Mr Campbell, Discontinuities in Light Emission. 515 



passages for the light through the mercury, and corresponded to 

 the holes in the silver in the earlier arrangement. The mercury 

 acted also as a cooling bath, the heat being removed by means of 

 a stream of water flowing through a jacket covering the top and 

 surrounding the upper part of the iron vessel. Since mercury is 

 not nearly so perfect a reflector as silver, the filament could just 

 be seen on looking in through the tubes, but it could be dis- 

 tinguished only with difficulty from the bright background 

 furnished by reflection. 



With this arrangement it was found that the relative intensity 

 of the beams emerging from the tubes, as indicated by the position 

 of balance of the measuring instrument, did not vary appreciably 

 with the total intensity of the light. A change of 10 °/^ in the 

 current through the lamp was not accompanied by any change in 

 the position of balance which could be observed. The explanation 

 of the change in the relative intensity of the beams from a Nernst 

 lamp with a change in the total intensity of the light was now 

 obvious. The beams into which the light is divided do not come 

 entirely from the same parts of the filament, owing to slight errors 

 in the adjustment of the optical train*. When the resistance of 

 the filament varies, all parts of it do not undergo the same change 

 of temperature, and the light emitted by one part of the filament 

 increases relatively to that emitted from another. But, with the 

 arrangement now adopted, light from every part of the filament 

 emerges through both tubes, and changes in the relative intensity 

 of the light from different parts of the filament have but little 

 influence on the relative intensity of the light emergent through 

 the two tubes. 



§ 4. But though the main difficulty which had beset the work 

 hitherto was thus overcome, a solution of the problem had not 

 been attained. For, when an attempt was made to measure the 

 fluctuations obtained on balancing the currents through the two 

 cells, due to the light emergent firom the two tubes and incident 

 upon the cells, it was found that these fluctuations were very 

 much larger than those which had been obtained under similar 

 conditions with light from two wholly independent lamps. 6t^ 

 was some 12 — 18 times larger for the two "dependent" than for 

 the two " independent " sources : and, moreover, the magnitude of 

 the fluctuations was much more variable, so that it was impossible 

 to obtain measurements of them which showed any useful agree- 

 ment. Since on neither of the theories of light which are being 

 examined the fluctuations due to " dependent " sources should be 

 greater than those due to " independent," it was clear that some 

 important influence had been overlooked. 



* See § 19 of the previous paper, where it is noted that the effect depends on the 

 adjustment of the optical train and not on that of the measuring system. 



VOL. XV. PT. VI. 34 



