xvi.] METHODS OF COMPARISON. 213 



In photographing the long and short lines of metallic spectra, 

 it was found that this object could not be well attained with the 

 electric lamp in its usual position (with vertical poles), as the 

 central column of dense vapour, as a rule, extended across the 

 arc, i.e. from pole to pole, and gave all the short lines. In 

 order to obviate this a horizontal arc was employed. This 

 was accomplished by placing the lamp on its side and firmly 

 securing it in that position. An image of the horizontal arc 

 was then thrown on the vertical slit in the usual manner. 



The lines thus photographed are pointed at either end, and 

 disappear from the centre in the order of their length, so that 

 an exquisitely symmetrical double or duplicate determination of 

 their lengths is thus obtained. 



Although the lengths, thicknesses, intensities, and reversals 

 of the lines were thus readily recorded, we had so far no very 

 convenient method by which to fix their positions, or to deter- 

 mine their coincidence or non- coincidence with solar lines. In 

 order to accomplish these ends, I .resolved to photograph the 

 solar spectrum on each plate immediately above or below the 

 metallic spectrum under examination. To do this an extension 

 of the method of working hitherto in use was introduced, 

 depending upon the following considerations : 



It is obvious that when we observe a spectrum its breadth will 

 depend upon the length of the slit. "When we at the same time 

 illuminate different portions of the slit with rays proceeding 

 from different vapours, the spectra of the different light-sources 

 are seen at once. But when we introduce photography we can 

 more conveniently obtain results by illuminating successively 

 different portions of the slit, the effect being that the various 

 spectra will successively record themselves on different portions 

 of the photographic plate. 



Acting on this principle I first covered up the upper half of 

 the slit, allowing the image of the horizontal arc to fall centrally 

 on the slit, so that in this way there was impressed an image of 



