658 PHILOSOPHICAL TRANSACTIONS, [aNNO 1/03, 



produce, I)\' making use of -,v of the contained arcs for refraction: this height 

 exceeds that in General Roy's table by 31 feet; but we are not certain of its beintr 

 nearer the truth: only it may be remarked, in the table, p. 240 (Phil. Trans, 

 vol. SO,) that between the several stations from High Nook to Botley Hill, the 

 mean refractions are very great. From the reciprocal observations at Leith Hill, 

 Banstead, and Shooter's Hill, the height of the last station is 446 feet, which is 

 the same, in fact, as that obtained in the following manner. General Roy found 

 by levelling, that the floor of the upper story of the Bull Inn at Shooter's Hill 

 was 444 feet above the Gun Wharf at Woolwich; and he allowed 22 feet for 

 the fall to low water at the sea; the sum is 466 feet. In 1794, we levelled from 

 the inn to the station, and found the latter 21 feet lower than the floor, 

 which taken from 406, there remains 443 feet for the station's height. 



Notwithstanding this consistency, and also that in the height of St. Ann's 

 Hill, found by different methods, it is evident from the observations at Dunnose, 

 Rook's Hill, and Butser Hill, that relative heights deduced from elevations, or 

 depressions, cannot always be depended on to less than about 10 feet, even sup- 

 posing those heights are the means of 2 or 3 independent results, except perhaps 

 reciprocal observations were made exactly at the same time. The very great 

 difference in the observed elevations of St. Ann's Hill, proves that no dependance 

 can be placed on single observations. But that was not the only instance; for, 

 at the station on Rook's Hill, we found the depression of the ground at Chanc- 

 tonbury Ring, vary from l' 41" to 2' 30*. The observations however on which 

 the tables are founded, were made in close cloudy days, or toward the evenings, 

 when the tremulous motion in the air is commonly the least. 



It has been conjectured, that the variations in terrestrial refraction, depend on 

 the changes in the atmosphere indicated by the barometer and thermometer: 

 this however, cannot be the case when the rays of light pass near the earth's 

 surface for any considerable distance. Mr. De la Lande, in his Astronomy (Art. 

 Terrest. Ref.,) remarks, that the mountains in Corsica are sometimes seen from 

 the coasts of Genoa and Provence, but at other hours on the same days, they 

 totally disappear, or are lost as it were in the sea. And the late General Roy 

 frequently mentioned an instance of extraordinary refraction, which he and 

 Colonel Caklerwocd observed on Hounslow Heath, when they were tracing out 

 the base. Their levelling telescope at King's Arbour was directed towards Hamp- 

 ton Poor-house, where a flag-stafF was erected at that end of the base; this for a 

 long time they endeavoured in vain to discover, till at last, very unexpectedly, it 

 suddenly started up into view, and so high it seemed to be lifted, that the surface 

 of the ground where it stood became visible. This will appear the more extraor- 

 dinary, when it is considered, that a right line drawn from the eye at King's 

 Arbour to the other end of the base, would pass 8 or 9 feet below the surface of 



