Nov. 30, 1871] 



NATURE 



85 



SPECTROSCOPIC NOTES* 



On the Conslnidion, Arrangement, and best Proportions of the 

 Instrument with reference to its efficiency. 



"TTHE spectroscope consists essentially of three parts — a prism, 

 -*- or train of prisms, to disperse tlie liglit ; a collimator, as it 

 is called, whose office is to throw upon the prisms a beam of 

 parallel rays coming from a narrow slit ; and a telescope for 

 viewing the spectrum formed by the prisms. 



Supposing the slit to be illuminated by strictly homogeneous 

 light, the rays proceeding from it are first rendered parallel by 

 the object-glass of the collimator, are then deflected by the prisms 

 and finally received upon the object-glass of the view-telescope, 

 which, if the focal lengths of the collimator and telescope object- 

 glasses are the same, forms at the focus a real image of the slit, 

 its precise counterpart in every respect except that it is somewhat 

 weakened by loss of light and slightly curved, t 



If the focal length of the view-telescope is greater or less than 

 that of the collimator, the size of the image is proportionally in- 

 creased or diminished. 



This image is viewed and magnified by the eye-piece of the 

 telescope. 



If now the light «'ith which the slit is illuminated be composite, 

 each kind of rays of different refrangibility will be differently 

 reflected by the prisms, and form in the focus of the telescope 

 its own image of the slit. The series of these images ranged side 

 by side in the order of their colour constitutes the spectrum, 

 which can be perfectly pure only when the slit is infinitely 

 narrow (so that the successive images may not overlap), and 

 accurately in the focus of the object-glass of the collimator, which 

 object-glass, as well as that of the telescope, must be without 

 aberration either chromatic or spherical, and the prisms must be 

 perfectly homogeneous and their surfaces truly plane. 



Of course, none of the conditions can be strictly fulfilled. An 

 infinitely narrow slit would give only an infinitely faint spectrum ; 

 and no prisms or object-glasses are absolutely free from faults. 

 A reasonably close approximation to the necessary conditions 

 can, however, be obtained by careful workmanship and adjust- 

 ment, and it becomes an important subject of inquiry how to 

 adapt the different parts of the instrument to each other so as 

 to secure the best effect, and how to test separately their excel- 

 lence, in order to trace and remedy as far as possible all faults of 

 performance. 



With reference to the battery of prisms, several questions at 

 once suggest themselves relative to the best angle and material, 

 the number to be used, the methods of testing their surfaces and 

 homogeneity, and the most effective manner of arranging them. 



Angle and Aiatcrial of the Prisms. — As to the refracting angle, 

 the careful investigation of Prof. Pickering, published in the 

 American Journal of Science andArtiox May 186S, puts it beyond 

 question that with the glass ordinarily employed an angle of about 

 60° is the best. For instruments of many jjrisms there is an 

 advantage as regards the amount of light in making the angle 

 such that the transmitted ray at each surface shall be exactly per- 

 pendicular to the reflected. For ordinary glass, the refracting 

 angle determined by this condition somewhat exceeds 60° ; for 

 the so-called "extra-dense" flint it is a little less. 



The high dispersi\'e power of this " extra-dense " glass is cer- 

 tainly a great recommendation. But it is very yellow, power- 

 fully absorbing the rays belonging to the upper portion of the 

 spectrum, and is very seldom homogeneous. It is so soft also, 

 and so liable to scratch and tarnish, that it can only be safely used 

 Ijy casing it with some harder and more permanent glass, as in 

 the compound prisms of Mr. Grubb, and the direct vision prisms 

 of many makers. 



For many purposes these direct vision prisms are very con- 

 venient and useful, but they are hardly admissible in instruments 

 of high dispersive power designed to secure accurate definition of 

 the whole spectrum, the violet as well as the yellow. 



* By C. A. Young, Ph.D., Professor of Natural Philosophy and Astronomy 

 in Dartmouth College. Reprinted from advance-sheets of the Journal of the 

 Franklin Institute, by permission of the Editor. 



t The curvature arises from the fact that the rays from the extremities of 

 the slit, though nearly parallel to each other, make an appreciable angle 

 with those which come from the centre. They therefcre strike the surface 

 of the prisms under different conditions from the central rays, and are 

 differently refracted, usually more. The higher the dispersive power of the 

 instrument and the shorter the focal length of the collimator, the greater 

 this distortion, which is also accompanied by a slight indistinctness at the 

 edges of the spectrum. 



Test for Flatness of Surface. — For testing the flatness of the 

 prism surfaces, probably the best method is to focus a small 

 telescope carefully upon some distant object (by preference the 

 moon or some bright star), and then to scrutinise the image of 

 the same object formed by reflection from the surface to be tested. 

 Any general convexity or concavity will be indicated by a 

 corresponding change of focus required in the telescope ; any 

 irregularity of form will produce indistinctness, and by using a 

 cardboard screen perforated with a small orifice of perhaps ^ 

 inch in diameter, the surface can be examined little by little, and 

 the faulty spot precisely determined. 



Test for Homogeneity. — It is not quite so easy to test the homo- 

 geneity of the glass. Any strong veins may, of course, be seen 

 by holding the prism in the light, and if the ends of the prism 

 are polished, the test by polarised light will be found very 

 effective in bringing out any irregularitiesof density and elasticity 

 in the glass. A blackened plate of window glass serves as the 

 polariser ; a Nicol's prism is held in one hand before the eye 

 in such a position as to cut off the reflected ray, and with the 

 other hand the glass to be tried is held between the Nicol and 

 the polariser. If perfectly good it produces no effect whatever ; 

 if not it will show more or less light, usually in streaks ancl 

 patches. 



On the whole, however, the method of testing which has been 

 found most delicate and satisfactory is the following : — 



A Geissler tube containing rarefied hydrogen is set up verti- 

 cally, and illuminated by a small induction coil. 



A small and very perfect telescope of about six inches focus is 

 directed upon it from a distance of seventy-five or one hundred 

 feet, and carefully adjusted for distinct vision. 



The prism to be tested is then placed in front of the object- 

 glass of the telescope with its refracting edge vertical, adjusted 

 approximately to the position of minimum deviation, and tele- 

 scope and prism together then turned (by moving the table on 

 which they stand), until the spectrum of the tube appears in 

 the field of view. This spectrum consists mainly, as is well 

 known, of three well-defined images of the tube, of which the 

 ;v(i' image, corresponding to the C line, is the brightest and best 

 defined, and stands out upon a nearly black background. 



Supposing then thefatncss of the prism surfaces to have been 

 previously tested and approved, the goodness of the glass may 

 be judged of by the appearance and behaviour of this red image ; 

 and by using a perforated screen in the manner before described, 

 inequalities of optical density are easily detected and located. 

 Irregularities, which would hardly be worth noticing in a tele- 

 scope object-glass, where the total deviation produced by the 

 refraction of the rays is so small, are fatal to definition in a 

 spectroscope, especially one of many prisms, and it is very 

 difficult to find glass which will bear the above-named test with- 

 out flinching. Of course it must be conducted at night, or in 

 a darkened room. 



Number and Arrangement of Prisms. — The number of prisms 

 to be employed will depend upon circumstances. If the spectrum 

 to be examined be faint, and either continuous or marked with 

 dark lines, or by diffuse bands, either bright or dark, we are 

 limited to a train of few prisms. 



The light of the sun is so brilliant that, in studying its spec- 

 trum, we may use as many as we please. The light is abundant 

 after passing through 13, and I presume would still be so if the 

 train were doubled. 



Spectra of fine well-defined bright lines also bear a surprising 

 number of prisms. The loss of light arising from the trans- 

 mission through many surfaces is nearly, if not quite, counter- 

 balanced by the increased blackness of the background, and the 

 greater width of slit which can be used. 



As to the best arrangen.ent for the prisms, this also must be 

 determined by circumstances. 



Where exact measurements are aimed at, as, for instance, for 

 the purpose of ascertaining the wave-length of lines, or the dis- 

 persion co-efficient of a transparent medium, the prism or prisms 

 ought to be firmly secuied in a positive and determinable relation 

 to the collimator. A train of many prisms can hardly be safely 

 used in such work on account of the difficulty in obtaining this 

 necessary fixity, and if high dispersion is indispensable, it can 

 only be obtained by enlarging the apparatus. 



But for most purposes it is better that the prisms, instead 01 

 being fixed, should be mounted upon some plan which will 

 secure their automatic adjustment to the position of minimum 

 deviation. 



Having now thoroughly tried the plan which I proposed and. 



