26 MR SWAN ON THE CONSTITUTION OF FLAME. 
OBSERVATIONS OF THE BREADTH OF THE SPECTRUM OF A COAL-GAS FLAME. 




Line of 
Spectrum First Set of Observations. | Second Set of Observations. | Third Set of Observations. 
Observed. 
157 125 103 
152 124 105 
G 153 123 | 102 


The numbers in the table are, in the first set of observations, the means of 
seven micrometer readings, and in the second and third sets, the means of twenty- 
four and eight readings respectively. 
It will be seen that so far from the spectrum increasing in breadth towards its 
most refracted extremity, the measured breadths at the line G in the blue rays, 
are actually in every case less than those at the line B in the red rays of the spec- 
trum.* 
Having found the flame of gas, notwithstanding every precaution, to be sub- 
ject to anomalous fluctuations, and being desirous, moreover, to repeat Professor 
Draper’s experiments as faithfully as possible, I next observed the flame of a spirit- 
lamp fed by alcohol. It will be remembered that the spectrum of the flame of 
alcohol is that which he describes to illustrate his views. I did not anticipate, 
however, that such a flame would be convenient to experiment upon, from the 
faintness of its light; and, in effect, [found it was impossible to illuminate the field. 
so as to be able to see the wires without rendering the violet end of the spectrum 
invisible. I therefore observed with faintly illuminated wires on a dark field; 
and, instead of making contacts with the wires at different points of the spectrum, 
I varied the mode of observation, by carefully adjusting the wires to the breadth 
of the spectrum at its red end, and then “sweeping” its whole length. The 
wires seemed to jit the spectrum equally well at all points, so that no variation 
in breadth could be detected. 
In order to avoid the difficulty of observing a faint spectrum, and at the same 
time to obtain a flame which would remain sensibly of the same size for a con- 
siderable time, I devised the following method. Volatile liquids, such as naphtha 
or turpentine, seemed preferable to the fixed oils, because, from their greater 
fluidity and freedom from viscid matter, they flow more freely through a cotton 
wick, and less rapidly impair its transmitting power; while from their volatility, 
and consequent rapid evaporation, the wick remains cool, and is scarcely at all 
burned. Both conditions are highly favourable to the maintenance of a constant 
* It may be proper, in order to explain the different values for the same line of the spectrum 
in the different sets of observations, to remark that these were made on different days, and on flames 
of different dimensions. 
