VOL. XC.] PHILOSOPHICAL TRANSACTIONS. 67 1 



in which case, as the rays enter from a vacuum, the greatest possible effect of the at- 

 mosphere might be expected. But it must be remembered, that when a star appears 

 in the horizon, its rays intersect the superior strata of the atmosphere at an inclina- 

 tion of several degrees, and that they pass but once through the variations from 

 rarity to density; but, on the contrary, that in the experiments with red-hot iron, 

 the rays may pass actually in the direction of the strata, and that they are re- 

 fracted, not only by their entrance from the denser into the rarer medium, but 

 the effect is doubled, since the refraction caused by their emergence is equal to 

 that produced by their incidence. 



Though a stratum of air, heated by these means to so great a degree, affords 

 an erect, as well as an inverted, image of objects seen through it, the more 

 moderate warmth communicated to it from bodies heated by the action of the sun 

 on them, seems insufficient to produce both images; but the inverted image may 

 generally be seen, when the sun shines on a brick wall, or other dark-coloured 

 surface. While the eye of the observer is placed nearly in a line with the wall, if 

 another person, at 30 or 40 yards distance, extends any object towards the wall, 

 an image similar to it will appear to come out to meet it. It would be difficult to 

 ascertain with accuracy the degree of rarefaction capable of showing this appearance, 

 but it may be of some use to future observers, to mention the different degrees of 

 heat which I observed. In one instance, a thermometer in contact with the wall, 

 stood at g6°; but, at •§- of an inch distance, 82°. One morning, when the sun 

 shone bright, I examined the temperatures and refraction produced at the surface 

 of a deal bar painted green, about 8 feet long. A small thermometer in contact 

 with the bar, rose to o6°; at -± of an inch distance, it stood at 73°. The re- 

 fraction at the same time exceeded 20 minutes. 



To explain why red-hot iron occasions 2 images, while solar heat produces but 1, 

 I imagine that the intense heat in the former case rarefies the air for some small 

 distance uniformly, and thus affords the same series of variations as between other 

 fluids of uniform density ; but that, in the latter, the heat is conveyed off as fast 

 as it is generated ; so that, as there is no extent of medium uniformly rare, the 

 densities corresponding to the concave portion rm, fig. 6, of the curve before 

 described, do not take place, but the phenomena occasioned by the convex part 

 md are alone produced. It must be remarked, that the vertical position of the 

 surface contributes greatly to increase the effect; for since the heated air rises in 

 the direction of the surface, its ascent has in this case no tendency to blend it with 

 the adjacent denser strata, and hence very different degrees of density take place in 

 the thickness of ^ of an inch ; so that, as the increments of density are great, the 

 refractions will be proportionally so ; but where the heated surface is horizontal, the 

 ascent of the rarefied air into the superincumbent denser strata renders the variations 

 far more gradual; consequently a heated surface of far greater extent must be re- 

 quisite, to produce equal refraction. 



However, over extensive plains, when the sun shines,, some degree of inversion 



