MEASUREMENT OF HEIGHTS. 529 



of the transition of the one to the othei- is not known, vie 

 are compelled to decide arbitrarily, and it seems to me most 

 suitable to take the mean of the two values of a to be applied in 

 the calculation of the formula (11.). 



5. 



It appears to me needful to examine more closely the dif- 

 ferent suppositions by means of which I have obtained the for- 

 mula (11.)' The first assumption in all researches relating to 

 the pressure and density of the atmosphere at undetermined 

 heights, is that of its equilibrium. That this is not strictly cor- 

 rect, is not now said for the first time. Its incorrectness is shown 

 both in the oscillations of the barometer around its mean height 

 at each point of the earth, and in the difference of this height 

 at different points strictly at the level of the sea. The know- 

 ledge of this difference was first obtained by an investiga- 

 tion by Adolphe Erman in 1831*, in which he showed, partly 

 from his own observations made in his travels round the earth, 

 and partly from the observations of others in Northern Asia 

 and America, and on board the Russian corvette Ki'otkoi com- 

 manded by Captain Hagemeister, first, that in the zones of 

 the trade winds, the barometer stands higher at the boundary 

 most distant from the equator than at the boundary w hich is 

 nearest to it ; and secondly, that the mean height of the baro- 

 meter is different in different meridians. The first result rests 

 on observations collected in passing eight times through the 

 zone of the trade winds ; and has since been coiToborated in 

 Herschel's astronomically-memorable voyage to the Cape of 

 Good Hope. The second result rests on a comparison of ob- 

 servations made in the Atlantic and Pacific Oceans ; the differ- 

 ences amount to several lines, and leave no doubt that the mean 

 height of the barometer at the level of the sea is different at 

 different points of the earth's surface, and depends on the geo- 

 graphical latitude and longitude of the place. 



The oscillations of the barometer, which may be regarded as 

 accidental, must cause single barometrical determinations of a 

 difference of elevation to deviate fi'om the mean of several de- 

 terminations ; but the mean diversities, which depend on the 

 longitude and latitude, if not known, must produce errors, 

 which will not disappear in the mean even of many observations, 

 • roj^gendorfr, Ann. der Pliysik, vol. >.cix. (.\xiii. of the new series), p. 144. 



