of the Sea over the Gulf Stream. 075 



horizon is supposed to be raised by terrestrial refraction, one 

 fourteenth part of the depression due to the spherical figure 

 of the earth ; and the corrections for different heights, rigor- 

 ously computed from the dimensions of the earth, are re- 

 duced, accordingly, in that proportion, in the tables of the 

 most approved authorities. Experience has shown, that in 

 general, when the temperature of the air is colder than that 

 of the surface of the sea, the tabular depressions so computed 

 and reduced, are in error in defect,— and when the air is 

 warmer than the sea, in excess— of the true depression- the 

 proportion of the error to the difference of the temperature 

 being, however, too irregular, and the differences themselves 

 subject to exceptions of too decided a character, to allow any 

 practical rule to be established for a corresponding allowance 

 in correction. So long as the error of the table is confined to 

 a few seconds in amount, its occurrence may be safely disre- 

 garded in all the ordinary purposes of navigation ; but it was 

 a question only to be solved by experience, whether, in cases 

 of an extreme difference between the temperatures of the air 

 and water, the amount of error might not be so considerable 

 as to require attention, especially in deducing a ship's place 

 by chronometrical observations within three hours of noon 

 It was the purpose of having this question tried in the Gulf- 

 Stream, where the sea is frequently many degrees warmer 

 than the air, that Dr Wollaston contrived the dip sector 

 which, from accidental circumstances, had not been applied 

 in its original design until the present occasion. 



The following Table presents an abstract of the observa- 

 tions, (which may be confided in to less than five seconds,) by 

 which it will be seen, that, so far as their evidence can deter- 

 mine, a navigator may be right nine times in ten in appre- 

 hending a tabular error in defect when the sea is warmer than 

 the air; but that, with differences in the temperature of the 

 air and water, frequently amounting to between 10° and 20° 

 and once even so great as 29°, (the sea being always the 

 warmer,) the error of the Tables was not found, even in a 

 single instance, so great as two minutes. 



rations of Captain Sabine, taken from the work quoted in the preceding 

 article, and particularly from their connection with the subject of the 

 l mlf Stream — Ed- 



