21 



It is pretty clear that tlie suspeusion over the weight of atmo- 

 sphere of impalpable dust in the rarified regions may be prolonged 

 for an indefinite period. Not that it has no tendency to fall. 

 On the other hand, I conceive, as I have already hinted, 

 that it is subject to great alterations of level, depending pro- 

 l)ably on barometric pressure, which may intensify or minimise 

 the " sun-glow ;" though, as I shall afterwards show, this con- 

 nection is hardly determined. The upward currents in times 

 of low barometer at the earth's surface, when — allowing for 

 differences of altitude — a comparatively high or accumulated 

 pressure (as I have already shown) seems to prevail aloft, would 

 buoy it up, as an upward current does a feather or ordinary motes 

 of dust; and it would keep playing over the tops of the upward 

 currents on a plan analogous to the phenomena of the so-called 

 *' willow-leaves" on the solar photosphere which are reasonably 

 supposed to be the tops or crests of the currents in the sun. 

 If this is so, the objection to floating dust even higher than 

 40 miles — the traditional height of the twilight-producing 

 atmosphere — is overcome. Again, according to Faraday, 

 metallic gold, when minutely divided, required months to sub- 

 side when suspended in water ; and on this experiment I see 

 no reason why the dust-powder should not remain suspended 

 over the haze of vapour. But if objection is raised to these 

 explanations of the suspension, we have the repulsive agency 

 of electricity to account for it, as has already been suggested, 

 assuming that the dust, if from Krakatoa, is negatively electri- 

 fied. The particles would mutually repel each other, as in the 

 case of electrified gold-leaf, and be repelled and buoyed up by 

 the repulsive action of the earth. 



I have endeavoured to meet the chief objections to the dust 

 theory, and will now proceed to a few words in conclusion on 

 the atmospheric or possible secondary causes of the " sun- 

 glows." 



My meteorological and spectroscopic observations in con- 

 nection with the glow are very conflicting, and further discus- 

 sion of them is necessary before I can see my way to establish a 

 connection. That the glows have become intermittent, and 

 have been observed to be less intense during times of low 

 barometer, seems certain. This may be explained by supposing 

 an abundance of absorbing vapour, meteorologically derived, 

 which would be increased by the primary stratum of dust 

 favouring condensation. So the red light of less refrangibility 

 would be impeded or cut off and partly absorbed by the 

 vapour of a low-pressure area, and we might expect the " red- 

 glow" to be not so intense, while the more refrangible green 

 and blue tints of the sunset would not be interfered with. 

 I think Mr. Todd has also observed this minimum of intensity 



