Heat may be lost from a flame in three several ways: by direct radiation and conduction into the surrounding air, among the products of combustion, and by conduction into the body of the burner. Loss of heat by radiation and conduction to the air will be the greater as the flame exposes a larger surface, and as a more rapid current of cold air is brought into proximity with the flame. Loss of heat by conduction, into the burner will be the greater as the material of which the burner is constructed is a better conductor of heat, and as the mass of material in that burner is larger. Loss of heat by passage into the combustion products will also be greater as these products are more voluminous; but the volume of true combustion products from any particular gas is a fixed quantity, and since these products must leave the flame at the temperature of that flame--where the highest temperature possible is requisite--it would seem that no control can be had over the quantity of heat so lost. However, although it is not possible in practice to supply a flame with too little air, lest some of its carbon should escape consumption and prove a nuisance, it is very easy without conspicuous inconvenience to supply it with too much; and if the flame is supplied with too much, there is an unnecessary volume of air passing through it to dilute the true combustion products, which air absorbs its own proper proportion of heat. It is only the oxygen of the air which a flame needs, and this oxygen is mixed with approximately four times its volume of nitrogen; if, then, only a small excess of oxygen (too little to be noticeable of itself) is admitted to a flame, it is yet harmful, because it brings with it four times its volume of nitrogen, which has to be raised to the same temperature as the oxygen. Moreover, the nitrogen and the excess of oxygen occupy much space in the flame, making it larger, and distributing that fixed quantity of heat which it is capable of generating over an unnecessarily large area. It is for this reason that any gas gives so much brighter a light when burnt in pure oxygen than in air, (1) because the flame is smaller and its heat more concentrated, and (2) because part of its heat is not being wasted in raising the temperature of a large mass of inert nitrogen. Thus, if the flame of a gas which naturally gives a luminous flame is supplied with an excess of air, its illuminating value diminishes; and this is true whether that excess is introduced at the base of the actual flame, or is added to the gas prior to ignition. In fact the method of adding some air to a naturally luminous gas before it arrives at its place of combustion is the principle of the Bunsen burner, used for incandescent lighting and for most forms of warming and cooking stoves. A well-made modern atmospheric burner, however, does not add an excess of air to the flame, as might appear from what has been said; such a burner only adds part of