432 Mr. J. C. McConnel on the 



(3) These figures exhibit a rapid diminution in brightness 

 with increasing i/r, such as is found in observation, but is 

 wanting in the theory of thin plates. 



For the brightness of the first spectrum at 5° we obtain the 

 value 



•000164 a, 



and at the same distance for a cloud of thin plates, 



•0000045 a. 



Though the ice-crystals which fall as snow give us some 

 clue to the forms of those suspended in the clouds, we are 

 ignorant, at any rate, as to which is there the predominating 

 form. But, if we assume that plates and filaments occupy 

 nearly the same fraction of the field of view, these figures 

 show that the colours due to the filaments must far outweigh 

 the others in the near neighbourhood of the sun. So we are 

 in a position, not merely to declare that the observed facts 

 about the brighter colours are inconsistent with the hypothesis 

 of thin plates, but also to show cause why ice-lamina3 should 

 not produce these brighter colours. 



Granting, however, that thin plates are not the principal, 

 they might still be a subsidiary cause. Though incompetent 

 to produce the bright colours near the sun, they might yet 

 give rise to the faint tints at a distance. But I do not think 

 this probable. On days when the colours are visible unusually 

 far out, I have generally noticed several eases all at about the 

 same distance, e. g. on one day 21°, 18-^°, 21°, 21°, 19°, on 

 another 20°, 20°, 22^°. This points to a rapid falling-off 

 of intensity with distance, and therefore to the hypothesis of 

 diffraction. Some slight additional evidence may be drawn 

 from the figures given below. I have thought it probable 

 that the colours at 20° belonged to the fourth spectrum. On 

 this assumption, for -v/r = 20° the brightness is '0000037a, 

 while for any spectrum on the hypothesis of thin plates it is 

 •0000017a. 



The average diameter of the filaments that produce the 

 brighter colours I find to be about *013 millim. It might be 

 thought the chief cause of the brighter tints appearing at 

 about 5° was, that at this angle the colours escape, on the 

 one hand, the overpowering white glare which suffuses clouds 

 in the neighbourhood of the solar disc, and, on the other, the 

 enfeebling influence of greater distance. But I cannot hold 

 this view. There is one characteristic colour, which always 

 occurs at the boundary between the first two spectra, I mean 



