NA TURE 



{June 29, 1882 



and produced some which were nearly 2 inches square, though 

 he was rarely successful above if inches, having about 30,000 

 lines. These gratings were on speculum metal, and showed 

 more of the spectrum than had ever before been seen, and have, 

 in the hands of Young, Rutherfurd, Lockyer, and others, done 

 much good work for science. Many mechanics in this country, 

 and in France and Germany, have sought to equal Mr. Ruther- 

 furd 's gratings, but without success. 



Under these circumstances, I have taken up the subject with 

 the resources at command in the physical laboratory of the Johns 

 Hopkins University. 



One of the problems to be solved in making a machine is to 

 make a perfect screw, and this, mechanics of all countries have 

 sought to do for over a hundred years and have failed. On 

 thinking over the matter, I devised a plan whose details I shall 

 soon publish, by which I hope to make a practically perfect 

 screw, and so important did the problem seem, that I imme- 

 diately set Mr. Schneider, the instrument maker of the university, 

 at work at one. The operation seemed so successful, that I 

 immediately designed the remainder of the machine, and have 

 now had the pleasure since Christmas of trying it. The screw 

 is practically perfect, not by accident, but because of the new 

 process for making it, and I have not yet been able to detect an 

 error so great as I- 100,000th part of an inch at any part. 

 Neither has it any appreciable periodic error. By means of this 

 machine I have been able to make gratings with 43,000 lines to 

 the inch, and have made a ruled surface with 160,000 lines on 

 it, having about 29,000 lines to the inch. The capacity of the 

 machine is to rule a surface 6^ X 4} inches, with any required 

 number of lines to the inch, the number only being limited by 

 the wear of the diamond. The machine can be set to almost 

 any number of lines to the inch, but I have not hitherto attempted 

 more than 43,000 lines to the inch. It ruled so perfectly at this 

 figure that I see no reason to doubt that at least two or three 

 times that number might be ruled in one inch, though it would 

 be useless for making gratings. 



All gratings hitherto made have been ruled on flat surfaces. 

 Such gratings require a pair of telescopes for viewing the spec- 

 trum ; these telescopes interfere with many experiments, absorb- 

 ing the extremities of the spectrum strongly ; besides, two tele- 

 scopes of sufficient size to use with 6-inch gratings would be 

 very expensive and clumsy affairs. In thinking over what would 

 happen were the grating ruled on a surface not flat, I thought 

 of a new method of attacking the problem, and soon found that 

 if the lines were ruled on a spherical surface, the spectrum 

 would be brought to a focus without any telescope. This dis- 

 covery of concave gratings is important for many physical inves- 

 tigations, such as the photographing of the spectrum both in the 

 ultra-violet and the ultra-red, the determination of the heating 

 effect of the different rays, and the determination of the relative 

 wave-lengths of the lines of the spectrum. Furthermore, it 

 reduces the spectroscope to its simplest proportions, so that 

 spectroscopes of the highest power may be made at a cost which 

 can place them in the hands of all observers. With one of my 

 new concave gratings I have been able to detect double lines in 

 the spectrum which were never before seen. 



The laws of the concave grating are very beautiful, on account 

 of their simplicity, especially in the case where it will be used 

 most. Draw the radius of curvature of the mirror to the centre 

 of the mirror, and from its central point with a radius equal to 

 half the radius of curvature draw a circle ; this circle thus passes 

 through the centre of curvature of the mirror, and touches the 

 mirror at its centre. Now if the source of light is anywhere in 

 this circle, the image of this source and the different orders of 

 the spectra are all brought to focus on this circle. The word 

 focus is hardly applicable to the case, however, for if the source 

 of light is a point, the light is not brought to a single point on 

 the circle, but is drawn out into a straight line with its length 

 parallel to the axis of the circle. As the object is to see lines 

 in the spectrum only, this fact is of little consequence, provided 

 the slit, which is the source of light, is parallel to the axis of 

 the circle. Indeed, it adds to the beauty of the spectra, as the 

 horizontal lines due to dust in the slit are never present, as the 

 dust has a different focal length from the lines of the spectrum. 

 This action of the concave grating, however, somewhat impairs 

 the light, especially of the higher orders, but the introduction 

 of a cylindrical lens greatly obviates this inconvenience. 



The beautiful simplicity of the fact that the line of foci of 

 the different orders of the spectra are on the circle described 

 above, leads immediately to a mechanical contrivance by which 



we can move from one spectrum to the next, and yet have the 

 apparatus always in focus ; for we have only to attach the slit, 

 the eye-piece, and the grating to three arms of equal length, 

 which are pivoted together at their other ends, and the condi- 

 tions are satisfied. However we move the three arms, the spectra 

 are always in focus. The most interesting case of this contriv- 

 ance is when the bars carrying the eye-piece and grating are 

 attached end to end, thus forming a diameter of the circle with 

 the eye-piece at the centre of curvature of the mirror, and the 

 rod carrying the slit alone movable. In this case the spectrum 

 as viewed by the eye-piece is normal, and when a micrometer is 

 u-ed, the value of a division of its head in wave-lengths does not 

 depend on the position of the slit, but is simply proportional to 

 the order of the spectrum, so that it need be determined once 

 only. Furthermore, if the eye-piece is replaced by a photo- 

 graphic camera, the photographic spectrum is a normal one. 

 The mechanical means of keeping the focus is especially im- 

 portant when investigating the ultra-violet and ultra-red portions 

 of the solar spectrum. 



Another important property of the concave grating is that all 

 the superimposed spectra are in exactly the same focus. When 

 viewing such superimposed spectra it is a most beautiful sight to 

 see the lines appear coloured on a nearly white ground. By 

 micrometric measurement of such superimposed spectra we have 

 a most beautiful method of determining the relative wave-lengths 

 of the different portions of the spectrum, which far exceeds in 

 accuracy any other method yet devised. In working in the ultra- 

 violet or ultra-red portions of the spectrum we can also focus on 

 the superimposed spectrum, and so get the focus for the portion 

 experimented on. 



The fact that the light has to pass through no glass in the 

 concave grating makes it important in the examination of the 

 extremities of the spectrum where the glass might absorb very 

 much. There is one important research in which the concave 

 grating in its present form does not seem to be of much use, and 

 that is in the examination of the solar protuberances ; an instru- 

 ment can only be used for this purpose in which the dust in the 

 slit and the lines of the spectrum are in focus at once. It might 

 be possible to introduce a cylindrical lens in such a way as to 

 obviate thr difficulty. Hut for other work on the sun the con- 

 cave grating w'll be found very useful. But its principal use 

 will be to get the relative wavelengths of the lines of the 

 spectrum, and so to map the spectrum ; to divide lines of the 

 spectrum which are very near together, and so to see as much as 

 possible of the spectrum ; to photograph the spectrum so that it 

 shall be normal ; to investigate the portions of the spectrum 

 beyond the range of vision ; and lastly to put in the hands of 

 any physicist at a moderate cost such a powerful instrument as 

 could only hitherto be purchased by wealthy individuals or 

 institutions. 



To give further information of what can be done in the way of 

 gratings I will state the following particulars : — 



The dividing engine can rule a space d\ inches long, and 4} 

 inches wide. The lines, which can be 4^ inches long, do not 

 depart from a straight line so much as l-ioo,oooth of an inch, 

 and the carriage moves forward in an equally straight line. 

 The screw is practically perfect, and has been tested to I -100,000th 

 of an inch, without showing error. Neither does it have any 

 appreciable periodic error, and the periodic error due to the 

 mounting and graduated head can be entirely eliminated by a 

 suitable attachment. For showing the production of ghosts by 

 a periodic error, such an error can be introduced to any reason- 

 able amount. Every grating made by the machine is a good one, 

 dividing the I474line with ease, but some are better than others. 

 Rutherfurd's machine only made one in every four good, and 

 only one in a long time which might be called first-class. One 

 division of the head of the screw makes 14,43s lines to the inch. 

 Any fraction of this number in which the numerator is not 

 greater than say 20 or 30 can be ruled. Some exact numbers to 

 the millimetre, such as 400, Soo, 1200, &c, can also be ruled. 

 For the finest definition either 14,438 or 2S.S76 lines to the inch 

 are recommended, thejfirst for ordinary use, ana the second for 

 examining the extremities of the spectrum. Extremely brilliant 

 gratings have been made with 43,314 lines to the inch, and 

 there is little difficulty in ruling more if desired. The following 

 show some results obtained : — 



Flat grating, I inch square, 43,000 lines to the inch. Divides 

 the 1474 line in the first spectrum. 



Flat grating, 2X3 inches, 14,438 lines to the inch, total 

 43,314. Divides 1474 in the first spectrum, the E line (Ang- 



