©F THE ATMOSPHERE IS SUPPLIED. J87 



the general opinion were juft, it would be fubject to perpetual 



variation. From this fingle fact we infer, that affinity or at- The fupply is 



traction is concerned in keeping up a conftantand regular fup- ^[° e ^ J hcmica j 



ply, and that however unequal at different times the con- affinity j 



fumption may be, that power will frill continue to act in a 



proportionate degree. Several years ago, I conjectured that 



the water in a ftate of folution in the atmofphere, or that which 



compofes the clouds, might be decompofed by the action of 



light, and confequently furniih a fufficient fupply. With this not perhaps 



view I made feveral experiments, by expofing water in glafs ntio°n of aqlfeoua 



veflels to the influence of the rays of the fun, but could never vapor, 



fucceed in producing a fingle bubble of any fort of gas, although 



confined with air of different degrees of denfity, and in vacuo, 



and expofed for feveral months. At that time I attributed the 



want of fuccefs to the difficulty of imitating nature, therefore 



did not abandon the idea entire-ly. 



Meditating on the fubject fome time after, I conceived it but from the 



extremely probable, whatever may have been alledged to the ?jJ?JSi 



contrary, that azotic gas has fome degree of affinity for oxi- withoxigen. 



gen, and that the combined action of this fluid and light might 



be fufficiently powerful to decompofe water. To determine 



this point, I made the following experiment, which fucceeded 



beyond expectation. Into a fmall tranfparent glafs retort, Experiment. 



filled with frefh drawn pump-water, I introduced 17 ounce Azote gas was 



. r r expofed to water 



meafures of pure azotic gas of the temperature or 56°, and j n an inverted 



inverted it into a bottle filled with water, and funk to the neck veffeL 

 in a pot of fand ; the mouth of the retort defcended to within 

 about an inch of the bottom of the bottle. After accurately 

 marking that part of the neck of the retort at which the water 

 flood, it was placed on the outfide of a window fronting the 

 fouth, April 22, 1801. On the firlt three or four days of ex- 

 pofure, a number of very minute bubbles appeared on the fides 

 of the glafs and rofe to the furface. They after this entirely 

 ceafed, and the water became fome what turbid. In about 

 three weeks, having recovered in a great degree its transpa- 

 rency, an infinite number of fmall yellowifh particles were 

 uniformly diffufed through it, and which continued to the end 

 of the experiment. The volume of included gas was per- 

 ceived to be increafed fome time after expofure, and after fome 

 days of uninterrupted funjliine. Some time in the beginning of 

 October, long after the gas had ceafed to gain any additional 



bulk. 



