1 lb. of crude CaCl, in a pint of water, and, when the solution has cooled, diluting it to exactly the right gravity.

The coal is crushed, avoiding the production of dust as far as possible, until it will all pass through the half-inch sieve. The sample is then thoroughly mixed and a weighed amount of it (from 3 to 5 kg.) sifted over the 12- to 1200-in. sieves. This divides it up into sizes 2 to 3%, 3% to 14, 14 to 1%, % to 16, 16 to 132, 132 to 164, etc. The quantity in each size is then weighed and expressed in percentages of the whole. The various sizes are now treated separately in the chloride of calcium solution as follows: A glass tank holding about a gallon is nearly filled with the solution, the coal put into it a little at a time and well stirred to wet it thoroughly and detach all air bubbles. The coal rises to the surface, while the slate and pyrite settle to the bottom. Enough of the coal should be put in at a time to make a layer about an inch thick when it rises to the top. This is now skimmed out with a little dish or a dipper and dropped into a large funnel, the neck of which is closed with a wad of glass wool. More coal is now added to the solution and the operation continued until all of the given size is thus separated. The coal in the funnel is then thoroughly cleaned from chloride of calcium by repeatedly pouring water over it, allowed to drain, then spread out on paper, air dried and weighed. The chloride of calcium solution is then carefully poured off from the heavy material in the tank, which is then washed out, filtered dried and weighed. Each of the above products should then be analyzed for ash and sulfur. As a check, the analysis of the original coal should be computed from the analysis of these products and should agree very closely with that of the original sample.

A little practice is necessary in order to manipulate the separation properly. From a comparison of the results shown on the various sizes, it may be desirable to repeat the experiment, crushing the whole of the coal 14 in. or even tog in. in order to separate very finely disseminated pyrite and secure a coal low in sulfur. But as the loss in washing falls almost wholly on the finer coal, the increase of the proportion passing the smallest sieve must be carefully noted.

When many tests have to be made it is well to use the "Delatester" made by the Tyler Co.

REFERENCES: See DROWN, Trans. Am. Inst. Min. Eng., 13, 341. STOEK, J. Soc. Chem. Ind. (1897), 304.

CHAPTER XLI

THE ANALYSIS OF GASES

The analysis of gases is comparatively simple in theory, but no branch of analytical chemistry requires more skill and experience for accurate results. Very careful attention to detail of manipulation, both in sampling and in analysis, is necessary if the results are to be really dependable.

The methods of gas analysis may be classified as: First, those based upon volume changes; second, titrimetric; third, gravimetric. Most determinations belong to the first class, that is, they are based upon measurement of the decrease in volume of a gas mixture when some constituent is removed by chemical or physical methods.

The gas mixtures which the metallurgical chemist is likely to be called upon to analyze are tabulated below with their approximate compositions. Of course, these compositions are merely typical.

Besides the gas constituents given in the above table, the gas chemist is required frequently to determine the amounts of H2O, H2S, CS2, "light oils" (C6H6, C7Hs, CsH10), naphthalene (C1oHs), cyanogen (CN), SO2, NH3, oxides of nitrogen and, at smelters, the constituents of fume in gas, such as As2O3, PbO, ZnO, etc. Perhaps the most difficult determination of all is that of the dust in blast-furnace gas.

Mine air, which requires many analyses, is simply air in which part of the oxygen has been absorbed by coal and to which has been added

various small amounts of CH4, CO2 and, in case of mine fires, CO and other hydrocarbons which may be present.

Sampling. The sampling of gases as they are found in the mains, gas producers and furnaces about metallurgical plants is sometimes a very simple matter and sometimes very difficult. For instance, the sampling of gas in a gas main, a considerable distance from the source of the gas, may be merely a matter of taking gas out of the gas-cock, but the sampling of gas in the hearth of a blast-furnace or at the exit of an openhearth steel furnace or even in the flues leading from steam boilers is sometimes very difficult, that is, it is very difficult to get an accurate sample. The diversity of conditions under which it may be necessary to take gas samples makes it impossible to give directions here other than sufficient to illustrate the general principles.

In the sampling of gases two kinds of samples are recognized-the accumulative or "long-time" sample and the control sample. The accumulative sample is one which is taken continuously through the entire period of an operation running from one-half of an hour to 24 hours. For instance, in operating a gas retort, the composition of the gas derived from the coal in the retort varies continually from the beginning to the end of the carbonization, both in quantity and composition. A gas sample, truly to represent the average composition of the gas given off during the entire period, would have to be taken continually at a constantly varying rate bearing constant ratio to the rate of gas making. This is best done by a proportioning meter.

Control samples are those taken in a short period of time and which represent the gas merely for that short period of time, sometimes only a few seconds. All reports should, of course, state the way the sample was taken.

Sampling Flue Gas.-The gas should be drawn from the flue by a pipe that crosses it at right angles and extends to within 6 in. of the further wall. The end of the tube should be closed with a cap, and the gas should be drawn into it through a number of holes about 16 in. in diameter, drilled along the side of the tube at regular intervals not greater than 6 in. The nearest holes should not be less than 6 in. from the side of the flue. The diameter of the tube should be at least 12 times the diameter of the holes in the side. This will insure a uniform sampling across the flue.

Professor Lord carefully tested this point by inserting such a tube in air and gas for different portions of its length and analyzing the gas drawn from the tube. The composition of the issuing mixture was always proportional to the number of holes in the gas and in the air, air and gas being under atmospheric pressure.

The sampling tube can be made of iron, if the temperature of the flue does not exceed 340°C., as at that temperature iron, even if rusted or covered with soot, is without action on flue gas, neither the CO nor the CO2 being affected. This point was tested by the writer by passing

H

E

flue gas containing CO through a glass tube filled with iron tacks, also with rusted tacks and with soot and tacks. The tube was immersed in a bath of melted lead and the temperature of the lead measured by a nitrogen-filled hightemperature thermometer. If soot and rust are present on the iron, action begins about 340°C., and is rapid at 400°C., CO being oxidized by the Fe2O3 and oxygen consumed by the soot, forming CO2. With clean iron, however, there is practically no action at 400°C. As 340°C. is above the temperature at which most flue or blastfurnace gas is drawn off, the use of iron tubes is generally permissible. For flue temperature higher than 340°C., water-cooled tubes must be used for withdrawing the samples. For rapidly withdrawing a single sample from the interior of a furnace, an iron tube, open at the end and wrapped with 14 in. of sheet asbestos tied on with wire, can be used. The asbestos cover is well soaked with water and the tube run into the furnace and the sample drawn. A tube so protected can remain in a white-hot furnace for two or three minutes without the asbestos drying or the tube heating beyond a safe point. A silica tube is convenient for sampling gases from a very hot place, such as at the ports of an openhearth furnace. Where there is a strong draft, as in a chimney flue, it is important that the opening by which the tube passes through the wall be well plastered up with clay, or air may be drawn in and reach the nearer holes and affect the sample. From the end of the sampling tube the gas is drawn continuously by a water or steam aspirator. If the gas sample is to be kept any time before analysis, it must be borne in mind that gases containing CO2 cannot be preserved over water, as the CO2 is rapidly absorbed. If confined over water, the water should be covered

FIG. 20.

with paraffin shavings. The writer has found that if the water be well covered with paraffin shavings a sample of gas with 10 per cent CO2 will not lose 0.10 per cent CO2 in three hours.

In most cases it is possible and far preferable to make the analysis at the furnace, especially where a series of analyses is required. Where this is not possible, as, for instance, on a locomotive engine test, the apparatus shown in Fig. 20 will be found very effective and convenient for drawing a series of samples at short intervals.

The sample tube F, ƒ has a capacity of about 250 c.c.; the ends are closed with rubber tubes stopped with short glass rods J. A number of these sample tubes are provided; they are kept in a rack in a box, and are filled with water before starting out. In the apparatus itself, A is a bottle of about a liter capacity, containing absorbent cotton to filter the gas. C is a small bottle containing a little mercury; it serves as a trap to prevent reversal of the gas current. The gas enters through E, and is drawn out through D by an aspirator at the rate of about 150 c.c. a minute.

Thus the bottle A will always contain a gas representing the average of several minutes. H is a "pressure bottle," connected as shown.

In taking the sample, the apparatus is set up as shown; the gas is supposed to be flowing freely through the bottles A and D. A pinchcock (not shown) on the rubber connection between the Y-tube and the sample tube is opened and gas drawn in by lowering the pressure bottle H until the gas fills the tube and also the lower tube G. The cock is then closed and the pressure bottle lifted, so as to put the confined gas under a little pressure; but the lower tube G must contain gas enough to prevent any water getting into the sample tube. The rubber connection with F is now pinched tight and the tube disconnected. It is then closed by inserting the glass rod into the rubber. The lower end is closed in the same way. The rubber tubes and stoppers on the sampling tube can be wired if necessary to keep them tight. As the gas is under a little pressure, any leakage during the disconnection will be out from and not into the gas.

The tube is now replaced by a second one and the apparatus is ready for drawing a new sample.

Fig. 20a shows a convenient arrangement for drawing a single sample rapidly, as, for example, through an asbestos-covered pipe such as was described.

The bottle C contains a little mercury and serves as a trap. The bottles A and B contain brine covered with paraffin shavings P. The gas delivery tube is connected with the tube F. B is filled with water by raising A, the air escaping through the mercury in C. Now A is lowered