COMPOSITION OF THE ATMOSPHERE. 9 



2 1 '05. This also is strengtliened by the labours of Gay- 

 Lussac, Berthollet, Thomson, Davy, and Humboldt, all 

 men to whom we must attend. Cavendish could not de- 

 cide that the London air differed from that of the country. 

 (See Dr. Wilson's Life of Cavendish.) 



Dalton laboured long on gases j and although his thoughts 

 were generally greater than his experiments when a theory 

 could be found, few men could so persistently labour out 

 the facts where no theory existed. On the composition of 

 the air of Manchester he is moderate, and, it seems to me, 

 just. 



Dumas and Boussmgault in conjunction, and also Brunner, 

 gave results by weighing the oxygen. The air was con- 

 veyed in bottles, and the bottles themselves were filled by 

 sending them free of air, and allowing the air for analysis 

 to flow in. There are objections to this method. It is 

 extremely dependent on the accuracy of the apparatus used, 

 and, if I am at all right in my judgment, the tendency is to 

 diminish the amount of oxygen ; for although the specific 

 gravity of nitrogen does not differ greatly from that of 

 oxygen, it does differ, and is lighter. At any rate I believe 

 the list of analyses given to be a fair representation of the 

 work done. They may be taken as the evidence of the 

 scientific world on the subject. 



Begnault's analyses are very beautiful, and the uniformity 

 of his results predisposes us strongly in their favour. Their 

 number comes in to increase their importance. 



Another reason why I admire the results of Regnault is 

 one of a kind which affects all men, and which may be ex- 

 cused. They agree with my own, made in an open part of 

 Manchester. His results are 20*9 13-20*99 for Paris; and 

 all the numbers he made are above 20*9 until he comes to 

 unwholesome places with putrid waters. A similar reason 

 leads me to believe in the analyses by Bunsen, and those of 

 Dalton made in Manchester, as I found like results when 



