April 1, 1887] 



NA TURE 



533 



traced from the most generalised to the most speciaUsed 

 member of the Klephantidx-. So complete indeed is this 

 transition that not only is there no real line of demar- 

 cation between Mastodon and Elephas, but seveial of the 

 species of the two genera seem to pass so imperceptibly 

 into one another that it is not unfrequently a matter of 

 extreme difficulty (if indeed it be not an absolute impos- 

 sibility) to determine to which species individual teeth 

 really belong. 



In regard to geographical distribution, there appears 

 to be considerable evidence in favour of an easterly 

 migration of the mastodons having taken place from 

 Europe to India ; while the restriction of the stegodont 

 group of elephants to the latter country and the regions 

 to the eastward, points to the conclusion that the transition 

 from the mastodons to the higher elephants took place in 

 those regions. From this we may also infer that there 

 subsequently ensued a westerly migration of these higher 

 forms to Europe, and finally on to North America, where 

 the true elephants did not make their appearance till the 

 Pleistocene, and then appear to have been represented 

 only by two species, one of which ranged over the greater 

 part of the higher latitudes of the northern hemisphere. 

 The descriptive details are very usefully illustrated by a 

 number of woodcuts of the teeth and cranial bones. The 

 work, despite its name of Catalogue, is a most important 

 contribution to our knowledge of the subject 



LETTERS TO THE EDITOR 

 [ Tkt E itor does not hold himself responsible for opinions ex- 

 pressed by his correspondents. Neither can he undertake to 

 return, or to correspond with the writers of, rejected manu- 

 scripts. No notice is taken of anonymous communications . 

 [ The Editor urgently requests correspondents to keep their letters 

 as short as possible. The pressure on his space is so great 

 that it is impossible othenuise to insure the appearance even 

 of communications containing interesting and novel facts. \ 



Iridescent Clouds 



This phenomenon is very common here in the winter, occur- 

 ring, with few exceptions, whenever there are scattered clouds 

 near the sun. The colours are often brilliant enough to catch 

 the attention of the most casual observer, but at other times 

 they can only be m.ade out with the aid of dark neutral-tint 

 spectacles. These reduce the intensity of the glare near the sun 

 to a point favourable to the discrimination of colour. 



I have lately been watching the somewhat complicated phe- 

 nomena, and taking rough measures of the angular distances of 

 the various colours from the sun, and I have little doubt that the 

 colours are due to diffraction of light by fine particles of ice. 



Within a circle, radius about 2', the clouds are white, perhaps 

 faintly tinged with blue ; but it is difficult to observe a delicate 

 shade of colour so near the sun. This circular space is sur- 

 rounded by a ring of yellow or orange. The region of most 

 vivid hues is comprised between 3° and 7°, the most striking 

 being purple, blue, orange, green, and red. These are not 

 arranged in rings, but are distributed over the thinner parts of a 

 cloud in irregular patches. Beyond this region the only colours 

 visible are green and red, becoming fainter as the distance from 

 the sun is increased. I have detected them in three or four 

 cases at a distance of 21°. At some distance from the sun the 

 greens and reds are frequently arranged in bands parallel to the 

 edge of the cloud, sometimes as many as three bands of each 

 being visible. I have often seen a cloud completely encircled 

 by bands, the impression given to the observer being that the 

 colour depends on the thickness of the cloud. No doubt the 

 real explanation is that the ice particles are larger in the interior 

 of a cloud. We have thus two independent factors to determine 

 the colour of a particular portion of cloud : the distance from the 

 sun, and the average size of the particles. Near the sun slight 

 variations of the former are more important, so we get a 

 tolerably regular yellow ring. Far from the sun the variations 

 of the latter have overwhelming influence, and we have bands 

 along the edge of a cloud. At a medium <listance, in the region 

 of vivid colour, we have the two factors nearly equally powerful, 

 and indescribable confusion as the general result. 



On one occasion, when the edge of a large cloud passed 

 almost through the sun, I noted down the colours in order along 

 the edge, where the size of the particles would be tolerably 

 uniform : white, yellow, red ; blue, green, yellow, pink ; green, 

 faint red. This list consists evidently of three successive dif- 

 fraction spectra, and it is in satisfactory agreement with a 

 number of less complete series that I have obtained. The blue, 

 however, is often replaced by a brilliant purple, due to the first 

 and second spectra overlapping. Another method of discover- 

 ing the true order of the colours is by watching the changing 

 hues of a cloud approaching or leaving the sun. This tended to 

 corroborate the list just given, but 1 could seldom trace more 

 than two or three changes, so rapidly did the clouds grow or 

 dwindle away. I noticed, indeed, that the more rapid the 

 growth the more brilliant were the colours displayed. One 

 interesting observation deserves special mention. A cloud 

 showed faint colour at a great distance from the sun. It was 

 edged wiih green, with red inside. As it approached the sun, 

 the bands moved inwards, and red appeared on the edge, then 

 green, then red, then green. The last tint was very undecided, 

 and afterwards the whole cloud remained white. The inward 

 motion of the bands showed that the inner particles were larger. 



We now come to the question of the form of the diffracting 

 particles. The form most favourable to diffraction is the sphere, 

 as with a sphere the angle of diffraction for any given spectrum 

 depends only on its diameter. Thus if a cloud be composed of 

 spheres of uniform size, and be at -the angular distance from the 

 sun corresponding to the first spectrum for that .size, each sphere 

 will send its quota of light to the observer. Next to the sphere 

 comes the circular cylinder. In order that a cylinder may send 

 diftracted light to the observer, its axis must lie in or near the 

 reflecting plane. By the reflecting plane I mean the plane of a 

 mirror which would reflect sunlight to the observer. But, 

 when this condition is satisfied, the angle of diffraction, corre- 

 sponding to any particular spectrum, depends only on the 

 diameter of the cylinder. Any other form, a circular disk for 

 instance, gives different diffracting angles according to the 

 orientation of the particle. The spectra corresponding to 

 different orientations would be superimposed and white light be 

 the result. 



Now, among the manifold forms of snow-crystals there is, I 

 believe, nothing approximating to a sphere. But thin hexagonal 

 prisms are common, either separate, or attached as rays to hexa- 

 gonal disks. These would produce much the same effect as 

 circular cylinders ; for the extreme variation of the apparent 

 diameter of an hexagonal prism from its mean value is only 

 7 per cent. 



Granted that the diffracting particles are hexagonal filaments, 

 my measurements of the angular distances of the colours from 

 the sun supply data for determining the average diameter of the 

 filaments. For this purpose purple is a useful colour, as it 

 always arises from the overlapping of the first two spectra. I 

 took some five-and-twenty measures at various times, which 

 varied from 3^° to 64". These give diameters from '017 mm. to 

 •009 mm. Orange of the first spectrum ranged from 2i' to 5 J°, 

 six measures. These give diameters from '021 mm. to ■oiomm. 

 Blues of the second spectrum, four measures, 4i° to 6J°; diameters 

 •014 mm. to 'cog mm. If the colours I observed at 21° from 

 the sun were produced by filaments '013 mm. in diameter, they 

 must have belonged to the ninth spectrum. But the ninth 

 spectrum, in addition to being only one-fortieth as bright as 

 the first, is overlapped by four spectra of higher order and three 

 of lower, so it can hardly be distinguishable. At such a distance 

 from the sun the finer particles would have a great advantage 

 in colour-producing power, so I think it prob.able the spectrum 

 was of the fifth order, produced by particles of diameter '009 

 mm. 



The next difficulty is to explain why colours are not seen in 

 clouds composed of minute spheres of water. As explained 

 before, the spherical form has a great advantage. I find by 

 calculation that if the spheres were of uniform size, diameter 

 •013 mm., the colours of the first spectrum would be about 

 twenty times as brilliant as if the cloud were composed of fila- 

 ments of the same diameter, arranged at random, but occupying 

 the same fraction of the field of view. So we might n priori 

 expect that water clouds would be more brightly iridescent than 

 ice clouds. But it is not so. During the summer here I looked 

 frequently and vainly for iridescence. This want of colour must 

 arise from the minute drops not attaining sufficient uniformity of 

 size. So we have to find some cause of uniformity which acts 



