ON OUR KNOWLEDGE OF SrECTRUM ANALYSIS. 327 



down on the paper (stretched on a drawing-board), and the other to rule 

 the curve with a finely pointed hard pencil. The author of this report 

 employs a little drawing instrument consisting of a steel bar, mounted on 

 a brass base which rests on the paper. By means of clamping-screws the 

 steel bar can be held bent in the required curve, whether of equal curvature 

 throughout its length, or more curved in one part than another. 



A somewhat different method of procedure is described in a paper by 

 Mr. Wm. Dodgson in the sixth volume of the third series of the ' Memoirs 

 of the Literary and Philosophical Society of Manchester.' 



The best paper for the purpose is a paper ruled into millimetres and 

 centimetres made in rolls 69 centimetres broad, which may be obtained 

 through Messrs. Williams and Norgate, 14, Henrietta Street, Covent 

 Garden, or a somewhat similar paper to be obtained from Messrs. 

 Lechertier, Barbe and Co., of 60, Regent Street. These papers are more 

 uniform and free from shrinkage than any others. Another paper also 

 ruled in millimetres, in sheets 1 metre by 7 decimetres, is to be obtained 

 from C. Dupressoir, Rue St. Honore, 175, Paris. A paper, ruled in inches 

 and tenths, 24 inches by 15 inches, is to be obtained from Waterlow and 

 Sons, 60 and 61, London Wall, but it is hardly uniform enough for the 

 purpose. Some trouble expended in drawing a good curve will be very 

 well repaid. The line obtained in this way will generally be very much 

 curved, but the less curved it is the more easily is it drawn and the more 

 exactly can it be employed. A less curved line is obtained by using the 

 reciprocals of the wave-lengths instead of the wave-lengths themselves.' 

 The adoption of this scale of inverse wave-lengths or of oscillation- 

 frequencies is strongly recommended by a Committee of the British 

 Association, under whose superintendence a catalogue ^ of oscillation- 

 frequencies and a corresponding map of the Praunhofer lines have been 

 prepared. It is hoped that this catalogue will be extended to the bright 

 lines of metals not present in the sun's atmosphere. 



The map of oscillation-frequencies is intermediate between a diffraction- 

 spectrum and a dispersion-spectrum, the red end being less extended when 

 compared with the blue end than in Angstrom's map, and more extended 

 than in Kirchhoff's. A map drawn to wave-lengths is too much dis- 

 torted to be advantageously employed with a dispersion-spectroscope, and, 

 on the other hand, a spectrum mapped with a dispersion-spectroscope 

 does not sufficiently resemble the same spectrum seen with a diffraction- 

 spectroscope ; but a map of oscillation-frequencies, being intermediate 

 between the two, is not so different from either but that it is suitable for 

 use both with diffraction-spectroscopes and with dispersion-spectroscopes. 

 Further rays which are harmonically related are represented in the map 

 of oscillation-frequencies by equidistant lines and in the catalogue by an 

 arithmetic series whose common difference is equal to its first term. The 

 map accompanying this report shows the scale of a one-prism spectroscope 

 reduced both to wave-lengths and to oscillation-frequencies. It will be 

 seen that the second line is much less curved than the first. 



■ If the squares of the reciprocals be employed the interpolation curve will be 

 very nearly (but only nearly) a straight line. 



= British Association Report, 1878, Dublin Meeting. 



