1880.] Northern Sea-horizon. 355 



It may be added here that the barometer, an aneroid, stood 

 throughout at 30'25. My thermometer, unfortunately, was packed 

 up, and, I think, not accessible ; but I am sure that the tempera- 

 ture was not far from 50 F. During the preceding six weeks I had 

 repeatedly noted it between 11 P.M. and 1 A.M. at Vardo: this, and 

 my constant habit of observing it at home enable me to say so 

 much with confidence. 



The Tables of refraction used in order to obtain the reduced or 

 true position of the Sun's centre, corresponding to the observed 

 altitude of his limb, are those of Ivory, which extend to a zenith 

 distance of 90". Bessel's Tables are not, strictly speaking, com- 

 puted so far, but only to 85" ; the supplement to them, for zenith 

 distances between 85" and 89^°, is said by him, in the Preface to 

 his Tahidce Regiomontaiioe, to depend considerably on observations 

 of stars taken, not on the meridian, but when setting, on the 

 western hoi'izon, where it may be fairly said that their position is 

 not certain, but constructive \ It is clearly however of little 

 importance which Tables are used, as for 85" the two tables give 

 the same result within a second : for 89" 11' 40" (for example) Th. 

 50" F., Bar, 30"25, the refraction on Ivory's Tables is 26' 14"; on 

 those of Bessel 26' 44" for 89" 28' 45", Ivory gives 28' 51", Bessel 

 29' 31". Whatever therefore may be the comparative merits of 

 these two authorities, it is clear that they will not practically affect 

 my own results. 



The main feature to be noticed in these observations is that 

 although the observed point of the Sun while more than a degree 

 above the visible sea-horizon does not seem to have been more 

 seriously displaced than might have been expected considering the 

 uncertainty of the ship's position and the circumstances under 

 which the sights were taken, still that below that altitude, 

 and especially when the ship's latitude becomes almost absolutely 

 certain on account of its proximity to so marked and known a 

 point as the North Cape, the errors become very large indeed, 

 and uniformly of one type, i.e. they invariably make the refraction 

 less than it ought to be, the Sun's disk as observed being always 

 lower, at the extreme point by as much as eleven minutes, than is 

 required by the theories of the writers on this question. One 

 simple consideration will prove that the phenomenon, as observed 

 could not have been in any way illusory : it will easily be found 

 that in lat. 71" 10', with the Sun's declination taken at 18" north, 

 very nearly what it was on the evening in question, at intervals of 

 5, 10, 20, 40 minutes, before or after the Sun's passing the 



1 By this I mean that the Decliuation, and from that the refraction of any 

 celestial object is, as a rule, determined from its altitude on the meridian. But in 

 other parts of the sky, its proper altitude must be ascertained by geometry^ i.e. is 

 constructive. 



