1891.] Influence of Pressure on the Spectra of Flames. 223 



noticed to occur in the principal sections of the discontinuous 

 spectrum of the oxy-hydrogen flame. Whether the continuous 

 spectrum is due to the mutual action of the molecules of the com- 

 pressed gases may perhaps be best determined by some photometric 

 measures of the rate at which the brilliance increases with the 

 pressure. Frankland ('Exp. Researches,' pp. 892 et seq.) has made 

 some such measures, but not safficient to solve the question. We 

 have made an attempt to measure, not the total intensity of the light, 

 but that of rays of definite refrangibility. 



Photometry of Oxy-Hydrogen Flame under Pressure. 



The apparatus used for these measures was a spectro-photometer of 

 the pattern employed by Crova (' Annales de Chimie,' ser. 5, vol. 29, 

 p. 556). In this, the rays of one of the sources of light to be com- 

 pared are passed through two Nicol's prisms, and then reflected 

 into one half of the slit of the spectroscope, while the light from 

 the other source passes directly into the other half of the slit. 

 By turning one of the Nicol prisms, the light from the first source 

 can be reduced at pleasure, and any small section of the spectrum 

 can be separately observed by cutting off the rest by means of a 

 shutter in the eye-piece. We found it by no means easy to get 

 good concordant observations. A much larger vessel was used than 

 for the earlier experiments, one which contained several litres, and 

 so we may presume a more uniform pressure was maintained within 

 it. The results of the best series of observations on the photometric 

 intensity of the jet of oxygen burning in hydrogen are given in the 

 following table. The comparison light was a petroleum lamp. 



The first column gives the pressure of the gas, the second the 

 lean of four to six observations of the angular deviation of the Nicol's 

 nsms from the position of complete extinction, for each pressure. 

 The third column gives the squares of the sines of the angles in the 

 ind column multiplied by 100,000. 



It will be seen from the last column that the numbers in the third 

 jlumn, which should be proportional to the photometric intensities 



