MR. CHARLES DE WATTEV1LLE ON FLAME SPECTRA. 147 



orthochromatic films, developed with amidol and sodium sulphite, and for the prism 

 photographs LUMIERE plates, series A and B, one series being sensitive to the yellow 

 and green and the other to the yellow and red, CRISTALLOS being used as the developer. 

 All these preparations have given good results, except for their known insensitiveness 

 in the green region of the spectrum. This inconvenience I have not been able to 

 overcome by means of sensitizing agents. 



The times of exposure have been eight hours on the average with the grating, from 

 three to four hours with the glass prism, and six hours with the quartz prism. It 

 would certainly be advantageous to prolong these exposures still further for the ultra- 

 violet portion of the spectrum, where the quartz apparatus gives a greater intensity 

 and also a greater number of lines than the grating. 



III. METHODS OF MEASUREMENT. 



I have used several methods for measuring the photographs obtained. In the case 

 of the grating photographs the simplest method consisted in finding exactly the 

 distance between two lines, using a magnifying lens and finely pointed dividers, and 

 transferring it to a divided scale. Knowing the difference of wave-length corre- 

 sponding to 1 millim. on this scale, the difference of wave-length could then be 

 deduced. In this way one could obtain approximate results, the maximum error of a 

 of a millimetre in the readings corresponding to an error of 4 Angstrom units. 

 This approximation was sufficient when it was a question of determining easily 

 recognized lines, and one had rather, as in the present case, to identify the lines of 

 the Hame sjHJctrum with those of the adjacent and known spark spectrum of the 

 same metal than to make an exact measurement. In certain other cases this 

 summary method has not been sufficient, and I have then had recourse to photographs 

 obtained with the prism spectroscopes on which the comparison spectrum wa always 

 that of the oscillatory electric spark between iron poles. The position of the line has 

 then been obtained definitely by reference to the atlas of the iron spectrum prepared 

 by KAYSER and RUNOE. Finally, for metals whose spectra are very rich in lines, for 

 example the metals of the iron group, I have been compelled to use the ordinary 

 methods of measurement. The new measuring machine described by M. KAYSER.* 

 and which belongs to the Physical Laboratory of the University of Manchester, has 

 been of the greatest use to me. The time needed for the measurement of a photo- 

 graph is considerably shortened by its use, thanks chiefly to the automatic printing 

 apparatus with which it is provided. Any required wave-length is obtained, as is 

 well known, by interpolation between the wave-lengths of two neighbouring known 

 lines. 



The lines due to the carbon in the coal-gas and to the water- vapour are present in 



* ' Handbuch der Spectroscopie,' vol. i., p. 644. 



v 2 



