Several theories have been brought forward to explain this phenomenon, but the true one is that the burner abstracts so much heat from the flame at that point that it is unable to burn there, and this can be proved by the fact that where a cold object touches the flame, a dividing space, similar to that noticed between flame and burner, will always be observed, and the colder the object and the more diluted the gas the greater is the observed space.  If a cold metal wire or rod is held in a non-luminous flame, it causes an extinction of the gas for some considerable space around itself; but as the temperature of the rod rises, this space becomes smaller and smaller until the rod is heated to redness, and then the flame comes in contact with the rod.

In the same way, if the burner from which the gas is issuing be heated to redness, the space between burner and flame disappears.  It has already been shown that cooling the flame by an inert diluent reduces the illuminating value, and finally renders it more luminous; and we are now in a position to discuss the points which should be aimed at in the construction of a good gas burner.

In the first place, a sensible diminution in light takes place when a metal burner is employed, and the larger the surface and thickness of the metal the worse will be its action on the illuminating power of the flame; but this cooling action is only influencing the bottom of the flame, so that with a small flame the total effect is very great, and with a very large flame almost nil.

The first point, therefore, to attend to is that the burner shall be made of a good non-conductor.  In the next place, the flow of the gas must be regulated to the burner, as, if you have a pressure higher than that for which the burner is constructed, you at once obtain a roaring flame and a loss of illuminating power, as the too rapid rush of gas from the burner causes a mingling of gas and air and a consequent cooling of the flame.  The tap also which regulates the flame is better at a distance from the burner than close to it, as any constriction near the burner causes eddies, which give an unsteady flame.

These general principles govern all burners, and we will now take the ordinary forms in detail.  In the ordinary flat flame burner, given a good non-conducting material, and a well regulated gas supply, little more can be done, while burning it in the ordinary way, to increase its luminosity; and it is the large surface of flame exposed to the cooling action of the air which causes this form of burner to give the lowest service of any per cubic foot of gas consumed.  Much is done, moreover, by faulty fittings and shades, to reduce the already poor light given out, because the light-yielding power of the flame largely depends upon its having a well rounded base and broad, luminous zone; and when a globe with a narrow opening is used with such a flame—­as is done in 99 out of 100 cases—­the updraught drags the flame out of shape, and seriously impairs its light-giving powers, a trouble which can be got over by having the globe with an opening at the bottom not less than 4 inches in diameter, and having small shoulders fixed to the burner, which draw out the flame and protect the base from the disturbing influence of draughts.