in percentages, providing a ten gram sample were used and all the mois- 

 ture were recovered, that is, a volume of water recovered measuring one 

 cubic centimetre would read ten per cent, on the scale of our receiver. 

 After obtaining no reagent that would deliver all the water we concluded 

 that we would have to just decide upon some one reagent, and compare 

 the results from its use with those obtained from the official method ; 

 that is, heating a very accurately weighed sample in a water oven to 

 constant weight, reweighing and calculating the percentage of moisture 

 from the loss in weight from the heating process. Now, amyl butyrate 

 seemed to be a fairly desirable reagent, although it had some undesirable 

 features, namely, that it was rather expensive — especially the chemically 

 pure product; its boiling point was rather high, requiring a longer 

 time than we thought desirable to drive enough of it over to wash out 

 the inner tube of the condenser, and its odor, although pleasant under 

 ordinary circumstances, proved to be disagreeable when heated with 

 butter, or when in contact for any length of time with the hands. We 

 tried a large number of reagents, among which were eucalyptol, amyl 

 acetate and various distillation products of crude petroleum. In choos- 

 ing a reagent we had to consider the various requisites of one suitable 

 for our purpose, namely, its boiling point, specific gravity, ability to 

 prevent foaming of the sample, stability under the conditions to which 

 it would be subjected, itJ suitability for flushing the inner tube of the con- 

 denser and finally its cost. Although we could easily have chosen a 

 reagent consisting of a mixture of substances which would have per- 

 formed the work well enough, yet, as we were anxious to use our 

 reagent over and over again to reduce the expense of testing, we had to 

 discard the idea of a mixture, since the constituents of highest boiling 

 points would partly remain behind in the evaporating cup, thus altering 

 the composition of that portion of the reagent recovered. We finally 

 concluded that amyl acetate, when chemically pure, fulfilled these requis- 

 ites in the largest measure. 



After thus getting our method approximately complete, we concen- 

 trated our efforts upon solving the problem connected with its accuracy. 

 We first used it with butter, and after making an exhaustive number of 

 tests with this product we discovered that the loss of moisture due to the 

 reagent approximated a constant quantity, rather than being propor- 

 tional to the percentage of the moisture in the samples. We next used our 

 test with curd and cheese and found the same principle to hold here also. 

 The explanation of this appears to us to be that the moisture lost is that 

 absorbed by the portion of the reagent that volatilizes with the moisture, 

 the balance of the reagent being recovered in a moisture-free condition. 

 We thus found it necessary to correct the graduations on the receiv- 

 ing tube accordingly, so that it now gives readings closely approximat- 

 ing the results obtained bv the official method, as may be seen from an 

 examination of the table of results obtained by us. 



