of Edinburgh, Session 1870 - 71 . 467 
this inconvenience, which, with some slight recent improvements, 
I will now proceed to describe. My method was founded on a fact 
which I had observed many years ago, namely, that the mere pre- 
sence of a chemical substance in a flame frequently suffices to cause 
the appearance of its characteristic rays, and that it is not at all 
necessary that the substance should be consumed and dissipated. 
This dissipation is an accident, and if by any means it could be 
prevented, the flame would maintain its characters for a consider- 
able time. For instance, in Brewster’s Journal for 1826, vol. v. p. 
77, &c., I remarked that alcohol burnt in an open vessel, or in a 
lamp with a metallic wick, gives but little yellow monochromatic 
light, while if the wick be of cotton, it gives a considerable quan- 
tity, and that for an unlimited time. And I added that I had 
found other instances of a change of colour in flames, owing to the 
mere 'presence of a substance which suffers no diminution in conse- 
quence. Thus, a particle of muriate of lime on the wick of a spirit 
lamp will produce a quantity of red and green rays for a whole 
evening without being itself sensibly diminished. 
Mindful of these experiments of 1826, when a few years ago I 
wished to examine the spectra of thallium and other substances, I 
adopted the following plan : — A grain, or sometimes much less, of 
the substance was. placed in a piece of strong glass tube about one 
inch long. Short platina wires were inserted into the tube at each 
end, approaching each other within about half an inch. The ends 
are then sealed by a blow-pipe, leaving enough of the platina wire 
outside the tube to allow of its being soldered to a long copper wire. 
One of these copper wires (with the external portion of the pla- 
tinum wire soldered to it) was then coated with gutta percha for 
the space of three or four inches next the tube. To coat the other 
wire was found unnecessary. The mode of experimenting was as 
follows. The tube in a horizontal position, having the chemical 
substance nearly in its centre, was lowered into a glass of water 
about two or three inches below the surface. The two wires were 
then connected with a BuhmkorfFs coil, set in action by six of 
Grove’s cells. When the sparks were allowed to pass through the 
tube, they speedily ignited the substance, and caused it to give 
forth its characteristic spectrum. Even after the sparks have been 
passing for several minutes, the tube remains perfectly cold. This 
