KENT. — SPECTRA OF THE ZINC ARC AND SPARK. 95 



mediately above one another with different exposure times. The long 

 exposures seem to give a different center of intensity from the short, 

 if a line is unsymmetrically broadened to one side; whereas on the 

 other hand the real center remains clearly in the same position only 

 in the case of sufficiently short exposures. The long and short ex- 

 posures play the same role, however, as a greater or smaller density 

 of metallic vapor; therefore the shifts observed by Exner and Haschek 

 are to be considered only as apparent. Exner and Haschek then 

 tried to maintain their theory by referring the cause of the shifts to 

 changeable satellites, which cannot be resolved by a Rowland grating 

 and might therefore produce a shift. They studied the arc lines of a 

 series of elements by means of a 15 plate echelon and made the aston- 

 ishing discovery that a satellite often appeared upon the red side of 

 the line, especially when the arc flickered. With the plane parallel 

 plates at my disposal, which are more efficient than a 15 plate echelon, 

 I have been unable to verify the satellites which they reported." * * * 

 " It is possible that the satellites seen by Exner and Haschek with the 

 flickering of the arc arose from impurities in the carbon and the metal. 

 It is more probable, however, that they must be regarded as ghosts. 

 Ca X 4527 is supposed to be simple, but with a satellite arising on the 

 side of greater wave-length upon the flickering of the arc; whereas I 

 found no satellite near this strong line. On the contrary, I observed 

 a weak satellite of greater wave-length near Ca X 4586, while Exner 

 and Haschek did not. Ca X 5270 is supposedly a triplet, in which 

 with weak current the middle line is the brightest; with strong current 

 the two lines toward the red are the brightest. All my photographs 

 show this very strong line to be single; furthermore, Cu X 5218 is 

 supposed to have a red companion which grows more rapidly than the 

 head-line as the current is increased; I always found this very strong 

 line to be single. This very line seems to me proof that Exner and 

 Haschek were deceived by ghosts in their echelon. For if the head- 

 line is not very strong, the ghost can scarcely be seen; if the main 

 line becomes stronger, the ghost comes out more strongly; with 

 further increase in intensity, the main line, however, seems to gain less 

 rapidly than the ghost, since the eye (Exner and Haschek make 

 visual observations only) cannot distinguish differences in great 

 intensities so accurately as in the case of small ones. Nutting has 

 also used the ordinary arc for creating spectrum lines and worked 

 with an echelon of 30 plates, of 1| cm. thickness. The same remarks 

 as above made are valid in case of the use of the carbon arc." 

 Janicki reviews Nutting's results, characterizes them as extraor- 



