Mr Campbell, Discontinuities in Light Emission. 517 



currents in the mercury probably cause slight changes in the 

 reflecting power of the walls at various points. Additional 

 fluctuations will be due to this cause, to which may also prob- 

 ably be attributed the great irregularity of the magnitude of the 

 fluctuations for the reflecting lamp as compared with those of 

 the simple lamp. These irregularities are so great that it has 

 not been thought worth while to give the results of any of the 

 very large number of measurements which were made on the 

 fluctuations with a lamp of this nature. In some cases measure- 

 ments made at different times under apparently similar conditions 

 differed by more than 100 7o- 



§ 5. When the reflecting lamp proved a failure the ingenuity 

 of the author was insufficient to devise any new form of source of 

 light which appeared to offer any reasonable chance of success. 

 The difficulty which has rendered the experiment impossible was 

 not foreseen when the work was started, but has proved insuper- 

 able. Its nature may be described briefly once more. 



The object of the experiment was to compare the fluctuations 

 due to two beams from independent sources, i.e. beams from 

 different incandescent surfaces, with those due to two beams from 

 dependent sources, i.e. beams proceeding from the same incan- 

 descent source and divided, either by being emitted in different 

 directions or by some such process as partial reflection and trans- 

 mission. In order to be free from fluctuations due to changes in 

 the total intensity of the source, either a source of absolutely 

 constant intensity must be found, or some means devised of 

 splitting the beam in such a way that different parts of it arrive 

 at the cells without any changes in their relative intensity. 



§ 6. Though the main object of the research had to be 

 abandoned, some observations were made on the fluctuations due 

 to independent sources. It was hoped that these might throw 

 some light, both on the theory of the measurement of such 

 fluctuations and also, possibly, indirectly on the main problem 

 which had inspired the research. 



The independent sources were two 8-volt 16- watt Osram lamps 

 placed one opposite each cell. The total intensity of the light 

 falling on the cell was varied by changing the distance between 

 the lamp and the cell. Compensating resistances were introduced, 

 so that a change in the e.m.f. of the battery from which the lamps 

 were run in parallel did not cause any change in the relative 

 intensity of the light from the two lamps. The remainder of the 

 apparatus was exactly as described in the previous paper. The 

 observations were made and the fluctuations calculated as has 

 been described with one exception (§ 17). In place of series 

 consisting of 200 observations, series of 20 observations were 

 taken. The object of the change was to eliminate errors arising 

 from the constant drift of the position of balance : it was difficult 



34—2 



