1024 



the tops of the lines to be compared were in a line beside each other. 

 The distance between the dark lower ends was then measured. As 

 in tliis waj the two lines could be compared over a fairly great 

 length and both appear on the same back ground, the difference 

 in length can be prettj^ correctly measured. Not more accurately 

 however than to about 0.2^^ mm. 



With intensity-ratios of 1 : 3,95 : 8,80 as measured by the 

 spectro-photometer we found in the mean differences of length of 

 3,7 and 6,5 mm. Calculating from these the intensity-ratios we 

 find 1:3,6:9,4*. These deviations correspond to difference in length 

 of 0,2' and 0,2 mm. and are therefore of the order of accuracy 

 with which the length of the lines can be determined. We could 

 not find a systematic difference for the various lines ; the lines 

 measured were 5191—5170, 5461, 4359, 4047 and 3650. 



We also tried the sector for much greater intensity-ratios, for 

 instance 1 : 60 and 1 : 100. In these cases the short line was always 

 found too short. It is certainly not surprising that the sector is not 

 correct in the extreme fields. For some investigations, however, 

 these are not of importance; in which cases the method may be 

 usefully applied. The method^) was not sufficiently accurate for 

 us to be in a position to decide as to the validity of Bunsen and 

 Roscoe's law. 



To summarize the advantages and disadvantages of this method 

 we may say : it gives on a single photographic plate an easily read 

 image of the spectral intensity including the ultraviolet part of the 

 spectrum ; it enables us to determine quantitatively changes in the 

 intensity, the measurements do not cost much time and require 

 only a few photographic plates. The accuracy of the method, however, 

 is not more than about 10 7o- Greater intensity-ratios than about 

 1 : 20 cannot very well be determined by it. These characteristics 

 of the method make it very suitable for preliminary investigations, 

 which can subsequently be worked out in detail by a less rapid 

 and usually less easily readable method. Researches of this kind 

 concerning the emission of light by gases and their mixtures, have 

 been made by one of us which have demonstrated its usefulness. 



It remains to express our sincere thanks to Dr. G. L. F. Philips 

 for the friendliness with which he enabled us to carry out this research. 



Eindhoven. Laboratories of Philips's 



Incandescent lamp factories. 



1) As a matter of fact all photographic methods have only a relatively small 

 accuracy. A. E. Weber (1. c.) says for instance that if abnormal values are excluded 

 a possible error of about 8 % must be reckoned with. 



