3(32 Mr. R. H. M. Bosanquet on the Illuminating-power 



in the mixtures. It appears desirable to experiment with 

 mixtures differing little from marsh-gas in composition. 



Let us now examine the locus which represents the hydro- 

 gen mixtures. 



Differentiating the equation, 



we have 



dy 

 dx u 



In our diagrams the units of y are 100 times those of a. 

 So that if 6 be the inclination to the x axis, 



For the hydrogen-ethylene mixtures c = 3'6, 

 and tan 0= -036?/. 



(Similarly, for the nitrogen-ethylene mixtures c = 6'6, 



and tan 0= -066?/.) 



For very small percentages of ethylene it seems that no 

 light is given. But long before enough carbon is absorbed to 

 fill up the solvent capacity of the hydrogen present, light 

 appears. Raw hydrogen therefore presents a resistance to 

 the entire absorption of carbon of density 2. And, as far as 

 we can ascertain, this resistance continues present in all the 

 mixtures, as there remains unsaturated solvent capacity 

 throughout. The effect of this is that, in the regular part of 



the curve, -~ and tan 6 fall below the values they might be 



expected to have if the absorption were complete, part of each 

 addition of carbon being absorbed, and part adding to the 

 C.P. In the early part of the curve the inclination is 

 steeper, and the explanation is a little difficult. I think that 

 for the present these low values must be put aside. I 

 append a table showing what I suppose to be the rational 

 explanation of the behaviour of the different mixtures. To 

 explain the very low values, where no light is shown, we must 

 suppose the existence of a resistance to precipitation until the 

 first nuclei are formed. 



The solvent capacity in presence of raw hydrogen cannot 

 be taken as less than '166, as there is '165 of absorbed carbon 

 in the highest mixture. A higher value would suppose the 

 presence of an unsaturated capacity, which would have to be 

 balanced by an increased resistance to absorption. The con- 

 stancy of the solvent capacity is an assumption. It can hardly 



