TIIK CONSTITUTION OF THK KLECTRIC SPAKK 199 



be seen in the instrument traversing the spectrum horizontally. They served as 

 reference lines along which the displacements could be measured. 



Accurate measurements of wave-lengths were not necessary for our purpose as lone 

 as we could identify the different lines. As the prism was not disturbed during the 

 principal series of our experiments, the air linos were found to be sensibly at the same 

 distance from each other in all our photographs. Making use, therefore, of a few of 

 these air lines, which are sharp, and of the lines of zinc and cadmium, which are 

 easily identified, a curve may l>e drawn from which the wave-lengths corresponding 

 to any reading of the comparator may be determined in the usual way with quite 

 sufficient accuracy. 



4. Preliminary Experiments. 



Before entering into the main subject of our research, we desired to become familiar 

 with the appearance of the spark itself, and with such other phenomena connected 

 with it as could be elucidated by photographs of the spark itself. We describe the 

 results obtained, selecting out of a number of plates those which seem to present 

 distinctive features of interest. 



Fig. 11, Plate 9, November 10, 1897. Image of spark taken between zinc and 

 brass electrodes ; 5 Leyden jars were used, their total capacity was not measured, but 

 the jars were smaller than that of the battery of six used in the later experiments, 

 and the total capacity was estimated to be about half, i.e., about '015 microfarad. The 

 chief feature of this photograph is a slightly curved very luminous column, containing, 

 as was subsequently ascertained, the first discharge breaking through the air. This 

 is surrounded by a cloudy appearance, which is due to the metallic vapours generated 

 by the first discharge. 



Our later experiments, to be described further on (p. 211), point to the conclusion 

 that the oscillations following the first discharge pass through this cloud of metallic 

 vapour. Some sharp linear luminous filaments may be seen in the original on one side 

 of the main discharge. These set out from small projections on the electrodes and 

 precede the main discharge, as may be ascertained by carefully watching the image on 

 the focussing plate of the camera. The preliminary discharges may be avoided by 

 giving careful attention to the polish of the electrodes. 



Fig. 12, November 10, 1897. A photograph of the spectrum of the preceding 

 discharge, taken with 15 sparks. The image of the central column was thrown on 

 the slit. The spectrum is that of zinc and air, with one or two lines of copper, 

 and apparently the calcium lines H and K. A trace on the calcium line at 4227 also 

 shows on the original. 



Fig. 13, Plate 9, November 10, 1897. The same as fig. 12, except that the cloudy 

 portion of the discharge was focussed on the slit, which had to be widened a little in this 

 case. Thirty sparks had to be taken to secure an impression which could be com- 

 pared in intensity with the previous one. The absence of air lines, verified also by 



