PROCEEDINGS OF THE AMERICAN ACADEMY 



The second ten-gramme weight, which was much less used, re- 

 mained quite constant in value. 



Throughout the present investigation the method of. weighing by 

 tares was universally adopted. A vessel to he weighed was placed 

 upon the left hand scale pan, balanced with common gilded weights, 

 and then replaced by a similar vessel which weighed a few milli- 

 grammes less. The exact amount of this extra tare having been 

 determined w T ith the rider, the counterpoise was replaced by the 

 original vessel and the rider removed, in order to determine if the 

 centre point had changed. When only a slight change had taken 

 place, the reading for the counterpoise was compared with the mean 

 of the two readings for the original vessel. In the rare cases when 

 the change exceeded the equivalent of the thirtieth of a milligramme, 

 the vessels were alternately substituted for each other until con- 

 stancy was reached. A substance to be weighed was of course 

 placed in such a tared vessel, and after substitution the deficiency 

 of the counterpoise was made up with standard weights. The dif- 

 ference between the tares on the left hand scale pan indicated the 

 observed weight in air of the substance taken. It was found con- 

 venient to tabulate the results in the following form. 



The lowest right hand figure represents the true weight of the 

 substance taken, if the Sartorius ten-gramme weight is taken as the 

 standard. Reduced to the Washington standard the value becomes 

 1.80934; but this last correction is in no case applied in the work 

 which follows. 



The method used in the case of hygroscopic substances, and most 

 other precautions, are given at length in the paper already quoted. 

 In weighing a crystallized salt it was usually necessary to weigh 

 the crucible while filled with ordinary moist air, hence the counter- 

 poise crucible was exposed to the same conditions. 



