G9S rroft\^sor Piicar [Juno 10, 



bo duo to the sanio causo as that above suggested for tlie non-appear- 

 auco of the higher triplets, the smallness of the incaudesceut mass in 

 the spark. 



A triplet of linos in tlie are near U appear to be represented iu the 

 spark by an equally strong, or stronger, pair near but not identieal 

 in position. The possibility of sueh a shift, alTeeting these two 

 lines only in the Avhole speetrnm and aifeeting tlieni unequally, must 

 in the present state of our knowledge be very much a matter of 

 speoulation. Perhajis sut^eient attention has not hitherto been 

 directed to the probability of vibrations being set up directly by the 

 electric discharge independently of the secondary action of elevation 

 of temperature. Some of the observations above described, and many 

 lathers well known, indicate a selective action by which an electric 

 discharge lights up certain kinds of matter in its patli to the exclu- 

 sion of others ; and it is possible that in the case of vibrations which 

 are not those most easily assumed by the particles of magnesium, the 

 character of the impulse may slightly atVect tho period of vibration. 

 The fact that, so far as observations go, the shift in the case of this 

 pair of magnesium lines is detinite and constant, militates against tho 

 supposition suggested. On the other hand, the ghost-like pairs of 

 lines observed in the spark below the triplet near L, suggest tho 

 idea that some of the particles have their tones ilattened by some 

 such cause. 



The strong pair at wave-lengths 2801, 2794, are accompanied in 

 the spark, but not usually in the arc, by a much feebler, slightly more 

 refrangible pair, but these have not the dilluse ghost-like character 

 of those just alluded to. 



These lines are phenomena of the high potential discharge in 

 which particles are torn off the electrodes with great violence and 

 may well be thrown into a state of vibration which they ^vill not 

 assume by mere elevation of temperature. 



Tliere are two lines in the spark besides the well-known line at 

 wave-length -l-lSl which have not been observed iu the .arc, but they 

 are feeble and would be insignilicant if it were not the fact that they, 

 as well as wave-length -l-lSl, all short lines seen generally only 

 about the poles, appear to be present in the solar spectrum. In tlio 

 sun we seem to have all the lines connnon to the flame, arc, and 

 spark, and possibly the strong triplet of the flame at M. "We have 

 noticed that when the spark is taken in hydrogen, the line at wave- 

 length -1570 appears stronger than that at wave-length -4700, while 

 the reverse is the case when the atmosphere is nitrogen. It is 

 possible then that the atmosphere may, besides the resistance it offers 

 to the discharge, in some degree affect the vibrations of the metallic 

 particles. 



The substantive result of the investigation is to prove that the 

 dicmical atoms of magnesium are capable of taking up a great variety 

 of vibrations, and by mutual action on each otlier, or (m j>articles 

 (>f nuitter of other kinds, give rise to a great variety of vibrations of 



