not an Oxy chloride. 357 



we were to fix 5 per cent of the whole mass as the proportion 

 changed by light. So that the amount of oxygen that would, 

 according to the oxychloride theory, he introduced into a given 

 quantity of chloride by prolonged exposure would not exceed 

 three twentieths or 045 of one per cent, of the material under 

 examination, a dangerously small quantity on which to attempt 

 to decide an important question, especially where the estimation 

 is indirect. If even a very careful determination of the silver 

 and the chlorine present should bring the sum of these to a 

 quantity represented by figures amounting to from 99 '80 to 

 99*90, would it be allowable to assume that the difference be- 

 tween this and 100 consisted of oxygen and so to take the 

 presence of an oxychloride as proved ? Such reasoning could 

 not be accepted ; the errors incident to the most careful anal- 

 ysis would too largely affect the point vitally at issue, not to 

 speak of the entire absence of proof that the deficient quantity 

 was oxygen. 



These considerations convinced me that it was not in that 

 direction that one should seek for proof of the presence or ab- 

 sence of oxygen in the the substance in question. I therefore 

 looked for what may be called proof by exclusion. 



Coal naphtha (refined petroleum) is a substance absolutely 

 free from suspicion of containing oxygen or moisture as im- 

 purities, so much so as to be universally used for the preserva- 

 tion of sodium. I have sodium that has been preserved in this 

 way for over thirty years. Silver chloride was precipitated 

 with excess of hydrochloric acid, was washed in a darkened 

 room and dried in a dessicator. From this it was transferred 

 to a porcelain crucible, covered and fused over a lamp. When 

 thoroughly fused (in this condition it is as fluid as water) it was 

 poured directly into naphtha. This naphtha had been placed 

 to the depth of an inch or more in a dry porcelain vessel, which 

 was first well wiped out with naphtha to remove the film of 

 atmospheric moisture which condenses on surfaces. 



The chloride congealed into a pale gray lump which, whilst 

 it remained under the petroleum, was absolutely free from all 

 possibility of contact with oxygen, free or combined. Without 

 removing it, the vessel was moved into the sunshine. When 

 touched by sunlight, the chloride instantly became as black as 

 ink. 



This experiment seems decisive as to the oxychloride theory. 



It seemed desirable not to stop here, but to find a means of 

 applying an equally decisive proof by exclusion to the converse 

 case. In the above instance, a photosalt was formed by reduc- 

 tion starting with normal chloride. The converse case would 

 be the formation of a photosalt by chlorination, starting with 



