METEOROLOGY. o83 



4. It is not to be denied that the phenomena in general are due lo 

 the refraction of the sun's rays in the aqueous vapor of the atmosphere ; 

 but in order to explain the superposition of color we must also con- 

 sider tlie reflection of colored rays from exceedingly small bodies at an 

 extraordinary height in the atmosphere. This does not constitute inter- 

 ference phenomena. [It has been called, very appropriately, selective 

 reflection.] 



5. The ordinary twilight is, to a certain extent, such a phenomenon 

 of reflection ; the astronomical twilight disappears at the horizon when 

 the sun is about 18° below that plane from which we compute the alti- 

 tude of tlie layer of haze producing it ; but if the duration of twilight 

 increases we must conclude that the layer of haze is i)roportionately 

 higher. Kroue'.s observations, however, in December, 1883, and Janu- 

 ary, 1884, seem to show that occasionally the twilight did not disappear 

 through the whole night, but that a reddish tint was visible in the hori- 

 zon and even half way to the zenith. That this reflecting material can 

 possibly be the finest dust from the volcanic eruption at Krakatoa, is, he 

 thinks, entirely improbable ; neither can it be due to the presence of ice 

 spiculcie, but it must be minute particles of water. Owing to the low 

 temperature in the upper regions, this water can only be frozen, but we 

 know nothing as yet as to the form of the particles of such very cold 

 aqueous vapor. (/>. M. Z., i, p. 277.) 



390." [The examination of certain thin hazy clouds that covered Paris 

 during January, 1882, was made by Fonvielle, who rose up into them in 

 his balloon, and found that they were formed of the minutest spheres of 

 water frozen at a temperature many degrees below freezing; the forma- 

 tion of the cloud was apparently due to radiation, as it stayed in the same 

 locality for several weeks. The atoms of water were quiescent, but if 

 set in motion among themselves crystallized into mi nute spiculte. Dufour 

 has shown that minute drops may be cooled to — 20^ 0., so long as the 

 surface tension is kept large without crystallizing or even solidifying; it 

 is therefore ])lausible that the twilight phenomena of 1883-'84 were due 

 to selective reflection and diffraction by aqueous spherules of the finest 

 size, such as must always exist to a greater or less extent in the upper 

 atmosphere. If the particles, whether dust or vapor, were electrified 

 before or while being carried up to the height of several miles, they 

 would in the thin air of that region not only be carried along by its 

 currents, but by their own mutual repulsion would tend to rise still 

 higher, forming an appendage to the earth, and reminding one of the 

 ascent of an envelope from the surface of the nucleus of a comet, where 

 the particles are distinctly seen to ascend on the hot side toward the 

 sun and then by mutual repulsion, as shown by G. P. Bond and other 

 astronomers, flow away from the sun.] 



391. Among the principal items relating to the sunsets and Kra- 

 katoa, we may mention the following : 



Lockyer's theory of the volcanic dust. {Nature^ xxix, pp. 148-174.) 



