^70 PHILOSOPHICAL TRANSACTIONS. [ANNO 1800. 



the truth of the 2d proposition appeared capable of confirmation by experiment* 

 for the deviation of a ray is there said to depend on the increment of density and 

 time of the ray's passage, jointly; accordingly the deviation caused by a given 

 increment should be in proportion to the extent of the medium. 



In order to try what effect a greater extent of medium would produce. 



Eocper. 3. I made a rectangular glass vessel, of which the sides were in the 

 ratio of 10 to 30.6; and, having put into it some clear syrup, with water on its 

 surface, I measured the greatest refractions through it in both directions, and 

 found them in the ratio of 10 to about 29. In another vessel, of which the sides 

 were as 10 to 40.4, the refractions were, on an average, as 1 to 4. Being now 

 fully satisfied of the effect of different fluids, I made the following experiment, by 

 which it appears, that the variations of density occasioned by difference of tem- 

 perature between adjacent strata of the same fluid, follow a similar law. 



Exper. 4. Having put some cold water into a square vessel, I covered its surface 

 with a piece of writing paper perforated with a few small holes, and then filled 

 the vessel cautiously with boiling water. The paper nearly prevented any mixture 

 of the hot and cold water; but, by floating gradually up, left them to communicate 

 their heat by contact alone. While they were in this state, I examined the ap- 

 pearance of remote objects through the varied medium, and found, that when my 

 eye was removed 4 or 5 feet from the vessel, the effects were the same as in the pre- 

 ceding experiments with different fluids; above any object viewed through the cold 

 water, I could distinguish 2 images of it, the one inverted, the other erect, as 

 usual ; but these appearances did not continue more than 5 or 6 minutes. 



Having thus established, by experiments sufficiently varied, that the contiguity 

 of 2 fluids of unequal density is capable of occasioning all the appearances that 

 have been observed, I shall proceed to show by what means the air may be made to 

 exhibit similar phenomena. 



Exper. 5. I heated a common poker red-hot, and held it so as to look along the 

 side of it, at a paper 10 or 12 feet distant. The rarefaction occasioned by it caused 

 a perceptible refraction, to the distance of about £ of an inch from the side of the 

 poker. A letter seen more distant from it appeared as usual ; within that distance 

 there was a faint image of it reversed; and still nearer to the poker was a 2d image 

 direct, and as distinct as the object itself, but somewhat smaller, as in fig. 9, in 

 which a section of the atmosphere surrounding the poker is represented. At the 

 bottom and sides it is nearly circular; but upwards the circular form is lost in un- 

 dulations, occasioned by the rapid ascent of the rarefied air. The greatest de- 

 viation produced in this case measured about 4 a degree. 



Exper. 6. By a red-hot bar of iron, 30 inches long, the refractions were much 

 greater, the extreme deviations amounting to full 1^- degree. 



The refractions observed in these experiments may, at first view, be thought 

 greater than could be caused by difference of temperature alone, being in one in- 

 stance more than double the greatest horizontal refraction of the heavenly bodies; 



