Diffraction-grating spectrographs are far more satisfactory for all photographic work, especially now that the excellent celluloid replicas can be obtained so cheaply. The principle of the direct. vision spectroscope is described elsewhere, and the same arrangement may be used for the spec trograph. A section of such an instrument, designed by Tallent, is shown at E, in which S is the slit, C the collimator lens, D the grating cemented to a narrow angle prism which refracts the central white beam to w into a small pocket of blackened wood ; r. is the camera lens, and P the dark-slide. The lenses are single landscape lenses, and their foci or that of the camera lens will depend upon the length of the spectrum desired. A lens of 14} in. will give a spectrum of about 4 in. in length between X 3,50o and X 8,000. The slit must be placed at the equivalent focus of the collimator lens, and all parts except just the grating aperture should be blocked out and the interior of the camera lined with black velvet or blackened to prevent reflections.
If it is not desired to use a prism grating, then obviously the plate must not be in a straight line with the grating, but at the angle of diffrac tion. This may be calculated as follows: The inch mm.
x 14490 b = •oo175292 The correct position of the plate may also be found by supporting the collimator lens on a block of wood, or temporarily attaching it to a card and placing the slit at its equivalent focus close to a strong light ; then, on inserting the grating close behind the lens, the first order spec trum will be seen on each side of the white image of the slit, and the distance from this will, of course, give the angle between the axial line and the grating. The grating may be placed at spectra formed by a diffraction grating lie on each side of the central white image, and to find the distance of the first wave-length or the angle of diffraction let F represent a spectroscope in which b is the grating space—that is, the width of one ruling and the adjacent space—and e the angle of deviation. What we wish to deter mine is either e, or the length of A B, then taking •000334o mm. as the first wave-length which it is desired to find the position of— A B = sin e = -000334o b and assuming that b = we have— sin e = •000334o .001752 = •1906 = practically.
The grating space, or b, is found very easily, thus : Assuming that we have a grating with 14,490 lines to the inch, then— right angles to the collimator lens, or preferably so that its plane cuts the angle between the axes of the collimator and camera lens—that is, so that the angles of incidence and diffraction are equal.
Fora plane metal or reflection grating the arrangement shown for this (see" Spectroscope ") may be adopted, replacing the telescope by a. camera, and fixing the collimator and camera at an angle of 45°, the grating being mounted on a revolving table.
Excellent replicas, cemented to sections of con vex lenses so as to form concave gratings, being now obtainable, it may be as well to give Eder's method for mounting concave gratings for spec trography. The idea is to keep the slit position constant, which is a great convenience when using a heliostat or fixed source of artificial light, H in G. The camera moves round a.
point G ; the slit s, which is joined to the tube T, always moves along s G ; the plate P moves on the arc r" P P"', which is struck from the centre G ; the slit s is joined by a rod s B C to C, which is the centre of the distance P G, G being the grating. It is obvious then that as the plate is moved round the arc r" P P"' the slit and the grating also move. The central point of the grating must be exactly over the point G, and the slit s must be exactly over the point s, at which the rod is joined to T. T is a collap sible tube, which can be shortened or lengthened as it approaches or recedes along s G.
When diffraction spectra are used, the spectra are not isolated, but overlap ; and this over lapping follows the law XI = a Att = 3 = 4 = 5 in which XI, etc., are the wave-lengths in the first, second, third, etc., orders. This overlap is shown in the following table, taking X 6,000 in the first order : First order spectrum 6,00o Second „ 3,000 6,000 Third „ 71 1,500 4,000 6,000 Fourth 4,3oo 6,000 Fifth „ 3,60o 4,800 Sixth „ If 3,000 4,000 be placed in contact with a plate and a long exposure given to the arc or magnesium ribbon to see that it does absorb the whole of the ultra violet.
It may also be found convenient to fix per manently in the dark-slide a scale either of wave For visual work this overlap is of no moment when using the first order only, because the eye is not sensitive to the ultra-violet ; but in spec trography this overlapping may be very trouble some, particularly when dealing with the ex treme red, for if we take the limit set by the absorption of the glass a_s X 3,400, as that which would act on the plate, this line would coincide with X 6,80o in the first order, and therefore the farther we proceed with the red the more active will be the ultra-violet of the second order, and thus totally erroneous conclusions may be drawn from the results.
This superposition of the second order can always be seen in negatives when using day light, the electric arc, or magnesium ribbon ; so that it is as well to use some absorbent material between the light source and the plate. This may be either in front of the slit or, prefer ably, in front of the plate itself. In the former position there may be used a cell containing a solution of xsculin or filter yellow K, or a gelatine screen stained with the above. For a screen in contact with the plate it is advisable to coat a sheet of thin patent plate with a 5 per cent. solution of gelatine, and when dry to stain up about half of the plate ; that .is, up to about where the D lines fall, with filter yellow K or dianil orange G. The screen should then lengths or one merely divided into known divi sions, and excellent glass scales io mm. in length and divided into millimetres can be obtained commercially. With such a scale, which is impressed on the plate at each exposure, it is easy to calculate with approximate accuracy the wave-length of any line or the limits of an absorption band, etc. For instance, having determined that the distance between and C is exactly 90 mm., we have only to divide d —that is, their difference in wave-length—by 90 to find the number of wave-lengths to a millimetre, thus