26 Messrs. Nichols and Snow on the Character of 



simultaneous. At this stage the luminescence of the oxide is 

 very marked. In these diagrams ordinates express the bright- 

 ness of each wave-length in terms of that of the corresponding 

 wave-length in the spectrum of the comparison-lamp. 



In the course of our experiments it was found to be more 

 and more difficult to obtain consistent measurements as the 

 temperature of the film increased ; and it soon became evident 

 that we had to deal with an unstable state of incandescence, 

 the oxide being much brighter when first ignited than at any 

 later period. The changes were most rapid during the first 

 ten minutes after the foil was brought to incandescence, after 

 which time the rate of change was not such as seriously to 

 interfere with the accuracy of the determinations. All the 

 results given thus far were obtained from films which had 

 been heated for a sufficient time to ensure their permanence. 

 Values for films freshly ignited were also obtained. They 

 were always higher than those of the older films. Such 

 results, owing to the fact that they pertained to surfaces the 

 radiation of which was fugitive, have been excluded from the 

 tables and curves. It was deemed desirable, however, to 

 determine as accurately as possible the manner in which the 

 incandescence of fresh films diminished as a function of the 

 time. For this purpose a set of rapid readings were made. 

 A film newly prepared was ignited, and settings of the inten- 

 sity were made at intervals of ten seconds from the time when 

 the current was applied. These readings were less accurate 

 than those made at leisure, and one setting, instead of a 

 series of five or ten, had to be taken to represent the radiation 

 at a given instant of time. The same method was then pur- 

 sued for another wave-length, a fresh film being prepared for 

 the purpose, and this was repeated until a series of time- 

 curves were secured which included the entire spectrum. The 

 results for each wave-length gave a curve showing the rate 

 of diminution with the time. The curves were all of the same 

 general type ; but they showed the time-effect to be least in 

 the red, and to increase as the wave-length of the region 

 observed decreased. From these data it was also possible to 

 construct a series of intensity-curves, showing the relative 

 brightness of the spectrum and its character at any given time 

 between the limits covered by the observations. These curves 

 for thirty seconds, sixty seconds, three hundred seconds, and 

 six hundred seconds from the moment of ignition of the oxide 

 are given in fig. 10. They bring out in the most striking way 

 the evanescent character of the light emitted by the oxide at 

 the temperature in question, which was 1013°. This " time- 

 effect " became verv noticeable at about 900°; and it seems 



