

 necessary to apply a special coefficient for all concentrations. In using the 

 gravimetric method corrections, not altogether satisfactory, are to be 

 applied for the solubility of the condensation product; on account of the 

 instability of solutions of phenylhydrazine, this method is also associated 

 with great inconvenience. Jolles's method of titration with sodium bisul- 

 phite has many advantages. However, as the proportion of pentose is some- 

 times very small and the furfural solution therefore exceedingly dilute, the 

 method is by no means reliable. Other and perhaps better substances for the 

 precipitation of furfural have been suggested ; however, these do not solve 

 the main difficulty. 



It must also be borne in mind that pentoses are by no means the only 

 substances that yield furfural on treatment with mineral acids. The 

 hexoses, cane-sugar, and many other substances found in plants also form 

 furfural under these conditions. Davis and Sawyer 1 have shown that a 

 mixture of 0.01 gram of arabinose plus 25 grams cane-sugar yielded 20 per 

 cent more furfural than without the cane-sugar, while 0.02 gram arabinose 

 plus 0.25 gram cane-sugar (the proportion usually present in plant extracts 

 with which they were working) was 15 per cent high for pentose alone. 

 Besides the hexose and disaccharides, furfural is also formed from starch, 

 cellulose, and oxycellulose, as well as glucuronic acid. Furthermore, rham- 

 nose and the methyl pentoses by the treatment with hydrochloric acid yield 

 methyl-furfural, which by the usual methods can be easily mistaken for 

 furfural. 



In view of these facts it appears that the only reliable method of esti- 

 mating pentoses must be based upon a separation of these sugars from other 

 substances. At present the most practical way of doing this is by means of 

 fermenting away the hexoses. A good culture of baker's yeast serves this 

 purpose admirably, followed then by the procedure with alkaline copper 

 solution already described. Experimental mixtures of 1-arabinose or 

 1-xylose with dextrose and cane-sugar showed that the latter were entirely 

 removed, while the pentoses remained unaffected. The removal by distilla- 

 tion of the alcohol and other products formed in the fermentation is 

 naturally essential. It is, of course, necessary to make certain that the 

 non-fermentable residue is actually pentose and not some other non-ferment- 

 able sugar for instance the interesting d-mannoketo-heptose described by 

 La Forge, 3 which is occasionally present in plants. 



1 DAVIS, W. A., and G. C. SAWYEB. The estimation of carbohydrates. The presence 

 of free pentoses in plant extracts and the influence of other sugars on their 

 estimation. Jour. Agr. Sci., 6, 406-412, 1915. 



1 LA FORGE, P. B. D-mannoketoheptose, a new sugar from the avocado. Jour Biol 

 Chem., 28, 511-522, 1917. 



