EFFECT OF PRESSURE UPON ARC SPECTRA. SII.VKi;. 49 



spectrum. Though, for the case of silver, the pressure necessary for the formation of 

 the band spectrum is higher than that indicated by the amount of the displacement 

 as existing in suuspots, a lower pressure may suffice for the formation of the'titanium 

 oxide flutings. 



The transition from the banded to a continuous spectrum remains for discussion. 

 SCHUSTER and others have suggested that increase of pressure should ultimately 

 produce a continuous spectrum. For metalloids it is known that such is the case : if 

 the pressure of hydrogen in a vacuum tube is gradually increased the lines broaden 

 until the resulting spectrum is continuous. HARTLEY,* discussing the fluted flame 

 spectrum of silver, puts an idea of SCHUSTER'S (' Brit. Assoc. Report,' 1880) into the 

 following words : " Thus we see that there is really no essential difference between 

 the constitutions of the matter which enters into vapours of metals and metalloids ; 

 there is, in fact, something in their constitution common to lx>th which is apparently 

 dependent upon their vapour pressures, and probably due to the actions of the 

 molecules upon one another when their mean path is so extended that their motions 

 become rhythmical. Reduce the freedom of their motions and the result is a 

 continuous spectrum." 



The results of the present experiments bear out this conclusion, and the develop- 

 ment of the banded into a continuous spectrum is regarded as the outcome of the 

 widening of the individual bands. 



It remained to be seen if this banded spectrum were identical with any known 

 spectrum of silver. HARTLEY had investigated the flame spectrum and had photo- 

 graphed flutings, but not in the region of the spectrum under present observation, 

 and, since other methods for flame production also showed no flutings, the writer, 

 with the help of Mr. R. Rossi, B.Sc., undertook some experiments with a carbon-tube 

 furnace to discover if any bands were produced by this method in this region of the 

 silver spectrum. A great number of more or less regularly spaced lines, or flutings, 

 were found. These were photographed with a 1 m. grating spectrograph.t arid 

 measured and compared with the bands obtained in the present experiments. In the 

 last column of Table II. are given all the measurements made in this part of the 

 carbon-tube spectrum. In appearance the spectra are not the same, but it must be 

 borne in mind that the pressures of the two were very different, and how far this 

 factor is effective in masking any resemblance it is difficult to say. There are 25 

 lines in the carbon-tube furnace spectrum differing by less than 1 A.U., or 29 lines 

 differing by less than 2 A.U., from the measurements made under pressure, out of a 

 total of 55 furnace flutings and 63 arc flutings. The evidence is thus not entirely 

 favourable to the two being the same spectrum, especially as the strong violet edges of 

 bands in the furnace spectrum (4266'6, 4294'9, 4322'2) are absent under pressure, as 

 are also the strong flutings at 4505'1, 45G4'5, 4600'8, and 4607 '8. But these 



* HARTLEY, ' Phil. Trans.,' A, vol. 185, p. 166, 1894. 



t DUFFIELD and Rossi, ' Astrophys. Journ.,' vol. XXVIII., p. 371, 1908. 



FOL. WXI. A. H 



