68 Messrs. Paterson and Dudding on the 



a 



The photometric work involved has been laborious. In 

 order to eliminate any uncertainties of individual lamps and 

 to reduce the probable error, each of the six sets of lamps 

 consisted of 10 or 12 individual lamps, and each set was 

 compared with the next one, in order of colour, some six or 

 eight times by each observer. This entailed in all between 

 2000 and 3000 standardizations, or between 10,000 and 

 15,000 observations. The work was started several years 

 ago, but the difficulty experienced at first in obtaining metal 

 filament standards which were sufficiently constant caused 

 much delay in the early stages. 



Fig. 2 indicates the scheme of the cascade comparison 

 graphically, and the relation of each group of standards to 

 the remainder. 



Set 1 represents a series of 11 sub-standards whose radia- 

 tion matches in hue that of the pentane lamp, or the radiation 

 from a black body at 1610° C. Set 2 indicates a series of 11 

 Fleming-Ediswan sub-standards operating at an equivalent 

 temperature of 1700° 0. Set 2 was compared against set 1 

 by the usual substitution method of photometry, using a 

 Lummer-Brodhun photometer, and a comparison lamp, the 

 colour of whose radiation lay midway between that of sets 1 

 and 2. In a similar way, set 3, with an equivalent tempera- 

 ture of about 1750° C, was compared against 2 through 

 another comparison lamp, which also divided the colour- 

 difference, and so on to set 6. It will thus be seen that by 

 such a substitution method as this the colour difference 

 between sets 1 and 6 is divided into 10 approximately equal 

 steps — that is to say, at any one measurement there is only 

 a colour difference of one-teuth that of the difference between 

 sets 1 and 6. The difference of colour in the photometer 

 throughout the investigation was, therefore, never greater 

 than that resulting from a difference of temperature of 

 about 45° C. All the measurements were made with an 

 illumination at the photometer screen of 10 metre-candles. 

 Finally, having by this means, and for each observer,, 

 assigned candle-power values to each of the lamps in set 6 r 

 the latter were compared in one step with set 1. 



It is of interest to note that the colour of the radiation 

 from melting platinum is the same as that of set 3, marked 

 by line 3 in fig. 1 *. 



Before passing to the details of the work the authors desire 

 to point out one of the advantages of the cascade system as- 

 applied to electric sub-standards. The possession at a stan- 

 dardizing laboratory of a regular gradation of sub-standards 



* Loc. cit. 



