160 Dr. Dal ton on the Constitution of the Atmosphere, 



In the first three experiments no oxygen was found in the 

 residuary gas; in the fourth a trace of oxygen was found; and 

 in the fifth and sixth, considerable quantities of oxygen were 

 found in the residues. 



It is obvious that Mr. Cavendish began intentionally with 

 an overdose of hydrogen, probably expecting the diminution 

 to be a constant quantity till the hydrogen became deficient, 

 and then of course the diminution must be lessened; this was 

 not the case exactly; but the reason is easily discovered, and 

 it proves the accuracy of the observations. 



Hydrogen gas is rarely obtained quite pure : it frequently 

 holds two or three per cent, of common air, detached from 

 the water through which it bubbles and by other means; this 

 air increases as more water enters the hydrogen bottle, till 

 sometimes it amounts to ten per cent, at the last, as every one 

 knows who has had a due share of experience. Now as Mr. 

 Cavendish does not mention the purity of his hydrogen, we 

 must try it by the means now generally known, as the re- 

 ported results will guide us in the investigation. 



On looking at the column headed " diminution on firing " 

 it is easy to see there is a discrepancy in the first three ex- 

 periments in that column; if the hydrogen used contained any 

 oxygen the diminution on firing ought to have continually 

 decreased, whereas it was greater in the third than in the se- 

 cond experiment. This it must be allowed is a proof of in- 

 accuracy in one or both of the experiments ; but it is no greater 

 error than usually occurs if we trust to a single experiment 

 with any gaseous mixture. The average of two or three ex- 

 periments on mixtures of the same proportions should be 

 taken. The fourth experiment clearly shows that the hydro- 

 gen contained oxygen as well as azote ; for a diminution of 

 61*2 would denote the union of 20-4' oxygen with 40*8 hydro- 

 gen ; hence there must have been 1*5 common air in the hy- 

 drogen. I have formed the column "amendment" by as- 

 suming the hydrogen in ail the experiments to contain 4^ per 

 cent, common air. If we combine the results of the third 

 and fourth experiments, either by assuming Mr. Cavendish's 

 diminution or that of the amendment, we shall obtain a very 

 good approximation to the quantity of oxygen in atmospheric 

 air, the former experiment giving too great diminution by 

 reason of the excess of hydrogen and that containing some 

 oxygen, and the latter giving too little diminution for want of 

 the requisite quantity of hydrogen ; the former will give 

 20-98 per cent, oxygen, and the latter 20*92 per cent, oxygen 

 in atmospheric air. If any doubt should remain as to Mr. 

 Cavendish's hydrogen containing oxygen, it is removed by 



