AT A: TRANSACTIONS OF SECTION B. 
region in which it is immersed. As ordinary flames consist of thin shells of 
burning gases, on either side of which there is a very rapid fall of temperature, it 
is necessary to use thin wires, and to dispose them so that there is no appreciable 
drain of heat from the junction. By using wires of different gauge for the 
couples it is possible by extrapolation to arrive at a temperature for a couple of 
infinitely small cross-section, and it is also possible to make a correction for the 
superior radiating power of the couple as compared with the flame-gases. Without 
this last correction a maximum temperatnre of 1770° was obtained for the Bunsen 
flame by Waggener in Germany, and 1780° by White and Traver in America. 
Correcting for radiation, Berkenbusch found 1880° as the maximum temperature. 
M. Féry, by an ingenious application of his beautiful optical pyrometer to a 
flame containing sodium, gives 1871° as the highest temperature of the flame of a 
Bunsen burner burning coal-gas. 
The consideration of flame-temperatures has become of increasing importance 
in the arts owing to the use of the Welsbach mantle as a means of deriving 
light from coal-gas, The great improvements which have been made in the 
efficiency of atmospheric burners depend primarily on the fact that the smaller the 
external surface we can give to a flame consuming gas at a fixed rate the higher 
must be the average temperature; and since the emission of light from a mantle is 
proportional to a high power of the absolute temperature, a small increase of 
temperature is of great effect on luminosity. 
The acetylene-oxygen flame in which a temperature of about 3500° prevails, 
not very different from that of the electric are, is the hottest of the hydrocarbon 
flames, and finds some important practical uses. 
I have already said something about the luminosity of flames so far as relates 
to the separation and glow of solid carbon. But there remains the more general 
question of the luminosity of flames containing nothing but gases. The older 
explanation of the emission of light from combining gases said no more than that 
the energy liberated during the reaction and appearing as heat raised the product 
to ineandescence—that is to say, so increased the velocity of its molecules and the 
violence of their collisions that vibrations were set up whose wave-lengths lay 
within the limits of visible radiation. This explanation has long been questioned, 
and there is now, I think, a very general agreement that it will not suffice. The 
average temperature, in fact, prevailing in a flame, if attained in the product of 
combustion by the supply of heat from outside, does not suffice to make that sub- 
stance luminous. We are therefore thrown back upon the conclusion that the 
generation of light in a flame is not a consequence, though it is an accompaniment, 
of the elevation of temperature. The question now is, Can we go any further ? 
To do this we are led to consider individual molecular transactions instead of 
statistical averages, and the view presents itself that the combining atoms may, 
in losing their chemical energy, form directly systems of independent vibration 
where the radiation is such as to fall within the limits of visibility. If we picture 
such vibrating systems momentarily formed, it is easy to see that by their collision 
one with another they may acquire in a secondary way increased translational 
motion, and so lead to a state of things where the greater part of their energy is 
degraded in the form of heat. The high temperature of a flame would then be a 
consequence rather than a cause of its light. 
This subject of the mechanism of luminosity, however, like so many others, has 
now become involved with the theory of electrons, and a chemist may be excused 
if he hesitates to pursue the subject further. Some years ago I called attention to 
the scantiness of our knowledge of the chemical changes that take place when 
metallic salts are used in flames for the production of spectra. Though there was 
general agreement that, for example, the yellow flame produced by common salt 
was due to the liberation and glow of metallic sodium, there was no agreement as 
to how the sodium was set free. 
Arrhenius, pursuing the analogy which exists between the laws governing 
matter in the gaseous state and in the state of dilute solution, had previously been 
led to the view that the electrical conductivity of flames containing salt-vapours 
was due to ionisation of the salt throughout the volume of the flame. It appeared 
