which gives as the weight of gas per lb. of pig yielded

Weight of dry gases 8b571.

and Mr. Bell's complete analysis gave

1lb .002

2lb .591

4lb.893, a difference of 0.085, or less than 2%

8lb .486.

Let us now take, as a second example, the blast furnace of 1866 at Clarence Works, height 80 feet, capacity 11,500 cubic feet. The consumption per lb. of pig is 1.125 lb. of coke 1.020 of pure carbon, and 0.683 of flux and limestone.

which gives the following as the quantities of gas per lb. of

N. 3.965, a difference of 0·004 or 0·1 %,

we see thus that the agreement of the formulas is as exact as possible, and indeed, by pure chance, greater than we had

any right to expect; for the method of analysis, or rather the method of taking the specimens of gas for analysis, adopted by Mr. Bell, does not admit of perfect exactness.

Although the results given by Mr. Lowthian Bell are the means of several specimens, it cannot be admitted that specimens taken almost instantaneously really give the true composition of the gas of a blast furnace. Ebelmen himself said, in speaking of the analysis of the gases at Seraing, "We must conclude from this that the analysis of the escaping gas of the furnace does not represent the mean composition of the gaseous current.”* And yet Ebelmen looked only to taking the means of the different gaseous currents at a given instant, whilst in order to have a clear idea of the working of a blast furnace, not only must we have the exact mean of all the currents escaping at a given instant, but of the true mean of a period of several hours. This important point has not been realized by any of the apparatus hitherto employed, and in taking specimens the following system is therefore proposed.

§ 9. A method of taking specimens securing a mean of all the gases during several hours.

This system is borrowed from the researches of MM. Scheurer-Kestner and Ch. Meunier on the products of combustion of coal.*

In order to obtain an exact mean a certain volume of the total gaseous current which passes from the top of the furnace to the boilers or heating stoves, etc., is drawn off continuously during several hours, and as it would be impracticable

* Annales des Mines, t. xix. 4e série, p. 127; also Note I. Appendix. † Bulletin de la Société Industrielle de Mulhouse, 1868.

to collect the totality of the gases thus drawn off, a certain fraction is withdrawn in the same manner and continuously into a Mariotte's jar filled with mercury, containing about 3 litres. The general arrangement of the apparatus is as follows:

1 50

Into the main pipe, which carries off the gases, there is passed a copper tube m n of 1 centimetre in diameter to 11 centimetres (a half-inch tube), fig. 13, of a length about double the diameter of the main pipe in question. The part inside the main pipe has a slit throughout that length, as at p. q, about a millimetre wide ( inch). This allows of the gases being drawn uniformly from every part of the current. The part of the copper tube outside the main pipe passes through a refrigerator on Liebig's system. Lastly, the extremity of the tube communicates by means of an Indiarubber junction with the leaden pipe which serves as exhaust. This is composed of a kind of trompe a b, provided with a cock, which allows of the current of water being regulated, and with a branch c d of greater or less length, according to the locality. This latter is soldered to the vertical tube a b, a little below the water-cock, and is also provided with a cock, which, working in concert with the other, serves to regulate the flow of gas. It is the end of this branch c d which is united to the copper tube m n by a joint-piece of India-rubber. Lastly, the lower end of the trompe opens into a cistern of water, which would in fact allow of measuring the gas drawn off by receiving it for a certain number of minutes under a glass receiver.

It might happen in many cases where the pressure of the gases in the main tube is strong that the trompe could be dispensed with, and the gas be taken off at once-into the

But in

It is a

lower cistern without using the current of water. any case it is more prudent to provide the trompe. simple means of regulating the flow of the gases. It may be made to come at the rate of three or four litres per minute.

Now, to collect a certain fraction of this gas-two or three litres in the course of several hours-it is only necessary to use the Mariotte vase above mentioned. For this purpose the tube m n is provided with a small tube h not far from its outer extremity. A bit of India-rubber tubing makes the communication with the straight tube of the Mariotte vase. The flow of mercury is regulated by a bent tube with cock, which may be raised or lowered at pleasure. And then a second upper tube of the Mariotte vase has also a tube with a cock g, which is only used when the gas is drawn off for analysis. When the Mariotte vase has to be filled, the mercury must be let in by the vertical tube till it is full, and overflows by the cock g to expel all air. This cock is then closed, and the mercury rises to the top of the straight tube. Then by means of the India-rubber the connection between this straight tube and the tube h of the copper tube m n is made. If the gas does not come off spontaneously by the India-rubber, a slight aspiration must be applied so as to expel all air it may contain, and then make the joint with the straight tube of the Mariotte vase.

What has been said above is sufficient to explain the working of the apparatus. It appears that by a double system of aspiration specimens representing as exactly as possible the average of the gases flowing from the furnace during several hours may be taken for analysis. It will of course be well, should the tension at the moment of charging be very different from the general tension, to desist from