I 



•ii:Uh Remarks on iencsirial HrJ) action. 211 



In calculation, it is customary to divide the observed sum of 

 the refractions equally between the angles of elevation and 

 depression ; but this method, Biot remarks, is in most cases a 

 mere approximation, the error increasing with the extent of 

 the arc. 



So much for theory ; let us next learn how far it has been 

 confirmed by actual experiment. 



Delambre informs us, that he found the refraction subject 

 to great variety, being at one time negative, but states the 

 mean at yl-j. The angles were measured with the repeating 

 circle, which giving the double of the zenith distances cor- 

 rects the error of collimation and irregularities of the level at 

 every pair of observations. 



In England, the refraction as determined by Roy, Mudge, 

 and others, with the great theodolite of Ramsden, seems to 

 have had no limits to its range. The mean as stated in one 

 part is l-12th, in another l-15th; and, if I mistake not, the 

 data in the 3d volume of the Survey will be found upon 

 calculation to determine it equal to 1-17, subject as before to 

 considerable irregularities. 



In India, the observations of Major Lambton, made with a 

 similar instrument, give a mean refraction of about 1-1 6th; 

 not free, however, from very marked exceptions. 



Subsequent operations to ascertain the height of the Hima- 

 laya mountains appeared, when the ray was confined to the 

 highly elevated regions of the atmosphere, to indicate a very 

 sensible diminution of refraction*. This is precisely what 

 theory would teach us to anticipate. Within the limits of 

 2000 or even 3000 feet it would however be scarcely percep- 

 tible, nor will it be sensibly affected by the ordinary changes 

 of the weather. 



Captain Warren has given us the details of his experiments 

 on terrestrial refraction, and seems to attribute its changes to 

 the greater or less degree of moisture contained in the atmo- 

 sphere. 



Should the steel rod of the upper telescope of the theodolites 

 be discovered to be conical, the estimated error of collimation 

 will, if I mistake not, be found incorrect. Were the ends of 

 the rods reversed in position, the angles would in such case 

 differ from the preceding ones. 



To conclude : the observations of Dr. WoUaston across the 

 Thames, tend to prove that when the surface of the water or 

 ground around the observer is unwontedly heated or cooled, 



• Sec Phil. Mag. for 1821, vol. lix. pp. 130 and 190. 



U (I 2 the 



