the gain obtained by reducing the loss up the stack. Curves Nos. 3 and 4 support this explanation.

Curve No. 3 shows that the combustion-chamber temperature rises in direct proportion with the per cent of black smoke. Rise in combustion-chamber temperature always indicates a decrease in air supply.

Curve No. 4 shows a gradual increase in CO in the flue gases. The unaccounted-for loss, represented by curve No. 2, also rises as the smoke increases. It is reasonable to say that the greater part of the increase in the unaccounted-for loss is due to incomplete combustion of hydrocarbon gases and to the escape of solid particles of carbonforming smoke.

Curve No. 5 shows but a small increase in oxygen in coal, and therefore does not indicate that oxygen in coal is a direct cause of smoke. Curves Nos. 6 and 7 indicate that, excepting the first point at the extreme left, good or bad coals have not been prevalent in any one group.

The direct cause of the smoke seems to be shown in curve No. 3, which indicates increased rate of combustion and decreased supply of air.

The table which follows indicates that eastern coals are about as apt to smoke as western. It should be stated here that if these coals had been handled as suggested below most of them would probably have burned entirely without smoke, and the remainder with less smoke than they produced as actually handled.

Average per cent of black smoke produced by burning coals from certain localities under a Heine boiler.

The following table includes coals from many localities; some of them are usually called "smoky coals." It is quite probable that most other coals which produced smoke could have been burned smokelessly if the boiler division had not worked for the greater part of 1905 under the idea that very high temperatures are needed to get good over-all efficiencies. As shown elsewhere, extremely high furnace temperatures and a high degree of completeness of combustion in the furnace are incompatible. More air should usually have

been admitted through the smoke-preventer openings to the space over the fire; this precaution might have prevented almost wholly in most cases the formation of the smoke without affecting the overall efficiency of the outfit.

Coals burned under Heine boiler without producing smoke.

"VOLATILE MATTER," "FIXED CARBON," "WATER OF COMPOSITION,"

"COMBUSTIBLE."

"a AND

There still lingers a trace of an old idea that coal consists of particles of carbon cemented together with a sort of natural bitumen. The present tendency is to regard bituminous coal as a mixture of dozens, or even hundreds, of organic compounds, most of them derived from cellulose and many of them containing nitrogen and sulphur organically combined. It is true that for most kinds of coal group formulas can be devised which represent very closely the chemical compositions of the respective groups as shown by ultimate analyses; but it must be distinctly borne in mind that the devisers of these formulas do not mean that any one coal under consideration consists of one compound expressed by the formula proposed. Such formulas are meant as collective only; in exactly the same way as physiologists speak of the normal or average man, whose height, weight, etc., bear certain accurate relations to each other, although there is not in the world any one man who fits the description exactly.

When coal is heated, some or most of its constituent compounds are always broken down more or less, but are broken down differently, according to the particulars of the distillation method. That por

a For definition of "water of composition," see the glossary, p. 183.

tion which is distilled off from coal when it is heated under certain "standard" conditions is called "volatile matter." But inasmuch as the "standard" conditions are seldom exactly alike, and inasmuch as "volatile matter" is the product of a change, it is a mistake to say that coal contains any stated percentage of "volatile matter," or even that it contains "volatile matter" at all. The terms "fixed carbon," "volatile carbon," and "water of composition" are equally misleading. Coals do not literally contain these constituents in any more exact sense than cane sugar consists of "water" and "charcoal," although hydrogen and oxygen are contained in cane sugar in exactly such proportions as form water and leave carbon when the sugar is heated.

GRATE AND COMBUSTION SPACE.

Any furnace consists of the grate and the combustion space. The combustion space extends from the top of the fuel bed to the opening into the tube chamber, and its rear portion is termed the "combustion chamber." The function of the grate and fuel bed is to distill the "volatile matter" and partly to burn the "fixed carbon" of the coal. The function of the combustion space is to burn the "volatile matter." With coals high in "fixed carbon" combustion is nearly complete a short distance from the top of the fuel bed; with highly volatile coals the combustion is incomplete, even at the rear of the combustion chamber.

Samples of furnace gas collected at the top of the fuel bed are very commonly rich in combustible ingredients In the subjoined table is given the chemical composition of gas collected at the top of the fuel bed and also from the combustion chamber, determined from samples collected with the water-jacketed sampler. The samplers projected about 10 inches into the furnace, both being inserted through holes in the side wall, the first sampler resting on the surface of the fuel bed.

Analyses of samples of gas collected at top of fuel bed and rear of combustion chamber.

It is difficult to say whether the samples collected with the waterjacketed sampler are really representative samples of the furnace gases. Some constituents of the gas may decompose and reunite in different ways when suddenly cooled by the water-cooled surface. Chemical analyses of gas collected at the end of the combustion chamber seldom show much CO or any H, and CH.

Samples collected at the base of the stack show dilution of 25 to 30 per cent. The following table gives the chemical analyses of some gas samples collected simultaneously at the base of the stack and from the rear of the combustion chamber:

The above analyses give 17.5 and 23.2 pounds of gas, respectively, per pound of carbon.

If the flue-gas analyses determine the control of the fire, it is important that the samples analyzed be collected before the gas is diluted. As it is almost impossible to have a perfectly air-tight boiler setting, it is perhaps best to take the sample from the combustion chamber.

VELOCITY OF COMBUSTION AT VARIOUS POINTS ALONG THE FLAME.

The chemical law of mass action for two reacting substances states that, other things being equal, the number of new molecules of resulting compound (in the present case CO2) formed per second is proportional to the product of the masses of the reacting substances present per unit of volume (of gases in the present case), these masses being expressed in terms of gram molecules. (The gram molecule of a substance is the weight of the substance, in grams, numerically equal to its molecular weight.) There are two atoms in an ordinary molecule of oxygen gas, each weighing sixteen times as much as an atom of hydrogen. Therefore the molecular weight of gaseous oxygen is 32. The molecular weight of CO is 12+16=28.

Actual volumetric analyses of the simultaneous compositions of gas samples collected from two points along the gas stream of combustion are given on the following page.

The hydrocarbon and hydrogen values are omitted for this problem, although they sometimes amount to several per cent at the surface of the fire; in this calculation the effect of considering them would simply be to intensify the final conclusion. At the top of the fuel bed there was present by volume 1.70 per cent of oxygen and 7.10 per cent of CO. Multiplying each percentage by its specific gravity (so as to get numbers proportional to the masses present), and dividing in each case by the molecular weights above given, we have:

1.70X1.105÷32=0.0588
7.10X0.967÷28= .245

If we assume that the reaction between CO and O, is trimolecular, which is probably the case, because two molecules of CO react with one of O, then the rapidity with which CO burns is proportional to COXCOXO2, or COXO,, and the rapidity of combustion of CO in the two places is, at the top of the fuel bed, 0.2452×0.0588=0.00353; at the end of the combustion space, 0.006812×0.284 = 0.00001337. 0.00353 is considerably larger than 0.00001337 and therefore the rapidity of combustion is much greater at the top of the fuel bed than at the end of the combustion space (the difference in temperature being neglected).

The curve of chemical activity would drop off rapidly as the gases proceeded along their path in the combustion chamber, so that if the rates of combustion were plotted as ordinates along a base of travel of gases, the resulting curve would look much like the expansion curve of an engine-indicator diagram.

The practical value of such calculations is that they afford a mathematical verification of a fact frequently observed that mere length of combustion chamber counts for little compared with some device for thoroughly mixing the gases of the flame stream; one good mixing wall or baffle is probably worth many feet of undisturbed flow.

The possibility that many of the molecules of oxygen gas are dissociated at high temperatures into their component atoms does not affect the above calculation, because there would then be more atoms of oxygen looking for molecules of CO, per volume of gases, in just the proportion that the total volume occupied by a given mass of gases would be increased by the dissociation.