369 



Dr Seller further insists that, even when no foreign elastic matter 

 exists in the atmosphere, there is reason to doubt if the specific gra- 

 vity of the air near the earth's surface is uniformly dependent on the 

 general pressure, the temperature, and the humidity. Among the 

 grounds for this doubt, he refers to the vast extent of the atmo- 

 sphere, the infinity of circumstances constantly tending to disturb ita 

 equilibrium, the considerable periods of time required on many occa- 

 sions to restore that equilibrium, if it can be said to have an ordinary 

 equilibrium, and, in particular, to the peculiar laws, in some de- 

 gree antagonistic of gravity and therefore of pressure, observed to 

 affect the distribution of gaseous bodies, whether placed simply con- 

 tiguous to each other, or already in a state of mixture. He concludes, 

 therefore, that the only mode in which any useful result in this sub- 

 ject, either as respects the etiology of epidemic diseases or meteoro- 

 logical phenomena in general, can be obtained, is by following the 

 example of Prout, and determining daily, by a direct process, the 

 weight of some certain volume of air. Neither does he regard the 

 efforts at present making by chemists to detect foreign bodies in the 

 atmosphere by means of chemical tests as necessarily superseding the 

 proposal to determine its daily variations of density by direct means. 

 Dr Seller considers the usual method of weighing air by compar- 

 ing the weight of an exhausted vessel with that of the same vessel 

 filled with air, as involving too much trouble for daily use. He sug- 

 gests that a near approximation to an exact I'esult may be made by 

 observing the difference between the weight of a light body in vacuo 

 and its weight in air ; the former being a constant quantity for every 

 place, while the latter varies in exact conformity with every change 

 which occurs in the density of the air. The larger such a light body 

 is, and the greater the difference of bulk between it and its counter- 

 poise, the nearer is the approximation to an exact result, while there 

 is the less need for extreme nicety in the process. The counterpoise, 

 with the exception of the mere grain weights, should be capable of 

 easy admeasurement ; for example, cubic inches of a heavy metal. 

 The sum of the weights of the body and its counterpoise in vacuo, di- 

 minished by the sura of their weights in air, is to be divided in the 

 ratio of their bulks for the weight of air which each displaces. The 

 weight of a body in vacuo, independently of its weight in air, can be 

 ascertained with precision in proportion as the following data, at a 

 certain temperature, are exactly known, viz., the weight in vacuo of 



