uf the Problem ojAiinosphnical Refraction. 3S 



nomers ; the greatest being that of Laplace 2026", and the 

 least that of Mayer 1885"; and M. Bessel now makes the 

 same quantity amount to nioie than 2160" by direct observa- 

 tion. Therefore all arguments must lose their force, which 

 suppose that no variation can take place in an element so im- 

 perfectly known. 



But it would be futile to scrutinize minutely the observa- 

 tions of an author whose chief solicitude is to say every thing 

 that will at all make for his purpose. He glides along, heap- 

 ing assertion upon assertion, without much minding the firm- 

 ness of the ground on which he treads *. 



2. In p. 376, I am charged with a mistake and a paradox. 

 The reason assigned is a very extraordinary one. The mis- 

 take is said to arise from correcting the expression of the 

 pressure so as to be = 1 at the surface of the earth, a proce« 

 dure, it is thought, which is not necessary. Now it is clear 

 that the pressure and density, denoted by the symbols y and z, 

 do not mean the absolute, but the relative quantities; that is, 

 the proportions of the pressure and density at any height in 

 the atmosphere to the like quantities at the earth's surface. 

 Therefore, in ascending froni the earth, we necessarily have 

 initially j/ = 1, ;: = 1. 



In the hypothesis of Kramp and Bessel, the rigorous quan- 

 tities are L/ ' '^.i-, 



tj =. r <^ ^'' ' 



but the expression of the density is made more simple ; and 

 the quantities used in calculation are as follows : 



// = r c. 



-,_ . -"-^■> . 



the function for heat being c"''". In order to show how 

 nearly one density is numerically equivalent to the other, 

 M. Bessel gives a table of the values of both computed for 

 the same values of s, extending the calculation to a great 

 height in the atmosphere. 



* It is saiil, p. ^IH, that I had discovered that the I'Vench Tables were 

 computed for the (Vecziiij; temperature. 1 made no iliscovery. The con- 

 struction of the tables is very fully explained by Delambre in the TnbUx 

 Aslronomif/iu'ii, \Hi)(j. 'Hie whole of tlie passage relating to the corrections, 

 p. 37H, is very strange. 



Vol.flr,. '\n. f{2l. ,/>7?/. IWLT). K \ow 



