horizontal RefraSlion of the Air. 59 



I do noi find that I examined the appearance of the cliff AB, 

 and its image ab', which, had there at that time been any 

 ftriking marks in them, would have determined whether the 

 objedl and its image were uf the fame magnitude. The 

 image a h was, however, ereft ; the boundaries on the top 

 of AB and a b agreeing together. Having examined this for 

 fome time, and taken a drawing of the appearance, during 

 which I could difcovcr no variation, I directed my lelefcope to 

 other objects; and, upon turning it again to the fame cliffs, 

 after the fpace of about fix or feven minutes, the images a b 

 and ViV were vaniflied; but, examining them again foon 

 after, the images were again vifible, and in every refpefil the 

 fame as they appeared before. A fhort time after, they dif- 

 appeared, and did not appear any more. 



Soon after the above appearances, I obferved a fliip C, 

 with the hulk below the horizon xy, pafl^ing by the fame 

 cliffs A B ; an inverted image D of which appeared againft 

 the cliffs, as reprefented in Fig. 6. The fliip was in motion, 

 and remained at the fame dillance on the other fide of the 

 horizon : I continued my obfervalions upon it till it had 

 paffed the cliffs for a confiderable diftance, but there was na 

 change of appearance. The cliffs were illuminated by the 

 fun, and appeared very difi:in6lly j but there was no image 

 above, as in the laft cafe. 



Continuing to obferve the fame cliffs AB, Fig. 7, I foon 

 after difcovered two partial elevations, vi, n, of the fea, by 

 the unufual rcfraftion ; they changed their figures a little, 

 and difappeared in the place where they firll appeared, and 

 were equally diftinft in every part. 



About this time, I obferved a very thick fog coming upon 

 the horizon from the other fide, rolling upon it wi.th a pro- 

 digious velocity ; curling as it went along, like volumes of 

 fmoke fomclimes out of a chimney. This appeared feveral 

 times. 1 conclude, therefore, that there was 3, confiderable 

 fog on the other fide of the horizon. 



The lall phenomenon which I obferved was that which is 



reprefented in Fig. 8.j where xy reprefents the horizon, ab 



two partial elevations of the fea, meeting at c, and conti- 



I 2 niied 



