ene 
ON THE GASES EVOLVED FROM IRON FURNACES. 147 
sequently diminish the quantity of carburetted hydrogen, while it increased 
that of hydrogen, we shall be the more inclined to attribute the erroneous 
results of his experiments to the uncertain methods employed by him in ana- 
lysis. The smallest quantity of oxygen remaining in the nitrogen with which 
the apparatus was filled previous to the experiment, the gases retained by the 
porous copper formed during the reduction, the carbon also retained by this 
copper, the smallest quantity of foreign substances which may attach them- 
selves to the combustion-tube, softened as it is by heat during the experi- 
ment,—all these must tend to increase the chances of an error of ~>4>5 in 
weighing ; a difference so small as even without the operation of these causes 
almost to be within the errors of observation, and sufficient to account for 
the erroneous results obtained by Ebelmen. But whatever may have been 
the grounds which induced Ebelmen to avoid referring to the original inves- 
tigations in Germany, when we consider the great labour which he bestowed 
on the inquiry; it will ever remain to be regretted that he did not introduce 
into his memoir an explanation of the grounds upon which he accorded the 
preference to his method of analysis, which differs from that of his prede- 
cessors in the inquiry more by its tediousness than its accuracy, and which 
we consider it necessary altogether to avoid in the following research. At 
the same time it cannot be denied, that eudiometric analysis, as usually 
performed, is little deserving of high commendation, or of universal adoption, 
although this is less owing to its incompleteness than to the neglect of the 
many precautions which should be adopted to procure accuracy. 
Before proceeding to our investigation, we thought it necessary to examine 
with great care all the conditions essential to obtain a proper degree of ac- 
curacy. It cannot, therefore, be thought superfluous to describe in detail 
the methods employed in the inquiry, especially as these must form the foun- 
dation for the reception of the conclusions which we draw from the experi- 
ments. 
The combustion and measurement of the gases is most conveniently and 
accurately performed in uniform glass tubes of 18-19 inches in length and 
about 0°6 inch internal and 0°8 inch external diameter ; in the closed end of 
the tube there is inserted by fusion two platina wires of the thickness of horse 
hair, for the purpose of passing the electric spark. The tube is divided into 
millimetres, and with this view, is covered with common etching paste, or 
still better, with a thin layer of wax containing a little turpentine, which may 
be laid very uniformly on the warmed surface of the glass by means of a 
hair pencil. The glass is then minutely graduated by a peculiar instru- 
ment, and subjected to the action of gaseous hydrofluoric acid, which, when 
evolved from a paste of fluoride of calcium and concentrated sulphuric acid 
placed in a vessel of lead slightly warmed, effects the etching in ten to fifteen 
minutes, and much more legibly than the liquid hydrofluoric acid usually 
employed in the graduation of thermometers. 
The capacity of the tube, which has thus been divided into millimetres, 
is easily determined by measurement. For this purpose, the tube is placed 
vertically with the table, its hermetically sealed end being downwards, and 
is then filled with successive portions of mercury carefully measured. The 
different lengths occupied by these equal volumes correspond to equal capa- 
cities of the tube. If the mercury in the successive parts of the tube ad, be, 
ed, de, &c. take up the lengths measured on the graduation L L! L'' L’", and 
_ the short parts of the tube ad, bc, ed, &c. be considered uniform in calibre, 
_ We obtain respective values of the divisional marks between ab, be, &c. with 
respect to the volumes corresponding to them expressed by the unity cor- 
responding to the length L, when L! L" L!", &c. are divided by L. On add- 
L2 
le oe ee 
