Coal gas, being much lighter than air, flows with greatest velocity in the upper floors of houses; hence the supply pipe may diminish in size as it rises, say from 1 1/4 in. at the basement to 3/4 in. on the 3rd floor. At a point near the commencement of the supply pipe it should be provided with a " siphon," which is simply a short length of pipe joined at right angles in a perpendicular position and closed at the lower end by a plug screwed in. As all gas-tubes should be fixed with a small rise, this siphon will collect the condensed liquids, which may be drawn off occasionally by unscrewing the plug end. When the lights flicker, it shows there is water in the pipes: the siphon prevents this.

The number of gas burners requisite for lighting a church or other large building may be computed thus. Take the area of the floor and divide this by 40, will give the number of fish-tail burners to be distributed according to circumstances. Example : a church 120 ft. long by 60 ft. wide, contains 7200 ft. area; divided by 40, gives 180 burners required for the same.

Burning gas without a ventilator or pipe to carry off the effluvia, is as barbarous as making a fire in a room without a chimney to carry oft* the smoke. If a pipe of 2 in. diameter were fixed between the joists, with a funnel elbow over the gaselier, and the other end carried into the chimney, it would be a general ventilator, Of course, an open ornamental rosette covers the mouth of the tube; or an Arnott valve ventilator over the mantelpiece would answer the same purpose.

In turning off the gas-lights at night, it is usual, first, to turn off all the lights, except one, and then turn off the meter main cock, and allow the one light to burn itself out, and then turn it off. The evil of this system is this, - by allowing the one light to burn itself out, you exhaust the pipes and make a vacuum, and of course the atmospheric air will rush in. The proper way is to turn off all lights first, and finally the meter, thus leaving the pipes full of gas and ready for re-lighting.

These few remarks have been derived from Eldridge's 'Gas-Fitter's Guide,' an eminently useful and practical handbook.

It was formerly the practice to make all gas burners of metal; the openings, whether slits or holes, from which the gas issued to be burned being small, in order to check the rate of flow. This was an error, for heat and light go together, and the metal, being a good conductor of heat, kept the lower part of the flume cold. The part of burners actually in contact with the flame is now invariably of some non-conducting material, such as steatite; and the effect of this simple improvement is most noteworthy. Bad burners show a great proportion of blue at the lower part of the flame, and the upper or luminous portion is small and irregular in shape, and dull in colour. These effects are due to gas issuing at too great velocity from small holes in burners, as well as to improper material in the latter. The illuminating power of coal gas depends upon the incandescence, at the greatest possible heat, of infinitesimal particles of carbon which it contains, invisible until heated. In the lower, or blue portion of the flame, the heat is not sufficient to render these particles incandescent; and it is necessary that this effect should be secured at the nearest point to the burner.

Unless this is done, the light is not only lessened, but the unconsumed carbon passes off and is deposited as soot on ceilings and furniture. Blackened ceilings are a measure of the badness of the burners. It will now be seen why a material which cools the flame should not be used for a burner, for the hotter the flame, the more perfect is the incandescence of the carbon for which in reality the consumer pays, and the less danger there is of blackened ceilings. But in addition to the better material, the construction of even the cheapest modern burners is very greatly improved; although even a good burner may be subjected to such conditions - e. g. allowing gas to be driven through it at a high velocity, a condition usually accompanied by a hissing or roaring sound - as to give a bad result. The capacity of burners should moreover bear a reasonable proportion to the quality of the gas for which they are required to be used. Thus with rich Scotch gas, burners with very small holes, consuming only about 1 1/2 cub. ft. hourly, are sometimes adopted for economical reasons. Occasionally these burners find their way South, but their use for the ordinary qualities of English gas is the worst possible economy.

It is difficult to lay down hard and fast rules for the sizes of burners, the purposes for which gaslight is required being so various. For an ordinary apartment, however, wherein distributed lights are adopted, 5-ft. burners with 14 or 15 candle gas, 4-ft. burners with 16 or 17 candle gas, 3 or 3 1/2 ft. burners with 18 or 20 candle gas, and 2 1/2-ft. burners with richer gas will be found to give satisfactory results. It may be remarked that these figures apply to burners regulated in some way to the given rates of consumption, and not to those merely reputed to be of the stated sizes. Various means are adopted for cheeking the flow of gas, not at the point of ignition, but at some prior point of its course; because it has been found that the slower the rate of flow at the commencement of combustion, the better the result obtained.

Clustering of gas lights is bad. All parts of a room should be as nearly as possible equally lighted, the only noteworthy exception to this rule being in the case of a dining-room, where concentration of light upon the table is not only permissible but is even demanded. Hence in most cases wall brackets give the best effect, and such masses of light as are afforded by pendants of many arms are to be avoided, or are only required in very large rooms where portions of the floor area would otherwise be insufficiently lighted. When it is desired to light a drawing-room with wax candles - than which nothing is more beautiful - they are distributed wherever support can be found for them. As every gas flame may be considered equal to 12 or 15 candles, with all their wicks together, the inadvisability of further concentration is evident. In fact, gas is if anything too brilliant for living-rooms, and if it were always properly distributed, man dimly-lighted apartment would be perfectly illumined with the same number of burners which, when massed, appear insufficient. Where concentrated ceiling lights are needed for dining-rooms, many-armed pendants are seldom satisfactory, owing to the shadows which most of them cast.

In these cases a single powerful argand light in a suitable reflecting pendant, or a cluster of flat flames similarly provided, will give a better result than the usual branched chandelier, and with a material saving in gas. For it is a curious and valuable property of gas, that large burners can be rendered much more economical in proportion than smaller ones. Thus, if the 4 burners of a branched chandelier give altogether the light of (say) 50 candles, the same illuminating power may be obtained from a greatly reduced quantity of gas when concentrated in a single burner of the most improved kind.