1881.] Investigations on the Spectrum of Magnesium. 201 



expansion. In the flame it is a broad band, and equally so in the arc 

 when magnesium is freshly introduced, but fines down to a narrow 

 line as the metal evaporates. 



Almost equal in persistence are the series of triplets. Only the 

 least refrangible pair of these triplets is seen in the flame, another 

 pair is seen in the spark, but the complete series is only seen in the 

 arc. We regard the triplets as a series of harmonics,, and to account 

 for the whole series being seen only in the arc we must look to some 

 other cause than the temperature. This will probably be found in the 

 greater mass of the incandescent matter contained in the crucible in 

 which the arc was observed. 



The blue line of the flame at wave-length 4570 is well seen in the 

 arc, and is easily reversed, but is always a sharp line, increased in 

 brightness but not sensibly expanded by putting magnesium into the 

 crucible. In the spark, at atmospheric pressure, it is only seen close 

 to the pole or crossing the field in occasional flashes ; but seems 

 to come out more decidedly at rather higher pressures, at least in 

 hydrogen. 



The series of bands near L, well developed in the flame, but not 

 seen at all in the arc or spark, look very much like the spectrum of a 

 compound, but we have not been able to trace them to any particular 

 combination. Sparks in air, nitrogen, and hydrogen have alike failed 

 to produce them. The very strong, rather diffuse triplet at M, with 

 which they end, so closely resembles in general character the other 

 magnesium triplets, that it may well be connected with that constitu- 

 tion of the magnesian particle which gives rise to the triple sets of 

 vibrations in other cases, but, if so, its presence in the flame alone is 

 not easily explained. 



The occurrence of this triplet in the ultra-violet, and of the re- 

 markable series of bands associated with it, as well as the extra- 

 ordinary intensity of the still more refrangible line at wave-length 2850, 

 which is strongly reversed in the spectrum of the flame, corroborates 

 what the discovery of the ultra-violet spectrum of water had revealed, 

 that substances at the temperatures of flames while giving in the less 

 refrangible part of the spectrum more or less continuous radiation, may 

 still give, in the regions of shorter wave-length, highly discontinuous 

 spectra, such as have formerly been deemed characteristic of the 

 highest temperatures. This subject we will not discuss further at 

 present, but simply remark what we have stated formerly, that " it 

 opens up questions as to the enormous power for radiation of short 

 wave-length of gaseous bodies at the comparatively low temperature 

 of flame with regard to which we are accumulating facts." 



In the arc and spark, but not in the flame, we have next a very 

 striking group of two very strong lines at wave-lengths about 2801 

 and 2794, and a quintuple group of strong but sharp lines above 



P 2 



