86 



NATURE 



\Nov. 30, 1 87 1 



published in this Journal last November, I am prepared to say 

 that I cannot imagine anything more effective and convenient. 



The an-.ingement of Mr. Browning and its extension by Mr. 

 Proctor, are equally effective so far as the adjustment of the 

 prisms is concerned, but are less compact and simple, and do 

 not afford the same facility in charging the number of prisms in 

 use. 



In my instrument the light, after leaving the collimator, falls per- 

 pendicularly upon the face of a half-prism, passes through the train 



=?F^ 



of prisms near their bases ; at the end of the train is twice totally 

 reflected by a rectangular prism attached to the last of the train 

 (which is also a half prism), is thus transferred to the upper 

 story of the train, so to speak, and returns to the view-telescope, 

 which is firmly attached to the same mounting as tlie collimator 

 and directly above it. Both are immovable, and the different 

 portions of the spectrum are brought into view by means of the 

 s-crew, which acts upon the last prism, and through it upon the 

 whole train. The adjustment for focus is by a milled head, which 

 carries the object-glasses of botli collimator and telescope in or 

 out together. Since they have the same focal length, this secures 

 the accurate parallelism of the rays as they traverse the prisms. 



The annexed diagram, taken from the paper already alluded 

 to,* exhibits the plan of tlie arrangement, and requires no ex- 

 planation, unless to add lliat, to avoid complication in the figure, 

 I have represented only two of the radial forks which maintain 

 the prisms in adjustment ; also, that the prisms are connected to 

 each other at top and bottom, not by hinges, but by flat springs, 

 preventing all shake. 



By adding another tier of prisms and sending the light back 

 and forth through a third and fourth story, the dispersion can 

 be easily doubled with very small .additional expense, except for 

 the prisms themselves ; the mechanical arrangements remaining 

 precisely the same. 



I desire, in this connection, to call attention to the great ad- 



vantages gained by tlie use of the half pi ism at the commence- 

 ment of the train, a point which hitherto seems to have escaped 

 notice. 



With a prism of 60", having a mean refractive index, ^, I '6, 

 and placed in its best position, the course of the rays is as shown 

 in Fig. 2. The side <; /' is just 15 times the cross section, a d, 



* After the appearance of the article referred to, I found that Mr. Lockyer 

 had anticipated me by some months, not only in respect to the method of 

 malcing the rays traverse the prism train twice, but also in the use of a half 

 prism at the beginning of the train, and the employment of an elastic spring 

 in the adjustment for minimum deviation. In all essential particulars his in- 

 strument is the same as mine, though in some matters of detail there are 

 differences which have proved to be of practical importance in favour of 

 my own. 



Mr. Lockyer has, however, never printed an account of his instrument, 

 and at the time of my publication I knew only the fact (which I then men- 

 tioned), that he intended to send the light twice through the prism train by a 

 total rctteclion. 



The beautiful instrument recently constructed for Dr. Huggins by Mr. 

 Grubb differs mainly in using compound prisms, and in producing the ad- 

 justment for minimum deviation by an arrangement of link work, which, 

 thou£h not theoretically exact, is practically accurate. 



of the transmitted beam. In other words a prism of the same 

 material and angle described, in order to transmit a beam one 

 inch in diameter, must be one inch high and have sides if inches 

 long. 



But when the light is received perpendicularly upon the face 

 of a half prism, as in Fig. 3, then, since ^'c = /'t--^cos 30°, the 

 length of the prism side, be, recjuires to be only i '155 times as 

 great as the diameter of the transmitted beam. 



Thus a train of prisms each i inch high, and Iiaving the sides 

 of their triangular bases each i '155 inches long, led by an initial 

 half prism in the way indicated, would transmit a beam i inch 

 in diameter, while without the initial half prism the sides would 



have to be i '667 long, the surface to be worked and polished 

 would be I '44 (; c. i (367-^1 '155) times as great, and the quantity 

 of glass required 2'oS {i.e. I ■44-) times as great. With a higher 

 index of refraction the gain is still greater. 



This advantage of course is not obtained without losing the 

 dispersive power of one half prism. But where the train is ex- 

 tensive this loss is comparatively insignificant, and may be made 

 up by a slight increase of the refracting angles. Indeed, in an 

 instrument of the form above described, it is necessary, if the 

 train is led by a liiholc prism, to reduce the refracting angle from 

 60' to about 55", in order that the reflecting prism at the end of 

 the train may not interfere with the collimator, while with the 

 initial half prism the full angle of 60° can be used, so that in this 

 case there is practically no loss whatever. 



It would seem to deserve consideration, whether in the con- 

 struction of spectroscopes to be used with some of the huge 

 telescopes now building, it may not be advisable to carry the 

 principle still further, by employing hM or more half prisms at 

 the beginning of the train in order to economise material and 

 weight. 



Dispersive Efficieney. — The dispersive efficiency of the spectro- 

 scope is its ability to separate and distinguish spectral lines whose 

 indices of refraction differ but slightly ; it is closely analogous to 

 the dividing fo:^'er of a telescope in dealing with double stars. 

 It depends' not only upon the train of prisms, but also upon the 

 focal lengths of the telescope and co'limator, the width of the 

 slit, and the magnifying power of the eye-piece. 



.\s has been said before, each bright line is an iiii.age uf the 



slit whose iiKixiiil/ii/e, referred to the limit uf distinct vision, 

 depends upon the telescope and collimator, but is independent 

 of the prism train. The distanee between the centres of two 

 neighbouring lines, on the other hand, depends upon the number 

 and character of the prisms, the focal length of the telescope, 

 and the magnifying power of its eye-piece, but is totally inde- 

 pendent of the collimator. 



In order that two lines may be divided, it is necessary that 

 the edges of their spectral images should be separated by a certain 

 small distance — a ?)tiiii/nHiii visibile, whose precise value is of no 

 particular importance to our present purpose, but which I 

 suppose to be about ^^ of an inch. 



* It is very common to describe the dispersive power of a spectroscope 

 as being equivalent to a certain number of prisms, or a certain number of 

 degrees from A to H. But either method fails entirely to convey an idea of 

 the appearance of the spectrum in the instrument, and it is much better to 

 name Ihc closest double line which it will divide, or else to give the distance 

 between the two D lines, either linear (referred of course to the limit of 

 distinct vision), or angular. If we know, for example, that the D lines 

 are separated 1% or, what comes to the same thing, appear to be one-si.\th 

 of an inch apart, we have a definite idea of the power of the instrument. 



