468 



NA TURE 



[Oct. 8, 1874 



scope of the spectroscope. To make our meaning clear, 

 let us suppose that the slit is illuminated with a sodium 

 flame, then the dispersive power of the prisms will pro- 

 duce in front of the eyepiece two images, or " lines," and 

 with the same lenses the distance of their centres will 

 depend upon the prismatic power ; but it is clear that if 

 the slit be widened, the two images will eventually widen 

 vmtil they touch each other or overlap. There is, then, 

 the same dispersion, but less separation, than when we use 

 the narrow slit ; and it would follow from this that with an 

 almost indefinitely small slit a prism of very small dis- 

 persion would give two separate images of a sodium- 

 illuminated slit, which could be magnified so as to have 

 their distance and width the same as would be given by 

 using a wider slit and greater prismatic dispersion ; but 

 with an eyepiece of the large power required, the lines 

 would be so diminished in brightness as to preclude this 

 arrangement ; and in order to see a spectrum as brilliant 

 as possible, the eyepiece ought to be as low in power as 

 possible consistent with reducing the cylinder of rays 

 sufficiently small that they all enter the lens of the eye. 



Let us now consider a spectroscope of any number of 

 prisms having the focal length of the collimator the same 

 as that of the telescope : then the image of the slit in the 

 focus of the telescope will be of the same size and of the 

 same brightness ; for we must, for this consideration, omit 

 the loss of light by reflection and absorption for the pre- 

 sent, as the slit itself, which we will first suppose illu- 

 minated by sodium light, so that two yellow images of it 

 will be visible in the eyepiece. Afterwards we will con- 

 sider the case in which sun-light is used. First let us 

 consider the efl'cct of opening the slit wider, say double 

 the width. By this means the images will be doubled in 

 width and the separation diminished ; the amount of 

 light will be doubled, but will be spread over double 

 the area, so the intensity of illumination will remain 

 the same ; therefore the slit should be as narrow as 

 possible consistent with the image being wide enough 

 to be visible. Secondly, let us double the length of the 

 collimator. Py this we halve the width of the image of 

 the slit, so that the separation is increased, but the dis- 

 tance between the centres of the lines remains the same ; 

 the angle subtended by the collimating lens will in this 

 case be halved, so that the amount o( light passing will 

 only be j of the original amount, but as the image 

 of the slit is reduced in like proportion, the intensity of 

 illumination remains the same ; the effect in this case is 

 therefore the same as narrowing the slit, with the excep- 

 tion of the lines being shorter, thereby reducing the width 

 of the spectrum — a matter generally of little moment, 

 which can be altered at ease by lengthening the slit. 

 Thirdly, we will double the diameter of the collimating 

 lens, and with it that of the telescope and the prism. 

 By these alterations the amount of light passing becomes 

 quadrupled, therefore the images of the slit will be four 

 times brighter; but the angle subtended by the telescope 

 lens at the image is doubled, so that in order to get the 

 whole of the light into the eye, the eyepiece must be 

 placed at half its distance from the image, and be conse- 

 quently doubled in power ; the images will by this be 

 reduced to their original brightness, but they will be mag- 

 nified at the same time, and the distance from centre to 

 centre doubled, the separation doubled, and the width of 

 the images doubled, so that the slit may be reduced in 

 width by 2, and yet leave each image as wide as at first. This 

 will increase the separation between the interior sides of 

 the image still more, so that by doubling the size of our 

 lenses and prisms we have obtained double separation of 

 centres of images, and more than double separation 

 between images, which is just what would be produced 

 by doubling the number or dispersive power of the prisms. 

 It is therefore obvious that in dealing with a bright-line 

 spectrum the power of the instrument depends on the 

 size of the prisms as much as on their number, and an 



increase in number means an increase in the number of 

 reflecting surfaces and loss of light, so that within prac- j 

 tical limits an increase of size is the more preferable. Prac- \ 

 tically, on increasing the size of the collimating lens, as 

 in this case the focal length should be increased, other- 

 wise the lens is injured in defining power, the effect of 

 this increase is, as shown in the second case, only equiva- 

 lent to closing the slit, so it is better to lengthen the 

 colhmator instead of touching the slit ; it is also better 

 to increase the focal length of the telescope glass, thereby 

 straining it less, and so increasing the size of the image 

 of the slit without altering the power of the eyepiece. 



Now let us consider the effect of these alterations on 

 sun-light or other light giving a dark-line spectrum ; and 

 there is this difference between the consideration of this 

 spectrum and the bright-line spectrum, for in this case the 

 dark lines are not images of the slit, but intervals between 

 them, and therefore their width and appearance depend 

 not so much on the separation between the centres of the 

 bright lines as on the separation of their adjacent sides, 

 and with the same \vidth of any two bright lines this 

 separation or width of dark line does not vary in the 

 same ratio as the distance between the centres of the 

 bright lines, or as what is called the dispersive power, 

 varies, but in a higher ratio. For example : suppose there 

 appear in a spectroscope the two sodium lines of appre- 

 ciable v/idth withthefinestpossiblcdarklinebetween them; 

 then, if the distance of their centres is doubled without 

 increasing their width, the black line becomes increased 

 by the increment of the distance of their centres, and 

 with this increment the original dark line becomes much 

 more than doubled ; this will be seen better by drawing 

 two bright lines of appreciable width on paper, and going 

 through the process just mentioned. It is therefore sepa- 

 ration, according to our definition of the word, that is 

 required for dark-line spectra. 



We will now consider the effect when using sun-light 

 instead of sodium light in a similar manner to our first 

 arrangement, namely, in our normal spectroscope, and let 

 us widen the slit as we did before. Every image of the 

 slit will then widen, and the separation between the sides 

 of any two images will diminish, and therefore the dark 

 lines will diminish in width as they are encroached on by 

 the light on either side ; the general spectrum will, how- 

 ever, increase in brilliancy, for although each image is 

 only increased in size, as was the case with sodium light, 

 still the images of each colour overlap, and so produce 

 greater intensity. From this we gather that to obtain 

 the greatest number and width of dark lines, the widili 

 should be as narrow as is compatible with sufficient illu- 

 mination of the spectrum, to show up the dark lines ; and 

 so with a dark-line spectrum as with a bright-line one, 

 the slit should be as narrow as possible. 



Secondly, as with the sodium light, let us lengthen the 

 collimator, say double it : then, as with the sodium light, 

 the images will be halved and the separation increased, 

 but only j of the light passes, and the spectrum is reduced 

 in width by h, so that its briUiancy is .\ what it was origi- 

 nally ; or we may account for the decrease in brilliancy by 

 considering that although, as we showed in the case of the 

 sodium light, the images of the slit are not reduced in 

 brilliancy, still there is less overlapping and so less bril- 

 liancy. So we see that in order to keep a sufficient bright- 

 ness of spectrum to show the dark lines, we must open 

 our slit if we lengthen our collimator, and I'/iV vt-rsif, so 

 that no power is gained by either of these methods, a.-, 

 was the case with the sodium light. Thirdly, we will double 

 the diameter of the collimating lens, and with it that of 

 the prisms and telescope object-glass. By this means the 

 brilliancy only of the spectrum is changed, and this is 

 quadrupled in the focus of the eyepiece, but the focal 

 length must be halved in order to reduce the cylinder of 

 rays small enough to totally enter the eye : this willjmagnify 

 the spectrum to double its original size in every direc- 



