182 Journal of the Asiatic Society of Bengal. [April, 1910. 



The method, however, is one very susceptible to error. 

 It makes considerable difference whether the volatile fatty 

 acids are filtered or not previous to titration with potash. 

 They should not be filtered or some caproic or higher acid not 

 very soluble in water may be removed. Again the titration 

 with potash gives different results according as it is performed 

 quickly or slowly. It must be performed slowly to allow the 

 slightly soluble caproic acid to be attacked by the potash. 

 Finally it makes a difference whether titration is stopped as 

 soon as the red colour of the litmus changes, or whether it is 

 continued until a blue is obtained. It seemed, therefore, very 

 desirable to estimate the ratio of butyric to caproic acid by 

 some other method. This has been done in the case of the 

 European cow by Duclaux. His method was, therefore, ap- 

 plied to the case of the Indian buffalo. 



Estimation of the ratio butyric to caproic acids by Duclaux' s 

 method. — By this process the volatile fatty acids in dilute aque- 

 ous solution are fractionally distilled and the various fractions 



** 



titrated with alkali. From these titration figures a curve 

 is drawn, and by comparing this curve with standard curves for 

 butyric acid, for caproic acid and for definite mixtures of these 

 two acids, the ratio of the two acids can be estimated. 



Duclaux' s results are given in the Comptes vendues (cii, 

 1022) , but in that paper the experimental data are not recorded. 

 A curve for butyric acid is given in Duclaux 's Traite de Micro- 

 biologic, Vol. II, pp. 384 — 395, but no curves for caproic acid 

 and mixtures of butyric and caproic were available. These 

 curves have, therefore, been prepared (Diagram 1). Merck's 

 purest acids were used in the preparation of these curves. 

 Here as in all the other experiments the distillates were not 

 filtered and were titrated slowly to allow the caproic acid to be 

 acted on by the alkali. Owing to some initial difficulties it 

 was investigated whether the shape of the curve for a given 

 ratio of the two acids depended on the shape of the distilling 

 flask, on the absolute concentration of the acids in the solution 

 distilled or on the presence of sulphuric acid or potassium sul- 

 phate in the solution. It was found that the shape of the 

 curve was practically the same under all these different condi- 

 tions (Diagram 2). Duclaux's curve for butyric acid is practi- 

 cally identical with that we have obtained (see also Dia- 

 gram 1). 



Curves for the volatile fatty acids in buffalo butter-fat were 

 now obtained by saponifying the fat, acidifying the product 

 with sulphuric acid, distilling over the volatile acids in frac- 

 tions and titrating. Curves closely corresponding with butyric 

 acid alone were obtained (Diagram 3). This appeared to 

 bear out the result obtained in the preliminary investigation 

 (Watson, loc. cit.). However, a sample of Danish butter-fat was 



