ON THE ACTION OF LIGHT ON THE UYDKACIUS OF THE HALOGENtJ. 61 



that the ether used was very carefully purified, and when tested by the 

 iodoform reaction it gave only a very minute trace of alcohol ; the oxygen 

 used was purified by passing through solutions of potassium iodide, sodium 

 hydrate, and water. The mixture of ether and water was tested in one 

 experiment after exposure ; it was found to be neutral to litmus, and 

 contained only a trace of alcohol, whilst the gas above the liquid was 

 free from carbon dioxide ; the ethereal solution, however, freely liberated 

 iodine from potassium iodide. 



The results of seven experiments are given in the following table : — 



It will be seen from these experiments that the amount of hydrogen 

 peroxide which is formed is dependent on the proportion of ether to 

 water present. Thus in No. 1 there was no excess of liquid ether, but 

 it was completely dissolved in the water ; again, when the water and 

 ether are present as gas no peroxide is formed (see No. 5) ; this is also 

 the case when oxygen is absent (No. 6), and when the solutions are 

 exposed to the same temperature changes in the dark in an atmosphere 

 of oxygen. 



The next experiments were made with water and oxygen, the ether 

 being omitted. 



Two bottles of 900 ccs. capacity were charged with 600 ccs. of water 

 containing 2'5 per cent, of pure suliihuricacid ; the space above the water 

 was filled with oxygen. In one experiment the exposure was made in 

 sunlight, whilst in the other case the liquid was exposed to rays of low 

 refrangibility only, by allowing the light to pass through a cell containing 

 a strong solution of potassium bichromate. 



The exposure was started on July 28, and 50 ccs. of the solutions 

 were tested from time to time with ^^ potassium permanganate. In each 

 case the number of ccs. of the permanganate required to produce a pink 

 tint in 50 ccs. of pure water was determined, and deducted from that 

 required for the same volume of the solution. The following table gives 

 the results of these analyses, the terras ' red ' and ' white ' being used 

 to distinguish between the two sets of experiments. In Nos. 2 and 4 the 

 analyses were made after prolonged periods of cloudy weather, whil.st in 

 7 and 8 the solutions had been exposed to bright sunshine. Nos. 5 and 6 



