Solar Eclipse. 299 



tains, are again transmitted to the eye of the observer. The 

 last is the supposition least exceptionable, yet is obviously far 

 from accounting for all the effects. Besides, were this the real 

 cause, the light would be more universally diffused, as is seen in 

 the new moon, when the whole unenlightened^part of the orb is 

 faintly illuminated by rays reflected in the same manner as is 

 here supposed. 



To apply the facts already discussed, to the explanation of 

 the present phenomenon, let it be observed that the lunar 

 mountains are probably arranged in chains of great extent*. 

 That one instance has been adduced in which reflections of the 

 sun's rays were separated into different streams by ridges appa- 

 rently in this situation ; and that, during the whole eclipse, the 

 luminous appearances previously mentioned, were most con- 

 spicuous near the cusps, simply because in that position the 

 inequalities of the periphery were most advantageously placed 

 for intercepting, and reflecting the solar rays, in directions 

 traversing the moon's disc. One of the larger and more per- 

 manent streams of reflective light, then, passing behind one of 

 these mountainous ridges, in a line. inclining to parallelism 

 with the moon's horizontal diameter, would either be altogether 

 invisible, or but faintly seen, according to the degree of its 

 elevation. If the ridge were continuous throughout its whole 

 extent, no new effect would follow; but if a discontinuity of 

 any considerable length occurred, the illumination becoming 

 then more diffused, Avould present the appearance of an irregu- 

 lar lucid spot, varying in extent, form and brilliancy, according 

 to the different modifying causes. The probable cause has 

 been pointed out why these streams generally appear to pro- 

 ceed from one of the cusps; striking on the elevations at the 

 other extremity of the opening, the rays would again be re- 

 flected, from a well known law in optics, in a direction tending 

 to the opposite cusp, in proportion as the point was nearer the 

 centre of the intervening circumference. Exactly the same 

 effects would follow, if instead of being diffused through an 



* Journal (if' Arts and Sciences, No. 21, p. 38- y. 



