242 PROCEEDINGS OF THE AMERICAN ACADEMY 



The low value obtained by Hampe has been until recently univer- 

 sally accepted, because of the care employed in his analytical work, 

 and the agreement between his results. Hence the present discussion 

 will be confined almost entirely to the two substances which formed 

 the bases of his operations. 



The following values for the atomic weights of various elements 

 entering into the present investigation will be assumed throughout 

 the paper. 



Oxygen = 16.000 Hydrogen = 1.008 



Barium = 137.10 Silver =^ 107.930 



Bromine — 79.955 Sodium == 23.053 



Carbon = 12.002 Sul^jhur = 32.060 



Balance and Weights. 



The balance used in the work to be described was made with especial 

 care by Henry Troemner, of Philadelphia, and was procured particu- 

 larly for the present research. The beam and pans are composed of 

 aluminium, and all the remaining metal-work is plated with gold. By 

 raising the centre of gravity, the pointer may be made to swing with 

 great constancy as much as eight divisions of the scale for one tenth 

 of a milligram ; but since this degree of sensibility is far beyond the 

 range of accuracy obtainable even in atomic weight investigation, the 

 balance was usually adjusted so that the pointer moved ten divisions 

 for a milligram. A reasonable increase in load altered this relation 

 but slightly. The balance was kept in a small closet lined with cur- 

 tains, and was protected as much as possible from changes in tempera- 

 ture. The air of the case was dried by means of potassic hydroxide 

 and sulphuric acid, and large dishes of the former substance were kept 

 in the closet outside. 



The standard weights have already been described.* They were 

 carefully rubbed with chamois skin and again compared before the 

 present work. The slight corrections, which differed scarcely at all 

 from those previously found, were applied to each weighing. 



A double rider attachment upon the balance made the method of 

 weighing by substitution accurate and easy ; and accordingly this 

 method was invariably adopted, except in the first series of experi- 

 ments upon cupric suli)hate. In general, the vessel to be weighed was 

 carefully tared with an adjusted set of common gilded weights, and the 



* These Proceedings. XXV. 196. 



