1883.] at Sunset near Trinidad and Rio. 409 



the interval was two minutes fifteen seconds. But on February 7, 

 in Lat. 28°' 5'. S. off the West coast of South America, the Sun 

 setting behind a low bank of clouds reaching perhaps 2' above the 

 apparent horizon, the interval was two miuutes thirty seconds : 

 and lastly, March 11, Lat. 23 0, 23'. S. near Rio Janeiro, the interval 

 was two minutes thirty-one seconds. I cannot however assign so 

 much value to these two observations as to the two others: in the 

 third case on account of the uncertainty of the horizon, and in the 

 last from the fact of my being in my cabin at the time, with the 

 port not more than six or eight feet above the water, the eye on 

 deck being on an average at the height of 20 feet, far more suit- 

 able to give a clear horizon. 



Only in the first case was I able to test the absolute value of 

 the horizontal refraction : in the other examples the uncertainty 

 of my geographical position deprives any calculations of real 

 value: I therefore defer for the moment to consider this point. 

 But if we compare the Sun's diameter with the arc which the 

 Sun's centre would traverse during the interval between the Sun's 

 first and second limb passing the horizon, we get, 



1st case. 2nd case. 3rd case. 4th case. 



Arc traversed : 30'- 0". 32'- 40" 31'- 30". 34'- 45". 



Sun's diam r . 3 2'- 36". 32'- 6" 32'- 30". 32'- 15". 



Difference. 2'- 36". 34"(-). 1'- 0". 2'' 30"(-). 



In a communication to this Society made March 8, 1880, 

 (Proc. Vol. in. 358) describing some observations made by me 

 with a theodolite in Norway on the Sun near the horizon I have 

 said that the " observations with the theodolite gave uniformly 

 the position of the Sun lower on the lower limb than on the 

 higher one " : in which case it is clear that the error would have 

 been less if the amount deducted for refraction had been that due 

 to the position of the Sun's centre and not of the limb observed. 

 But in the cases which I am now discussing, it would seem as if 

 . the Sun actually disappeared in a form differing on each occasion : 

 the retaining power, if I may so express myself, of the increase in 

 refraction between a point at a distance above the horizon equal 

 to the Sun's diameter, and the horizon itself, being variable. 



I desire to draw attention to this point, because I am not 

 aware that the effect of the Sun on Refraction has ever been con- 

 sidered separately : and some observations of the recent Transit 

 of Venus may perhaps have been made on the Sun when so near 

 the horizon that this point may become worth consideration. 

 I have no more to say on what seems to me a very curious fact : 

 except that at Trinidad the actual refraction seemed to me pretty 

 nearly what might be expected. Allowing for a dip on the ship's 

 deck of about 4', or a zenith distance of 90°- 4'. Bar. 30.0. Th. 84°. F. 



