proximity of a single luminary. If brackets are to be employed, let there be at least two in a room, and these disposed vis-a-vis, or as nearly so as possible.

Reflectors. The value of reflectors is not appreciated as it should be, and the reason is principally because few people, even those whose business is to make apparatus for artificial light and attend to the introduction of gas-fixtures, etc., are sufficiently acquainted with the laws that govern reflected light, and when so, they fail in the mechanical ability to properly arrange reflectors so as to obtain the proper effect. Reflectors should be made of a material that will not tarnish by the action of the atmosphere or the temperature they may be exposed to. A very slight film of dust, moisture, or smoke on a reflector will almost entirely destroy its value as a reflector. The surface of the reflector should be perfectly smooth, and free from scratches and abrasions. Hence it is apparent that metallic reflectors are not the best in that respect.

Glass reflectors are superior, inasmuch as they do not become tarnished, abraided, scratched, but their action is impaired if the glass is too thick, owing to the absorption of light. The late American invention of a mica reflector is advantageous on that account, because the plates or lamina are very thin. It has also the advantage of not being fragile or liable to fracture.

Reflectors are better placed overhead. A reflector which throws the light in a horizontal direction, unless neutralized by another opposite, will be very disagreeable, owing to the dazzling glare. As a rule, reflectors should be so placed that the reflective rays shall never reach the eye in a straight line. This will avoid the evil effects of glare. As a rule, all the direct rays of a lamp or burner thrown upward may be thrown downward by reflectors, producing a great economy of light, and an effectiveness of illumination very pleasant and satisfactory.

HEATING BUILDINGS BY GAS.

In the United States this art has lately acquired a new impulse, owing to the late discoveries and improvements in the art of manufacturing hydrogen and oxide of carbon gases, at a very trifling cost, compared to the cost of ordinary coal gas. These gases are especially adapted to heating purposes instead of solid fuel and for use in gas engines as a substitute for steam-power, and also for illuminating purposes when carburetted or charged with vapors of hydro-carbons.

The hydrogen and oxide of carbon gases are produced by the American process, under the Gwynne-Harris patent, which consists in decomposing superheated steam, by means of incandescent anthracite coal, in a peculiar manner, and in a simple yet novel apparatus.

Great improvements have also been made in the stoves and other apparatuses for heating and cooking, which overcome

most if not all the difficulties heretofore experienced in this department.

The use of gas as fuel has been tried to a considerable extent in France and other countries, but the progress has neither been rapid nor very satisfactory; one reason of this lies, perhaps, in the imperfection of the modes of combustion, although something has been done of late to remedy this; another is the natural hesitation of the directors of gas works to keep pressure of their gasometers all day for a small supply.

Still enough has been done to supply a certain amount of information on the economical part of the question, both as regards gas-cooking apparatus and stoves for churches and other large buildings. The average consumption of the cooking-stoves in use in France, which consume a mixture of gas and air, is found to be as follows: For a large fire, 260 litres per hour; for a moderate fire, 140 litres per hour; for a small fire, 50 litres per hour. When the stove is used, what the French call pot-au-feu, it is found that it is sufficient to keep up a large fire for about 20 minutes only, after which the gas may be turned down, and the cooking completed with a very small fire. Taking the average duration of this kind of cooking at 4 hours, and the cost of gas at 30c. per cubic metre, - the present price in Paris,—the consumption amounts to 1,050.20 litres, the expense of which is 31.20c., or little more than 14d. The cleanliness and handiness of gas as fuel, and the great economy arising from its instantaneous lighting and extinction, give it, in the hands of careful persons, a great advantage over charcoal, with few of its inconveniences, one of which is the impossibility of using it for broiling with a special arrangement, as the smallest quantity of fat falling upon heated charcoal fills the house with stifling fumes.

In a coal-using country, however, like England, the use of gas for the heating of apartments, and especially large buildings like churches, is of more importance than its application to cooking; and considerable improvement has been made of late in France in apparatus for the warming of ordinary rooms, to which we shall shortly have to refer more particularly.

The most important results yet produced refer to the heating of churches, which has been essayed on a large scale at Berlin. The method generally adopted is that of placing a horizontal gaspipe with 3 jets within a stove made of sheet iron, and over the gasjets a piece of brass wirework, of which the openings are not more than one-twenty-fifth of an inch in diameter. The cathedral at Berlin has a cubical contents of about 13,300 metres, and it is heated by means of 8 of these stoves, each of which has 22 of these brass gratings, 11 inches in length by 12 inches in width, making in all about half an inch square of grating for each cubic metre to be warmed. The consumption of gas in raising the air within the edifice to the required temperature- an operation which takes 3 hours-is 83,400 litres, or 4.82 litres per cubic metre; to maintain the same heat afterwards requires only seven-tenths of a litre of gas per cubic metre.

The parish church of Berlin, whose cubic contents is 13,800

metres, is heated by 4 stoves, each having 15 brass gratings, each rather more than 12 inches long by 14 inches wide, or little more than one-fifth of an inch of grating per cubic metre to be warmed. The annual consumption of gas in the cathedral above mentioned is 2.210 cubic metres, costing 20 pounds; this consumption is equal to 552 metres per stove, and 300 litres per four-fifths of an inch square of grating. The consumption in Parisian churches warmed by gas is found to agree very closely with that of the cathedral of Berlin, but other cases give different results.

The church of St. Philippe at Berlin has a contents of 2,780 metres, and is heated by two stoves 1.40m. high, and 1.10m. long, and 65 centimetres in width, each having 7 brass gratings 16 inches by two inches, equal to two-fifths of an inch square per cubic metre of the contents of the church. The annual consumption in this church is 1,485 cubic metres, or at the rate of 410 litres of gas per cubic metre of contents. But this church is only warmed 3 times a week.

The church of St. Catherine at Hamburg is heated by 8 gas stoves, each having 32 brass gratings, 12 inches long by rather more than 1 inches wide; the cubic contents of the edifice is 33,900 metres. The heating takes 34 hours, and consumes 220 metres of gas, costing about 27s. 6d., so that 3 litres of gas are required in this church per cubic metre of capacity; the temperature is kept up subsequently by the consumption of three-fourths of a litre per cubic metre and per hour.

In the churches of St. Mary and Nicholas in Berlin, and in Paris also, a kind of large rose burner has been substituted for the brass grating; these are known in France as mushroom burners (champigons). The result with these burners in the first of the above-named churches is as follows: The cubical contents of the building is 15,450 metres, and the consumption of gas in 4 hours is 150 cubic metres, costing about 35s., and as it is heated by 10 stoves, each having 3 of these rose burners, the consumption per hour is 14 cubic metres of gas per burner, and nearly 2 metres for each metre of the contents of the church. In this case only we have the effect as shown by the thermometer, which is to raise the temperature of the church from one degree below zero to 5o above, or from below 30° to 40° F.

In heating churches and large buildings the economy of gas exhibits itself quite differently as compared with its application to cooking; in the former case, the more continuous the operation the less the relative cost, whereas in the latter the more frequent the interruptions the greater the economy. The objection to gas on account of its vitiation of the atmosphere of a building is one which neither the wire grating nor the mushroom burner has yet obviated.

MIXTURE OF GAS AND AIR.

Professors Silliman and Wurtz conclude that,

1st. For any quantity of air, less than 5 per cent., mixed with gas, the loss in candle-power due to the addition of each one per

cent., is a little over six-tenths of a candle (.611 exactly); above that quantity the ratio of loss falls to one-half a candle power for each additional one per cent. up to about 12 per cent. of air; above which, up to 25 per cent., the loss in illuminating power is as shown by column 12 of the table, nearly four-tenths of a candle for each one per cent. of air added to the gas. In column 11 of Table 1, the ratio of loss in candle power is given in percentages for the several volumes, while in column 10 the destructive effect of air upon the illuminating power is most conspicuously exhibited, 12 per cent. of air destroying over 40 per cent. of the illuminating power. In the diagram this loss of power is represented by the numerals in the right-hand column, which are inverse to those in column 10, and stand with the maximum intensity 100.

2d. With less than one-fourth of atmospheric air, not quite 15 per cent. of the total illuminating power remains; and with between 30 per cent. and 40 per cent. of air it totally disappears.

In large gas works the liability to contamination by air accidentally introduced by various causes diminishes in proportion to the total make of gas, and an amount of air which, when diffused in a very large volume of gas, becomes insignificant, if confined to 10,000 or 15,000 feet daily product, will become a most serious injury to its illuminating power. This cause of deterioration in gas has been overlooked almost entirely by gas engineers; but in small gas works it deserves special attention, and we have no doubt that the low illuminating power too often obtained in such works is largely due to this cause.

Results of Messrs. Audouin and Berard.· We have already alluded to the results obtained by Messrs. A. and B., which form part of an important memoir published in 1860, under authority of the French Government, " upon the various burners employed in gas-lighting and researches on the best conditions for the combustion of gas." Their table shows "a considerably higher ratio of loss than we have obtained, being rather more than 6 per cent. loss for each one per cent. of air added to the gas, reaching a total loss of 80 per cent. with 15 per cent. of air added; while we obtain 57.53 per cent. loss with 16 per cent; and 93 per cent. loss with 20 per cent. air, while with the latter volume of air added we get 72.90 per cent. loss. These differences may be accounted for by the French trials being made upon a gas of not more than 12 candle-power, our trials being made on a gas averaging nearly 15 candles; also, by the fact that in the French experiments the gas was burned from a batswing burner, ours from a standard Argard.

It appears that the introduction of 6 to 7 per 100 of air suffices to diminish the intensity by one-half, and a mixture of 20 of air with 80 of gas leaves almost no illumination. Unfortunately Messrs. A. and B. do not record the actual illuminating power of their standard gas, which, however, we are led to believe cannot be more than 12 candles of the English and American standard. For a fuller discussion of this subject the reader is referred to the memoir of Prof. Silliman and Wurtz.

ON THE RELATION BETWEEN THE INTENSITY OF LIGHT PRODUCED FROM THE COMBUSTION OF ILLUMINATING GAS AND THE VOLUME OF GAS CONSUMED.

In photometric observations made to determine the illuminating power or intensity of street gas, it is the practice of observers to compute their observations upon the assumed standard of 5 cubic feet of gas, consumed for one hour; and in the constantly occurring case of a variation from this standard, whether in the volume of the gas consumed or in the weight of spermaceti burned, the observed data are computed by the "rule of three," up or down, to the stated terms. The standard spermaceti candle is assumed to consume 120 grains of sperm in one hour, a rate which is rarely found exactly in actual experience.

For example, a given gas, too rich to burn in a standard argand burner at the rate of 34 cubic feet to the hour, with an observed effect of 20 candles' power. This result, previously corrected by the same rule for the sperm consumed, is then brought to the standard of 5 cubic feet by the ratio 3.5: 20 = 5: 28.57.

The candle-power of the gas is, therefore, stated as 28.57 candles, and the result has been universally accepted as a true expression of the intensity of the gas in question, or the relative value of the two consumptions.

In common with other observers, I have long suspected that this mode of computation was seriously in error, as an expression of the true intensity of illuminating flames, and that there were other conditions, besides the volume of gas or weight of sperm consumed, which must influence and greatly modify the results. As most of these conditions are considered somewhat at length in a paper on "Flame Temperatures," prepared chiefly from researches conducted by Professor Wurtz and myself, and presented at the Salem Meeting of the Association, they need not be discussed in this connection.

The results of many trials, made with the purpose of determining the value of these photometric rations, indicate clearly, that the true ratio of increase in intensity in illuminating flames is, within certain limits, expressed by the following theorem, namely:

:

The intensity of gas flames, that is, illuminating power, increases (within the ordinary limits of consumption) as the square of the volume of the gas consumed.

As the first experimental demonstration of this theorem was made by Mr. William Farmer, the photometric observer at the Manhattan Gas Co.'s works in New York, I propose to speak of it as "Farmer's theorem." I am also indebted to Mr. Farmer, and to Mr. Sabbaton, the well-known and courteous engineer of the Manhattan Gas Light Company, for the free use of their experimental data, and the permission to employ them here.

The fundamental importance of this new mode of computation will at once appear, if, assuming it for the sake of illustration to be true, we apply it to the case already given above, which then becomes