OP ARTS AND SCIENCES. 243 



vibrations of the pointer were noted. After the removal of the vessel 

 standard weights were added to the left-hand pan until the same con- 

 dition of equilibrium was reached. The vessel was then once more 

 substituted for these weights, and any slight change of centre point 

 was of course manifest at once. In the rare cases when such slight 

 change appeared, the weights and vessel were alternately substituted 

 for each other until constancy was reached. In the case of hygro- 

 scopic substances the already ascertained weights were first placed 

 upon each scale pan, then the vessel was quickly removed from the 

 desiccator and substituted for the standard weights, and, finally, the 

 latter were again put in place. 



All desiccators were allowed to remain in the closet with the bal- 

 ance three or four hours before the weighing, and objects were often 

 weighed on successive days, to furnish assurance of constancy. In 

 every case the barometer and thermometer were read, and any correc- 

 tion due to change of relative buoyancy amounting to more than the 

 fiftieth of a milligram was applied to the result. With large vessels 

 an invariable slight loss of weight, amounting sometimes to as much 

 as one twentieth of a milligram, was noticed after the object had re- 

 mained for some time upon the balance pan. The loss may have been 

 due to the replacement of the perfectly dry air from the desiccator by 

 the less completely dried air of the balance case. Possible error from 

 this source was avoided by the employment of uniform conditions 

 suited to the particular substance in hand. 



All weighings were of course reduced to the vacuum standard by 

 calculation from the specific gravities of the substances and weights 

 involved.* Through the great kindness of Professor Mendenhall of 

 Washington, two of the Laboratory's ten-gram weights (one of brass 

 and one of platinum) have been compared as carefully as possible 

 with the standards of the Washington Bureau of Weights and Meas- 

 ures. Five comparisons of the Sartorius ten-gram weight with these 

 gave the following results for its value in vacuum : — 



1891. Grams. 



February 4. By comparison with the brass weight, 10.00025 



March 4. " " " platinum weight, 10.00025 



March 14. " " " " " 10.00026 



June 12. « « " " " 10.00026 



June 12. " « " " " 10.00024 



Average, 10.00025 



* These Proceedinp^s, XXV. 196. Specific gravity of brass = 8.3. 



