1910.] PUBLIC DOCUMENT — No. 31. 145 



The results rci)or1ccl were all made with G-iiicli bottles, because 

 we were attempting to find the most accurate method for read- 

 ing tests made under conditions existing at the experiment 

 station laboratory. It would have been quite instructive to have 

 been able to run more tests, using the 9-inch bottles, which 

 require a special machine on account of the longer neck. 



Conclusions. 



1. With one exception the readings taken to the top of the 

 upper meniscus were too high, the amount of error being in most 

 cases proportional to the diameter of the bottle neck. 



2. Where the readings were taken to the bottom of the upper 

 meniscus the results were much more uniform. In several 

 cases, however, there was considerable variation, due very likely 

 to the difficulty of determining accurately the lowest point of 

 the upper meniscus. 



3. The alcohol method, where used, showed more concordant 

 results, especially for the 10 and 30 per cent, bottles. The 

 difference between these results and the theoretical test was no 

 greater than might be expected between duplicate tests by the 

 gravimetric method. The tests made in the 50 per cent, bottles 

 varied materially, which might reasonably be attributed to the 

 difficulty of reading these bottles accurately on account of the 

 large diameter of the neck. 



It was suggested that the method might give somewhat dif- 

 ferent results if a mixture of a definite amount of butter fat and 

 skim milk was used in place of butter fat and water, the former 

 mixture more closely resembling milk or cream. In order to 

 test this ])oint butter fat was weighed into several 10 and 30 

 per cent, thoroughly cleaned bottles, together with sufficient 

 separator skim milk (which had been previously tested) to make 

 the contents of the bottles weigh 18 grams. Sulfuric acid was 

 then added and the test completed as usual. The results are as 

 follows : — 



