84 Professor Dewar [April 1, 



this way be separated by a distance of about |^ of a unit of Angstrom's 

 scale ; a quantity quite appreciable, though much less than the 

 distance between the components of tg, and about equal to the 

 distance of the components of the less refrangible of the pair of lines 

 E. We thought therefore that we might test theory by experiment. 



A preliminary question had, however, to be answered. What 

 lines could be seen in the flash of the exjDloding gases ? We were 

 pretty certain that the hydrogen lines could not be seen, but that 

 probably we might get sufficient dust of sodium compounds floating 

 in the gas to make the sodium lines visible. A preliminary observa- 

 tion was made on the flash of mixed hydrogen and oxygen in a 

 Cavendish's eudiometer, which showed not only the yellow sodium 

 lines, but the orange and green bands of lime and the indigo line of 

 calcium all very brightly, as well as other lines not identified. The 

 flash is very instantaneous, but nevertheless produces a strong im- 

 pression on the eye ; and by admitting the light of a flame into the 

 spectroscope at the same time as that of the flash, the identity of the 

 lines was established. That sodium should make itself seen was not 

 surprising, but that the spectrum of lime should also be so bright 

 had not been anticipated. At first we thought that some spray of the 

 water over which the gases were confined must have found its way 

 into the eudiometer ; but subsequent observations led us rather to 

 suppose that the lime was derived from the glass of the eudiometer. 

 The lime-spectrum made its appearance when the eudiometer was 

 quite clean and dry, and when the gases had been standing over water 

 for a long time. 



To obtain the high dispersion requisite, as already explained, we 

 made use of one of Eowland's magnificent gratings, with a ruled 

 surface of 3^ by 2i inches, and the lines 14,438 to the inch. One 

 telescope fitted with a collimating eye-piece served both as collimator 

 and observing-telescope ; and by this means we were able to use the 

 spectra of the third and fourth orders with good effect. 



Observations were made with this instrument on explosions in an 

 iron tube shown in section in Fig. 1, half an inch in diameter, fitted 

 at the end with a thick glass plate (a), held on by a screw-cap (c) and 

 made tight with leaden washers. Small lateral tubes {d, d), at right 

 angles to the main tube, were brazed into it near the two ends, for 

 the jDurpose of connecting it with the air-pump, admitting the gases, 

 and firing them. For this last purpose a platinum wire (h) fused into 

 glass was cemented into the small tube, so that an electric spark 

 could be passed from the wire to the side of the small tube when the 

 gases were to be exploded. 



To bring out the lithium lines, a small quantity of lithium 

 carbonate in fine powder was blown into the tube before the cap with 

 glass plate was screwed on. Powder was used because wo supposed 

 that it must be loose dust which would be lighted up by the explosion. 

 The lithium lines came out bright enough, and it was unnecessary to 

 put in any more lithium for any number of explosions. The tube 



