148 Scientific Proceedings, Royal Dublin Society. 
transferred to a strong tuve of hard glass closed at one end and 
contracted at the other; bright piano wire was now introduced, 
the liquid again boiled, and while it was boiling the tube was 
hermetically sealed. The tube was then exposed in an air bath, 
to a temperature of 150° C., for five hours. After this treatment 
the liquid contained a quantity of hydrated ferric oxide. 
2. This experiment was repeated with great care, the result 
being that a much smaller quantity of oxide was produced. 
From these results I conclude that a trace of oxygen, which 
does not reveal its presence by its action on iron at ordinary tem- 
peratures, produces a very decided effect at higher temperatures ; 
and it is also evident that sealing tubes in the way described is a 
most uncertain method, as it affords no guarantee that atmospheric 
oxygen is completely excluded. I, therefore, had recourse to a 
modification of the plan described by Bunsen* for obtaining air- 
free water. This method is most satisfactory ; even tubes of small 
diameter may be filled and sealed without allowing the smallest 
trace of air to enter. In this manner I prepared tubes of hard 
glass containing bright piano wire in (@.) a solution of 1 per cent. 
of potassium nitrite and 1 per cent. of potassium nitrate in dis- 
tilled water; (b.) a solution of 1 per cent. of potassium nitrite ; 
(c.) a solution of 1 per cent. of potassium nitrate; (d.) Vartry 
water concentrated on the water bath to one-quarter of its original 
volume. 
These tubes were placed in an air-bath provided with an auto- 
matic arrangement for maintaining a constant temperature, and 
for five consecutive days they were kept for five hours each day 
at a temperature of 150° C.; they were examined after each suc- 
cessive heating, but in no case could I detect any indication ot 
oxidation of the iron. As the experiment proceeded, the glass of 
the tubes became slightly corroded, and simultaneously small 
quantities of colourless floculent matter, resembling silica, appeared 
in the solutions. The action on the glass, and the consequent 
alteration in the composition of the solutions, would render 
negative results of little value, if previous experiments had not 
shown that with similar solutions oxidation of the iron invariably 
ensues if atmospheric oxygen has not been completely expelled 
from the experimental tubes. 
* Bunsen’s Gasometry, p. 142. 
