Tests of Glow- Lamps, 403 



The standard of light we used was a two candle-power 

 Methven screen fed with pentaned gas, and against this a ten 

 candle-power Bernstein glow-lamp was standardized to give 

 five candles, the Bernstein lamp being employed for the 

 actual comparison with the Edison-Swan lamps. 



One of the objects of employing this Bernstein lamp was 

 to test whether or no a convenient and trustworthy standard 

 of light could be obtained by using a low-voltage lamp with 

 its filament at a comparatively low temperature. From this 

 point of view, however, the results were not very satisfactory, 

 as at the end of some sixty hours of life the lamp had 

 blackened heavily. This secondary standard was of course 

 tested very frequently — as a rule, every time it was used. 



For the last few hundred hours of our tests an 8 C.P. 

 100-volt Edison-Swan lamp running at five candles was used 

 as an intermediate standard of light, and after the first few 

 hours it gave very fair results, the current required to pro- 

 duce exactly five candles dropping in almost a straight line 

 from 0*3893 to 0*3867 ampere in two hundred hours. It is 

 probable, for reasons which will be shortly apparent, that an 

 even better result would have been obtained if the lamp had 

 been run for some fifty hours at its normal brilliancy before 

 it was used as a standard of light. 



For measuring the light a Lummer-Brodham photometer 

 was used, and proved itself a very convenient instrument. 

 The lamp-stand was fixed at the end of a three-metre photo- 

 meter bench, the actual distance between the standard and 

 the Edison-Swan lamps being 319*5 centims. 



Some difficulty was found at first in measuring the candle- 

 power owing to the very considerable difference in colour 

 between the lights being compared. This difficulty was con- 

 siderably reduced by fixing the Bernstein lamp to the spring 

 frame shown in fig. 9, where A is the lamp, S, S are steel 

 springs, and W, "W are leaden weights. 



When measurements of the candle-power were being taken, 

 the Bernstein lamp was set gently swinging backwards and 

 forwards in the direction of the bench, and the effect thus 

 easily produced was found to be an improvement on that 

 usually obtained by moving the photometer first a little on one 

 side, and then a little on the other, of the neutral point. After 

 some practice the use of this contrivance enabled us to make 

 measurements of the candle-power which did not differ from 

 each other by more than 0*3 per cent. 



The first point noticed after starting the tests was that 

 candle-power, current, and candles per watt all rose as the 

 lives of the lamps increased; and as we were not prepared, 



