1884.] Spectroscopic Studies on Gaseous Explosions. 473 



put into the tube the continuous spectrum was very bright, but the 

 iron lines were still brighter. No line which could be identified as 

 due to magnesium was observed with certainty; there was only a 

 doubtful appearance of b. "With sodium, potassium, and barium car- 

 bonates, only the lines usually seen when salts of those metals are 

 introduced into a flame were noticed; but eye observations of this 

 kind are extremely trying, on account of the suddenness of the flash 

 and the shortness of its duration. Thallium gave the usual green line. 



Subsequently we had the interior of the tube bored out so as to 

 present a smooth bright surface of iron, and noted the iron lines 

 which were conspicuous in the flash. 



For the purpose of identification the pointer in the eye-piece was 

 first placed on one of the strong iron lines given by the electric 

 discharge between iron electrodes, and then, the discharge being* 

 stopped but the field sufficiently illuminated, the eye was fixed 

 steadily on the pointer while the gas in the tube was exploded. In 

 this way it was not difficult to see whether any given line was very 

 bright in the flash. The lines thus identified were those having the 

 wave-lengths about 5455, 5446, 5403, 5396, 5371, 5327, 5269 (E), 

 5167 (& 4 ). These lines were all many times observed in the way 

 described, and as a rule were always present in the flash. Lines with 

 wave-lengths about 5139 and 4352 were seen, and may possibly have 

 been due to iron, and several more lines were seen occasionally, but 

 were not so regularly seen that they could be well identified. The 

 lines \ 4923 and \ 4919 were specially looked for, but neither of them 

 could be seen. A group of blue lines were noticed, and were after- 

 wards identified by photography, a method much less trying than 

 observations by eye. To give intensity to the photographs ten or 

 twelve flashes were usually taken in succession without any shift of 

 the instrument, so as to accumulate their effects in one photograph. 

 For identification the spark between iron electrodes was also photo- 

 graphed, but with a shutter over the lower part of the slit, so that the 

 image of the spark should occupy only the upper part of the field. 



The following is the list of wave-lengths of the iron lines thus 

 photographed : — 



4414 7 4131-5 3920 



4404-2 4071 3902-5 



4382-8 4062-9 3898*4 



4325-2 4045 3885 



4307-2 4004-7 3877 -4 



4271-3 3967 3859-2 



4250-5 3929-7 3849 7 



4201-5 3927-2 3840-3 



4143-1 3922 3833 '6 



