102 Nichols and Franklin — A Spectro-photometric 



their brightness in other regions being expressed in terms of 

 that of the corresponding wave-length of this standard spec- 

 trum. This method of presenting the results of spectro-photo- 

 metric comparisons, seems to us to offer many advantages. It 

 was adopted by Mr. W. H. Pickering in an investigation made 

 some years since, and covering very much the same ground as 

 our own ;* and although other wave-lengths have been selected 

 by some writers, for instance that of a certain region in the 

 green (X=5570) which Otto Schumann has taken as unit of 

 brightness in his recent study of the spectrum of the incan- 

 descent lamp, there seems to be no good reason for abandoning 

 the region of the D line, as a reference point, in favor of any 

 other portion of the spectrum. 



Candles, Oil and Gas. 



Measurements were first made upon two standard candles 

 burning side by side in a double holder designed for use in the 

 Bunsen photometer. Every one who has had experience with 

 this light in photometry is aware that the candle is very far 

 from being a constant source of illumination. Pickering in 

 the paper already cited has expressed the opinion that, " of all 

 the artificial lights examined . . . none was so uncertain in 

 color as the standard candle." We concluded from the con- 

 sideration of many widely fluctuating values, taken in various 

 parts of the spectrum, that the color of the candle, under the 

 most favorable conditions of combustion, was very nearly the 

 same as that of the 16-candle lamp used as a standard in our 

 measurements, but that its average color was decidedly redder 

 than that of the incandescent lamp. 



Experiments upon various petroleum and gas flames showed 

 that these also were of very nearly the same color as the incan- 

 descent lamp. Nearly all of them were subject to slight fluc- 

 tuations in color and intensity, due to variations in the condi- 

 tions of combustion which were beyond control. These fluc- 

 tuations, although much less marked than in the case of the 

 candle, were sufficient to cause changes of color greater than 

 the average difference of color between the flame under inves- 

 tigation and the incandescent lamp. One of the gas flames 

 subjected to measurement, that of a lamp of the type known 

 as the " board of trade " argand, a sixteen-candle standard con- 

 structed especially for photometric work, was however of 

 remarkable steadiness both in color and intensity, and measure- 



* W. H. Pickering, Photometric Researches ; Proceedings of the American 

 Academy of Arts and Sciences, vol. xv, 1880. 



