THE SPECTRUM 313 



negative carbon and shorten the arc, the bright lines fade 

 and we gradually get dark lines. 



Weinhold's combustion-lantern, shown in figs. 168, 169, 

 would probably give good dark lines by keeping the door open 

 at the back, and sending the rays from the arc through the 

 flame ; but I have not been able to test it in this way. 



With the oxy-hydrogen flame, direct reversal is more 

 difficult, and is practically confined to the sodium line. I 

 gave ! the arrangement in fig. 171 for this light, after fairly 

 successful trial ; but I have since found that my success was 

 due almost entirely to the powerful jets employed, and that 

 even with these, such an arrangement cannot be depended 

 upon, the Bunsen-flame being not cool enough in proportion, 

 and so nearly in focus with the slit as to overpower the yellow 

 portion of the radiation spectrum. Mr. E. Cleminshaw was 

 quite justified in some criticism to this effect at the Physical 

 Society, 2 and I was very glad to find that his arrangements 

 gave better results than my own. They are as follows. 

 (a) The Bunsen burner and sodium-flame are placed at the 

 focus or crossing point of the lens, where the rays pass 

 through the smallest space, and can be passed through a 

 small flame where the vapour is most dense. The screen is 

 then easily shaded from direct rays, and the sodium-flame is 

 notfocussed on the screen. The latter is the great advantage 

 of this method. A spirit-lamp may be used instead of a 

 Bunsen burner, but with metallic sodium, not with salt. 

 (6) If the sodium-flame is used between the slit and the lens, 

 after vivid combustion has been produced, the Bunsen flame 

 is cooled by passing into it, along with common air, a portion 

 of carbonic dioxide, generated in a bottle as usual, and with two 

 vent-tubes, so that the supply can be graduated, (c) Another 

 plan is to arrange a spirit-lamp with four wicks, with a jet of 

 oxygen in the centre. Using oxygen in the flame, the bright 



1 Light: a Course of Experimental Optics. 

 8 Proc. Physical Society, vii. 53. 



