224 



NATURE 



IJan. 2 2, 1874 



ON DIFFRACTION SPECTRUM PHOTO- 

 GRAPHY, AND THE DETERMINA TION OF 

 THE WAVE-LENGTHS OF THE ULTRA- 

 VJ0LE7 RAYS* 



Illustrated by an ALiiEKT-rvrii Plate. 



THERE are, as is well known, two methods by which 

 spectra may be obtained : (i) by the action of a 

 prism ; (2) by a system of closely-ruled lines. In the 

 latter case it s convenient to speak of the contrivance 

 employed as a grating, and of the spectrum as an inter- 

 ference or diffraction spectrum. A casual inspection 

 shows that there is a great difference between the spectra 

 produced by these two methods, and close investigation 

 proves that the diffraction spectrum is by far the more 

 suitable for accurate scientific work. For this reason it 

 has seemed desirable to make a trustworthy map of those 

 parts of the solar diffraction spectrum which can be pho- 

 tographed on collodion, and to attach to it a scale for 

 reading the wave-lengths of the rays. 



The plate accompanying this memoir is from collodion 

 photographs made by myself, transferred to a thick piece 

 of glass, the latter process being known as the Albert-type. 

 For the entire success of this transfer I am indebted to 

 my friend Mr. E. Bierstadt, the owner of the patent in 

 America. The glass is then used in a printing press in 

 the same manner as a lithographic stone. The spectrum 

 is absolutely unretouched. It represents therefore the 

 work of the sun itself, and is not a drawing either made 

 or corrected by hand. 



The picture consists of two portions : first, the upper, 

 which gives all the lines of the spectrum from near G to 

 O, or from wave-length 4350 ten-millionths of a millimetre 

 to 3440. Above that is placed a scale, which is a copy of 

 Angstrom's from just below G to H.,, with the same-sized 

 divisions carried out from H., to O. The second, or lower, 

 is a magnified portion of the same negative, having Hj 

 and Ho about its middle, and e.\tending from wave-length 

 4205 to" 3736. 



It follows therefore that the fines in the solar spectrum 

 arc correctly represented in their relative positions. The 

 only errors are those which may have arisen from mal- 

 adjustment of the scale. The precautions that were 

 taken to avoid such errors will be described. ^Vith a 

 certain correction, to be mentioned hereafter, it may also 

 be stated that -the relative shadings and intensities are 

 preserved. 



The value of such a map depends on the fact that it 

 not only represents parts of the spectrum which are with 

 difficulty perceived by the eye (though they may be seen 

 by the method of Stokes and Sekulic), but also that even 

 in the visible regions there is obtained a far more correct 

 delineation in those portions which can be photographed. 

 In the finest maps drawn by hand, such as those in the 

 celebrated " Spectre Normal du Soleil" of Angstrom, the 

 relative intensity and shading of the lines can be but 

 partially represented by the artist, and a most laborious 

 and ]3ainstaking series of observations and calculations 

 on the part of the physicist is necessary to secure 

 approximately correct positions of the multitude of 

 Fraunhofer lines. Between wave-lengths 3925 and 4205, 

 Angstrom shows iiS lines, while my original negative has 

 at least 293. 



For such reasons many attempts have been made to 

 procure good photographs of the diffraction spectrum. 

 The earliest were by my father, J. W. Draper ; his results 

 were printed in 1843-44 ia a work entitled "On the 

 Forces which produce the Organisation of Plants." This 



* From the AmeriaiK Journal 0/ ^Science and Art,li'iz. 1873. Com- 

 mitnicated by the avUhor. 



memoir was accompanied by plates drawn fror^ a s 

 daguerreotypes, and the wave-lengths, which he first sug- 

 gested as the proper indices for designating the Fraunhofer 

 lines, were used as a scale. 



Since that time the most important experiments in this 

 direction have been by Mascarl and Cornu. These emi- 

 nent physicists have, however, resorted to the plan of 

 taking portions of the spectrum on a small scale and sub- 

 sequently making enlarged drawings therefrom. Thij 

 course introduces the defects of handwork, and the artistic 

 difficulties of copying intensity and shading, as well as 

 the omission of fine lines. 



In the photographs of the spectrum which I have taken 

 I have tried to get as large a portion as I could at once, 

 and on as large a scale as possible. I have usually ob- 

 tained images from below G (wave-length 4307) to above O 

 (wave-length 3440) of about 12 inches ("305 metre) long. 

 I have succeeded, however, in photographing from near b 

 (wave-length 5 167) to T (wave-length 3032) by resorting to 

 a ruled speculum plane and a concave speculum mirror, 

 but the photographic and optical difficulties in securing 

 an enlarged spectrum of that length are great.* 



Of course, in such a research as this an essential is a 

 finely and evenly ruled, plane of glass or other material. 

 Those which I have used were made by a machine de- 

 vised and constructed by Mr. L. i\I. Rutherfurd, whose 

 beautiful lunar and prismatic spectrum photographs are 

 so well known to the scientific world. The plate generally 

 employed is of glass ruled with 64S1 lines to the inch ; 

 the ruled part is lijy,, inch ('027 metre) long, and /j,*^ inch 

 ('016 metre) wide. It is unquestionably much more nearly 

 perfect than similar gratings made by Nobert and others, 

 for the character of the photographs and the uniformity of 

 the orders on either side of the normal, together with its 

 behaviour under a searching examination, show that it 

 leaves httle to be desired. As it is on glass, and gives a 

 bright transmitted spectrum, I have constructed the re- 

 mainder of the optical apparatus of glass achromatised, 

 according to the plan used by J. W. Draper in 1 S43, ex- 

 cept that I have not silvered the ruling, and therefore have 

 used the refracted, and not the reflected beam. The slit is 

 ^1t of an inch (o'2 metre) long, and yjg of an inch ('00023 

 metre) wide ; the jaws are of steel, and there is not only 

 a micrometer screw for separating them, but also one for 

 setting them at an angle. Occasionally I have taken 

 photographs with the jaws ,'0 of an inch ('00028 metre) 

 apart at the top, and -^\^ ('oooig metre) at the bottom, so 

 as to obtain different intensity in the two edges of the 

 spectrum. 



Most of the photographs have been of the spectrum of 

 the third order, which has certain conspicuous advan- 

 tages. In the first place it is dilated to such an extent as 

 to give a long image, and yet one not too faint to be 

 copied by a reasonable exposure of the sensitive plate ; 

 and in the second place, the spectrum of the second order 

 overlaps it in such a way that D falls nearly upon H, and 

 b upon O. These coincidences are serviceable in deter- 

 mining the true wave-lengths of all the rayj. 



The only point of special interest in connection with 

 the photographic part of the operation, is the device for 

 avoiding the unequal action on the sensitive plate of 

 different rays of the spectrum. It has been commonly 

 supposed, until the recent memoirs of J. W. Draper, that 

 there are in the spectrum three different types of force 

 in three difl'erent but overlapping regions. Heat was 

 supposed to be principally found at the less refrangible 

 end, light in the middle, and actinism at the more refran- 

 gible. But he showed that this error has partly arisen 

 from using prismatic spectra, which condense the red end 

 and dilate the violet, and do not present the rays in the 



* Since writinjj the above I have succeeded in photographing the lines o* 

 the visible specti-ura from b downward, and the picture comprises not only 

 the regions including E, D, C, B, a and A, but also the ultra-red rays. The 

 great groups, n, /?, 7, below A, discovered by my father in 1843, are distinctly 

 reproduced. 



