882 



SCIENCE 



[N. S. Vol. XXXVIII. No. 990 



building or the roof of a piazza, certain parts 

 of thin clouds, or edges of thick clouds, will 

 usually be seen tinged with red or green, the 

 colors often appearing together with red pre- 

 dominating. Occasionally the tint will be 

 straw-color or purple. The effect may be seen 

 at any time during the day, preferably when 

 the sun is at a considerable elevation above 

 the horizon. The colors are seldom intense, 

 but are, nevertheless, very beautiful. They 

 may be distinguished, when faint, by com- 

 paring them with any white cloud at an angle 

 of 30° or 40° from the sun. 



As the clouds in question are very brilliant, 

 one's eyes have to become accustomed to the 

 glare before the colors can be seen. Hence it 

 is better to use smoked glass or dark glasses. '^ 

 A smoked glass plate, on which the density of 

 the smoke deposit varies from one edge to the 

 other, is very convenient, as the best density 

 for any particular cloud may quickly be 

 found. 



The following facts indicate that the mech- 

 anism of the effect is totally different from 

 that by which the rainbow is produced. The 

 colors appear in irregular patches of various 

 sizes, and not in arcs of circles concentric with 

 the sun. In fact, two small clouds may be 

 close together, one being colored while the 

 other is pure white. The red and green do 

 not always appear together, the red occurring 

 alone more frequently than the green. The 

 same portion of cloud will frequently change 

 from one color to the other. 



It seems most reasonable to attribute these 

 colors to interference. To make this clear, 

 consider what must happen when white light 

 passes through a water drop or ice crystal. At 

 the surface where the light emerges, the ray 

 will be divided, part passing through, and 

 part being reflected back, to be reflected from 

 the upper, or incident, surface of the drop, 

 thence passing out through the lower surface. 

 This second part will afford interference with 

 the part of the ray that passed through un- 



1 A solution of a substance, having transmission 

 bands in the red and green only, would be beat for 

 observing the colors most frequently seen, namely, 

 red and green. 



reflected, for a certain wave-length, provided 

 a sufScient difference of phase, between the 

 two parts of the ray, has been introduced. 

 Owing to the shape of the drop, or particle, 

 only one particular ray wiU, after undergoing 

 this division, have both these parts sent in the 

 direction of an observer on the ground (just 

 as in the rainbow, each drop behaves like a 

 prism, to an observer, but only for light that 

 passes through one particular plane). If, 

 further, we suppose that there are many drops 

 of very closely the same diameter, then an 

 observer should see light of the same color as 

 that transmitted through a thin film, e. g., a 

 soap film or thin mica, of a thickness equal to 

 this diameter. 



Certain evidence supports the above expla- 

 nation. The phenomenon is especially promi- 

 nent in clouds that are increasing or decreas- 

 ing in density. For example, in one partic- 

 ular cloud that was observed, which was 

 increasing in size, the edge was first red, then 

 green, then gray. Further, a cloud was oc- 

 casionally seen with the red and green ar- 

 ranged in three or four alternate bands, strik- 

 ingly suggestive of Newton's rings, or the 

 fringes produced by an interferometer. 



If the explanation here given is correct, 

 these colors, besides of interest as being pos- 

 sibly the only sky colors produced by inter- 

 ference, may also be of some meteorological 

 importance, namely; in giving an idea of the 

 degree of homogeneity of size of drops in por- 

 tions of thin clouds, by the intensity of the 

 color; of the extent of these portions, by the 

 area occupied by the color, and of the size of 

 the drops, by the particular color present. 

 Perhaps more information could be obtained 

 by a spectroscopic method, whereby the 

 spectrum of a small portion of cloud would 

 show dark bands, corresponding to the wave- 

 lengths removed from the light by interfer- 

 ence. Egbert H. Goddaed 



Worcester, Mass., 

 November 2, 1913 



ORIGIN OF MUTATIONS 



Gates, in a personal letter, has kindly called 

 my attention to a misstatement contained in 



