468 



NATURE 



[December 8, 1^921 



(2) That if the surfaces of such a prism be covered 

 with finger marlvS, the radiant spectrum becomes very 

 much more intense. 



(3) That a prism with very scratched or poorly 

 polished surfaces gives an intense radiant spectrum. 



(4) But if now the surfaces be oiled to obliterate 

 the scratches the radiant spectrum is much diminished 

 in visibility. 



If the surfaces of the prism itself are responsible, as 

 the above evidence would seem to show, the question 

 arises : Why should the radiant spectrum always appear 

 to have its achromatic centre at apparently a fixed 

 position in the ultra-violet part of the ordinary spec- 

 trum of the source? 



The following explanation may be advanced : When 

 light passes through an optical surface which is 

 marred by random imperfections, a portion of the 

 light is lost in the formation of a vast number of 

 impure spectra of various dispersions, orders, and 

 meridians, which are oriented about the centre formed 

 by the image of the source. 



In all these spectra the short wave-length rays are 

 less deviated, and therefore appear closer to the source, 

 than do the rays of longer wave-length, which shows 

 that the spectra are produced by diffraction and not by 

 refraction. 



It will be observed further that the spectra are 

 arranged according to their length, short spectra 

 being nearest to the source, long spectra farthest 

 from it. 



If now this congeries of scratch spectra be looked 



5. V. 



I I 



s. 



I 



O 



( — r 



I I 



Fig. I. — " Radiant spectrum" before refractii 





V- G-. (?. 



<5-. 



R . 



Fig. 2.—" Radiant spectrum " after refraction. 

 R, red ; G, green ; V, violet; S, source. 



at by an observer through a prism, the position of the 

 colours in the scratch spectra will be altered : on one 

 side of the source the scratch spectra will be still 

 further dispersed since the dispersion of rays of 

 different length already produced by diffraction is 

 added to by the dispersion of the prism, as shown at 

 A in Figs, i and 2, while on the other side they will 

 be less dispersed, as shown at B, because the 

 dispersion produced by the prism is in the oppo- 

 site direction to that produced by diffraction. But 

 since the scratch spectra vary in length according 

 to_ their angular distance from the source, there 

 will be in a certain position a scratch spectrum 

 the dispersion of which is equal in length but 

 the opposite wav round to that of the prism through 

 which the scratch spectra are being observed ; the rays 

 in this spectrum will thus be recombined to form 

 white light, and this will form the achromatic centre 

 of the apparent "radiant spectrum." The other 

 scratch spectra will from similar reasoning rearrange 

 themselves in order about the achromatic centre 

 according to their length. 



It should follow from this that the achromatic 

 centre should always be beyond the violet end of the 

 spectrum of the source by a distance roughly equal 

 to the length of the spectrum of the source itself. That 

 is, that the displacement of the achromatic centre 

 should vary with the dispersion of the prism and not 



NO. 2719, VOL. 108] 



with its deviation. Experiment shows that this is 

 the case. This explanation obviously covers the case 

 where the optical surface setting up '"scratch " spectra 

 is that face of the prism which is situated farther 

 from the observer. With regard to the near face, it 

 will be seen that the final image presented to the eye 

 is the same, whether diffraction effects produced by 

 the imperfections of an optical surface are after- 

 wards refracted by a prism, or whether the prism 

 first produces the refraction spectrum, which is, in its 

 turn, diffracted, so long as parallel rays are used to 

 illuminate the two systems. But a small bright 

 source situated some distance from the eye produces 

 approximately parallel light, and therefore, if the path 

 difference of optical surface and prism is small, the 

 same effect will be produced whichever is placed 

 nearer the eye. 



The hypothesis that the "radiant spectrum " is pro- 

 duced by diffraction due to imperfections in the prism 

 surfaces, therefore, fits in with all the observations that 

 I have been able to make. But it does not exclude 

 the possibility that the optical surfaces of the eveiadd 

 their effects to those produced by the prism surfaces. 

 Experiment shows, however, that they are verv' slight, 

 which points to the surfaces of the cornea and crystal- 

 line lens being very free from irregularities. 



It remains to consider Brewster's final observations 

 on the "radiant spectrum." He found that the centre 

 of the figure was closer to the direct spectrum bv 

 red light, and farther from it with blue light than it 

 was with green or yellow light. These facts can be 

 readily explained if the differences in the relative dis- 

 persion of the rays in the cases of prismatic and 

 diffraction spectra be considered. Since the red rays 

 in the prismatic spectrum suffer small relative dis- 

 persions, short scratch spectra close to the image of 

 the source will be overfolded, and the centre of the 

 figure will therefore be close to the direct spectrum. 

 Since blue and violet rays in the prismatic spectrun? 

 suffer large relative dispersions long scratch spectra 

 far from the image of the source will be overfolded. 

 and the centre of the figure will therefore be some 

 distance from the direct spectrum. By green and 

 vellow ravs, which suffer medium relative dispersion, 

 the centre of the figure should be a medium distance 

 from the direct spectrum, as Brewstpr found was the 

 case. H. H.artridge. 



King's College, Cambridge. 



Microscope Illumination and Fatigue. 



I w.AS much interested in the letter from Mr. H. J. , 

 Denham on the above subject in Nature of Novem- | 

 ber 17. It is gratifying to know that the importance 

 of adjustment of the intensity of illumination in tl 

 microscope is recognised. At the National Institu'- 

 for Medical Research several workers are provided 

 with appliances for modifying this intensity, and they 

 are agreed that it increases their comfort and ■ 

 efficiency. But I am in disagreement with the method 

 that Mr. Denham adopts. The subject is too lengthy 

 to discuss in full, but it is, I think, an accepted prin- 

 ciple that the dominant wave-length in any light used 

 for the microscope should be as short as possible, and 

 that any light-modifier should reduce intensity without 

 any alteration of aualitv. It is also well known that 

 the wave-length of the dominant radiation is inversely 

 prooortional to the absolute temperature of the 

 radiating substance, which in the case of the " Pointo- 

 lite " lamp is tungsten. To alter the intensity of such 

 an illuminant, therefore, by means of a resistance 

 results in such an alteration of quality that the light 



