6o8 



NATURE 



[April 27, 1899 



The resolviii}; power of this combination is ;«/;, just as in 

 the case of ordinary gratings ; and thus, with a battery 

 of twenty elements, each 5 mm. thick (corresponding to 

 w/ = 50oo), the resolving power would theoretically be 

 100,000, which is as high as that of the best gratings 

 available at present." 



Although, as stated above, the resolving power is inde- 

 pendent of the number of plates, being determined by 

 the total thickness, yet for any given purpose the number 

 of elements has to be considered. This is evident when 

 it is remembered how closely the successive spectra 

 follow each other. With a small numtjer of plates the 



-The Michclson Echelo 



(Scale J.) 



overlapping may be so inconvenient as to render the 

 lines unrecognisable, so that, in designing an echelon, 

 this has to be taken into account. If the spectrum to be 

 examined consists of a few sharp lines, the overlapping 

 may be considerable before it causes confusion, and a 

 fe-,u tliick plates will suffice. If, however, the investiga- 

 tion of doubling or widening of special lines in ordinary 

 spectra is needed, then the echelon could advantageously 

 be constructed of a larger number of elements of less 

 thickness to give the necessary resolution. 



The echelon made by Mr. Ililger consists of fifteen 

 plates, each 75 mm. in thickness and 45 

 mm. high, the width (s) of each step being 

 I mm., and the order of spectrum observed 

 being therefore about the 8000th. Its ap- 

 pearance is shown in the annexed cut 

 (Fig. 2), the cell in which it is usually held 

 having been removed. In use the echelon 

 is laid horizontally on the table of an ordin- 

 ary spectrometer provided w ith collimator 

 and telescope, and, as the line of vision is 

 almost direct, it can be pla( ed at once 

 approximately into position. If, however, 

 the slit of the collimator be illuminated 

 with either white light or any source 

 consisting of many radiations, the over- 

 lapping of the successive spectra will 

 simply give an almost white band across 

 the field. To obviate this, it is necessary 

 to only illuminate the slit with light 

 which is sensibly monochromatic. In 

 the case of sodium or mercury vapour 

 there need be nothing more than the 

 source itself, as the light is sufficiently simple in their 

 case, but in general some light-filter will have to be 

 used. This may be done by first passing the light 

 through an auxiliary spectroscope, and focussing the 

 spectrum given by it on the slit of the echelon spectro- 

 meter, thus isolating a small region which will be 

 approximately monochromatic. This arrangement is 

 illustrated in Fig. 3, which shows how the echelon (k) 

 is fitted up in actual work. The form of apparatus for 

 obtaining monochromatic light will depend on the 

 resources of the observer. Obviously a direct vision 



NO. 1539, VOL. 59] 



prism will be most convenient, as then the source of 

 light can be kept stationary while the spectrum is caused 

 to traverse the slit in order to transmit the different lines. 

 If only an ordinary spectroscope with deviating prism is 

 available, then the light source must be secured to the 

 collimator and moved round with it. There is a very 

 simple method, however, of converting an ordinary prism 

 into a direct-vision one, viz. by fixing a plane mirror to 

 the back of the prism. The arrangement is described, 

 by Dr. Fuchs in Zeitschr.fiir Instr Kunde, vol. i. p. 352 

 (1881), and has later been developed by Wadswortb 

 ("Astronomy and Astro-Physics," vol. xiii. p. S44). 

 In this case, all that is necessary is to revolve the 

 prism-mirror combination about the centre of the baclcl 

 of the prism, the axial emergent ray being alwaysl 

 at minimum deviation. This combination is shown f 

 in Fig. 3, 1' being the prism-mirror system between 

 collimator and telescope ; the collimator (c) of the 

 echelon (1;) being then directed axially with the tele- 

 scope of the auxiliary instrument. The observing 

 telescope (t) will be almost directly axial with the col- 

 limator, as the deviation seldom amounts to more than 

 half a degree for the line under examination. 



With the above echelon the " D '' lines of sodium are 

 separated by about sixteen minutes of arc, this being 

 about ten times the dispersion of an ordinary good spec- 

 troscope. The successive spectra are only separated by 

 about two minutes of arc, so that generally two orders of 

 spectra are visible in the field together ; in practice the 

 echelon can be slightly inclined, thus reducing the in- 

 tensity of one of these almost to zero, so that measure- 

 ments can be made on the remaining one with less risk 

 of confusion. 



One is surprised at the first experience with the instru- 

 ment by the lirightness of the spectra, considering the 

 enormous dispersion and the path traversed by the light. 

 This is no doubt partly due to the incidence on the plates 

 being very nearly normal, thus diminishing the losses 

 by reflection. 



In making the plates it was a very delicate matter to 

 obtain all of exactly the same thickness. This was 



done by first producing a large plane-parallel plate, and 

 then cutting out the required pieces from it. In the mak- 

 ing of this pl.ite the ordinary tests for plane parallelism 

 were found quite inadequate, and every part of it was 

 tested by viewing the interference bands produced 

 between the two surfaces while illuminated by a parallel 

 beam of monoc hromatic light. Each elementary section 

 of the surfai e was then refigured until the interference 

 patterns were similar throughout. This will give some 

 idea of the patience and skill required to produce a 

 successful result. 



