108 PROCEEDINGS OF THE AMERICAN ACADEMY. 



a case such as that recorded on page 101 under Zn 4722 at 8.8 amperes. 

 It is impossible for the main hne to disappear and the ghost remain; 

 and again, even if ghosts were present, there is no reason why these 

 should appear in the case of any one line with the spark as a source, 

 and not with the arc. The presence of neither a symmetrical nor 

 unsymmetrical ghost structure could produce the enhancement of 

 the red satellites in the spark. 



A certain objection may, however, be made: namely, that the 

 presence of the diffraction pattern between the orders when the 

 instrument is in a double order condition, might cause satellites which 

 are of low intensity to appear (when otherwise they would not) in 

 much the same manner as fogging a photographic plate will carry the 

 exposures of "low lights" up along the intensity curve so that they 

 will become visible. ^^ In response to this objection, it may be said 

 that the satellites in question are not always of low intensity, either 

 visually or photographically; and they even come up on the right 

 side when the diffraction pattern lies to the left. 



We must conclude, then, that there exists for some unknown reason 

 a fairly progressive increase in the intensity of the red satellites of 

 these three zinc lines with decreasing inductance. There follows at 

 once the unsymmetrical broadening to the red of the images given by 

 instruments of less resolving power, namely, prism or grating spectro- 

 scopes. 



The unsymmetrical satellite system may be produced by the high 

 potential gradient in the spark; why, the writer, of course, cannot 

 state. Disruptiveness is not a determining factor, for in the same 

 spark we obtain from different parts of the gap different line structure. 

 Vapor density probably does not of itself determine structure, but may 

 influence the potential gradient. In the arc high density seems to 

 produce a tendency toward complexity of structure, but not an asym- 

 metry of a regular or enduring type. 



All the writer's observations, both visual and photographic, confirm 

 the results obtained by Nutting, dealing with arc structure. The 

 results of this study also confirm the shifts found by the writer ^^ to 

 exist at lower dispersion, shifts, — great at the end of a fairly large 

 gap of a non-inductance spark between terminals of the pure metal, 

 lessened or removed entirely by the addition of inductance, and by 

 the use of the central region of the gap; and lessened also by the use 

 of an alloy. In this former work the standard of reference employed 



22 R. W. Wood actually used this method. 



23 Astrophysical Journal, 22, No. 3, Oct. (1905). 



