OF THE ATOMIC WEIGHT OF ALUMINUM. 
1025 
potassium; and a sufficient quantity of the metal for the intended experiments was 
found to yield no appreciable trace of impurity. The surface of the specimens to he 
used which had touched the cutting pliers was again cleansed by acid and water. 
This pure aluminum did not differ much in physical character from the ordinary metal 
of commerce ; it seemed, however, to be somewhat whiter, was distinctly softer, and 
had a little higher density, the mean of three closely-agreeing determinations made at 
4° C. giving the number 2 A 8 3 as referred to water at the same temperature. 
Production of hydrogen by aluminum, and measurement of the gas.—Details of 
method used. —The pure metal was used for the determination of its atomic weight by 
acting upon a known weight of it with a strong solution of sodium hydrate and 
determining the amount of hydrogen evolved. The advantage of using an alkali 
rather than hydrochloric acid, as in Tekreil’s experiment above quoted, lies in the 
non-volatility of the former, only vapour of water having to be separated from the 
hydrogen, while sulphuric acid is not available on account of the resistance to its 
action of aluminum. In 1863 Fr. Schulze* proposed to measure the volume of 
hydrogen given off by the action of an alkaline solution on commercial aluminum as 
the means of approximately deciding on the comparative purity of different specimens. 
The nature of the reaction was established by preliminary experiments, which proved 
to me that normal sodium aluminate alone is formed, so that each atom of aluminum 
liberates three atoms of hydrogen. The sodium hydrate was prepared from metallic 
sodium, and was used in the form of a solution so strong as scarcely to lose a sensible 
amount of water by the passage through it of a dry gas at common temperature— 
such an alkaline solution, so far as strength is concerned, as would be used to absorb 
carbon dioxide in an organic analysis. The quantity taken for each experiment was 
but a few cubic centimetres, and but little beyond the exact amount required for the 
solution of the metal; a small excess was, however, always allowed, so that the action 
might not become very languid towards the end. This strong alkaline solution was 
prepared with water which had been boiled to expel air, and the solubility in it of 
hydrogen w T as ascertained to be so small that any correction on this account would 
have fallen within the limits of inevitable error, and might be safely neglected. 
To secure accurate measurement of a somewhat large volume of hydrogen, two stout 
flasks were selected, one holding about a litre and the other about half as much, and 
with narrow necks of rather more than usual length ; and on the necks a simple 
millimetre scale was marked. One of these flasks having been carefully filled with 
mercury and inverted over the mercurial trough, the hydrogen was collected in it, such 
a quantity of aluminum being used in each experiment as previous trials had shown 
would yield gas enough to bring the level of the mercury within the range of the scale 
on the neck. To thus obtain such a volume of gas as could be accurately measured in 
the narrow part of the flask it was necessary to note in advance roughly the prevailing 
* Fr. Schulze, “Die gasvolumetrische Analyse,” S. 18, quoted in v. Wagner, ‘ Jahresbcricht,’ 
u. s. w., 1864, S. 23. 
6 q 
MDCCCLXXX. 
