1881.] on the Oriijin and Llcntifi/ of Spectra. G87 



series of flutings and a lino in the bluo are also seen. ... On 

 passing the spark all tlicso bnt the two C()nij)()nont8 of h arc aboliHhcd. 

 We get tlio wide triplet replaced by a narrow one of the same f<.rni, 

 the two lines of h being common t > both. 



"May wo consider tlie existence of these molecular states as 

 forming a true basis for Dalton's law of multiple proportions ? If so, 

 then the metals in dillerent chemical combinations will exist in 

 diilercut molecuhir grcnipings, and we sliall be able, by spectrum 

 observations, to determine the particular heat-h^vel to wliicli the 

 molecular complexity of the S(did metal, induced by cliemical afhnity, 

 corresponds. . . . Examples. — None of the lines of magnesium 

 special to the flame spectrum are visible in the spectrum of tho 

 chloride either when a flame or a spark is employed." 



In order to ascertain the true cause of the variations in tho ma^'- 

 ncsium spectrum, tho following exj^eriments and observations wero 

 made, and they demonstrate that the views of Mr. Lockyer on this 

 question must also be regarded as resting on faulty experimenting : 



1. Observations on the Spark between Magnesium Points in Nitrogen 

 and Carbonic Oxide at various Pressures. 



The points were pieces of magnesium wire. Round one end of 

 each a platinum wire was tightly coiled and fused into tlie side of a 

 glass tube. This tube was attached by fusion at one end to another 

 tube filled with phosphoric anhydride, wliich in turn was connected 

 with a Sprengel pump. The other end of tlie tube was connected 

 by a thick rubber tube, capable of being closed by a pinchcock, with 

 a gasholder containing nitrogen over strong sulphuric acid. Tho 

 tube having been exhausted and filled with nitrogen two or three 

 times, it w^as found that no line at 5210 was visible in tlie spark. 

 The tube was now gradually exhausted, and the s])ark watched as 

 the exhaustion proceeded. No line at 5210 was seen, although tho 

 exhaustion was carried nearly as far as the pump would carry it ; nor 

 was any hydrogen line (C or F) visible, either with or without tho 

 use of a jar. The communication with the gasholder was now 

 opened, and the tube refilled with nitrogen at tho atmospheric 

 pressure ; a communication was then made with another vessel con- 

 taining hydrogen, which was allowed to difi'uso into the tube for a 

 very short time. On now passing the spark, the line at 5210 at onco 

 appeared, although the quantity of hydrogen difl'used into tho 

 nitrogen must have been very small. The experiments with nitrogen 

 at reduced pressure were repeated several times, with tho same 

 result. It was found necessary to have the phosphoric anhydride, as 

 without it traces of moisture were left or found their way through 

 the pump into the tube, and then, when the exhaustion was carried 

 far enough, both the line at 5210 and the hydrogen lines, C and F, 

 made their appearance. We have never, however, been able to detect 

 the line at 5210, in nitrogen, without being able to detect C or F 



