ON THE ATOMIC WEIGHT OF CHLORINE. 185 



packed into the small U-tube B (fig. 1), which was then fused to the apparatus for 

 the preparation of hydrogen. 



Purification of the Mercury used in the Pumps. The mercury was frequently 

 cleaned as follows : It was placed in a suction flask, and on to its surface was poured 

 a weak solution of nitric acid. The side tube of the flask was attached to the water 

 pump, which drew air through the mercury by means of a glass tube held in position 

 by a cork in the neck of the flask. 



This stream of air, coupled with the intimate mixing of the mercury and the nitric 

 acid, resulted in the rapid oxidation and solution of all metallic impurities contained 

 in the metal. When this had been accomplished, the mercury was thoroughly 

 washed with water, dried with filter paper, and filtered, by means of very fine holes, 

 through clean white paper. 



Cleaning of Glass Apparatus. Before use, all glass apparatus was filled with a 

 hot mixture of potassium dichromate solution and concentrated sulphuric acid and 

 allowed to stand for six hours. It was then washed out with boiling distilled water, 

 and filled with hot concentrated nitric acid and allowed to stand overnight. The 

 next morning the vessel was emptied, thoroughly washed out with hot distilled 

 water, and steamed for three hours. Finally, a current of hot air, filtered through 

 cotton-wool and dried through sulphuric acid, was passed through it until it was 

 completely dried. 



6. Weighing the Bulbs, 



The balance, made specially for atomic weight determinations, was placed on a 

 stone pedestal in a cellar, situated in the basement of the chemical laboratories. 

 Observations with a maximum and minimum thermometer showed that the tempera- 

 ture in this cellar varied but little. Three filter funnels filled with calcium chloride 

 were kept inside the balance case ; the air in it was assumed to be half dried. The 

 doors of the balance case were closed and half-an-hour allowed to elapse before a 

 weighing was made. 



The vibrations of the pointer over the scale were viewed through a mirror by 

 means of a telescope. Assuming the number of divisions on the scale to be 1000, and 

 the average zero at no load 500, then the range of the zero variations, during our 

 experiments, was 9 divisions, between 49G to 505. 



The sensibility of the balance, during the weighings of the chlorine bulb, was 

 approximately 206 divisions for 1 milligramme, with a range of variation of 8 divisions. 

 During the weighings of the hydrogen bulb, the sensibility was approximately 

 198 divisions for 1 milligramme, with a range of variation ecpual to 10 divisions. The 

 method of weighing adopted was GAUSS' method of reversals. Generally, five 

 weighings were taken on one side and four on the other. The concordance of the 

 individual weighings showed that their mean could be relied on to 4 divisions or 

 00002 gramme. 



VOL. CCV. A. 2 B 



