358 Mr. R. H. M. Bosanquet on the Tlluminating-power 



carbonic oxide. The points indicated by the ethylene mix- 

 tures lie nearly, except the two lowest, on a straight line, 

 which produced meets the y axis on the negative side at the 

 point # = 0, y = —10, or thereabouts. Thus we are led to a 

 conception of negative illuminating-power, which of course 

 consists in the capacity of neutralizing or absorbing illumi- 

 nauirg-power, and we may think of carbonic oxide as having 

 an illuminating-power of —19 candles. We can then calcu- 

 late the effect of the addition of carbonic oxide to any gas, so 

 far as the rectilinear process of the last paper serves us, by 

 joining the poinl: to # = 0, y — — 19, and finding the point 

 on the line corresponding to the new value of x. 



Air also presents a straight line, meeting the y axis in 

 y= — 35 nearly. But there are reasons for doubting whether 

 air can in general be represented in this way, though the 

 application of the above rule would be simple enough. Oxygen 

 presents series of points having values of y larger than that 

 of ethylene. It seems to be incapable of being dealt with by 

 the present method. A point could easily be determined by 

 direct experiment, w r hich would represent air or oxygen in 

 any given case. The difficulty lies in passing from these 

 ethylene mixtures to such a case as coal-gas. The two 

 lowest numbers for carbonic oxide belong to the class of cases 

 where y has values below 15, which fall below the places 

 indicated by the laws for higher values. 



The locus of the nitrogen points is distinctly curved, and 

 so to a less degree is that for carbonic acid. The rectilinear 

 method will therefore be regarded in such cases as only a first 

 approximation, and the whole of the material will be recon- 

 sidered from this point of view. 



We shall for the present assume that values of y below 15 

 are not to be depended upon in connexion with the laws of 

 higher illuminating-powers. 



Hydrogen. — Referring back to the figure in my last paper, 

 p. 126, we see that the hydrogen-ethylene line is drawn 

 through ethylene taken as #= *5,y = 65, and the lowest point, 

 #= '12, y = 6'7 nearly. Abandoning this point, it is possible 

 to sketch a curve starting from the direct ethylene point 68*5, 

 and lying fairly evenly among the remaining points. The 

 curve I first sketched passed through the origin. There 

 are, however, considerable deviations from any representation, 

 so that it is possible to vary the position of the curve a good 

 deal. On account of the importance of the case I give 

 two mathematical representations of the hydrogen-ethylene 

 mixtures. 



