PHOTOGRAPHS OF ARC SPECTRA. 109 



were obtained in the neighbourhood of 25 atmospheres, where some 

 plates gave vahies for the di>splacement twice those found on other 

 plates. This result has been discussed more fully elsewhere.* The 

 evidence favours the probability of there being tw'o values for the 

 displacement of a given line at any one pressure. This point, how- 

 ever, requires further examination before it is finally established. 



Between 100 and 215 atmospheres there is a marked increase in 

 the number of reversals in the ultra violet part of the spectrum, which 

 assists in the resolution of the .spectrum into series of lines, and also 

 into well-marked triplets. 



Copper. — Very few of the copper lines reverse under pressure 

 besides the strong members of the principal series, 3247, 3274, but 

 the broadening and displacement are pronounced: at 200 atmos- 

 pheres the displacement being nearly 2 Angstrom units. The most 

 striking feature is the disappearance of the lines belonging to the first 

 and second subordinate series within the region XX 4000-4600. The 

 systems responsible for these vibrations do not appear to exist at 

 high pressures. 



Gold. — Tavo rods of gold, lA in. long, g in. diameter, were ob- 

 tained from Messrs. Johnson and Matthey, of Hatton Gardens, 

 London, and these were screwed to the ends of the electrodes. The 

 gold arc burnt well under pressure, but not so brightly as other 

 metallic arcs. The lower poles burnt aw^ay rapidly, the molten metal 

 running down the sides much as does the wax from a candle. Slides 

 are presented showing the behaviour of this spectrum at 200 atmos- 

 pheres. 



Nickel. — Photographs have also been taken with the poles of this 

 metal at pressures up to 200 atmospheres. 



Silver. — Under pressure the spectrum of this metal undergoes 

 most remarkable changes. At atmospheric pressure silver gives a 

 line spectrum closely resembling that of copper, save that the doub- 

 lets are niore widely separated; but, as the pressure of the surround- 

 ing air is increased, this gradually vanishes, and gives place to a 

 banded spectrum which is not unlike that obtained when silver is 

 heated in a carbon-tube furnace, but actual identification of the bands 

 is difficult since the latter spectrum was obtained at atmospheric 

 pressure.! 



Increase of pressure up to 80 atmospheres effects the gradual 

 change of the banded spectrum into a continuous spectrum, which 

 may be regarded as being due to the broadening of each individual 

 band. Another peculiarity is the structural appearance of the 

 broadened lines of the first subordinate series (which, in the region, 

 XX 4000 to 4600 seem responsible for the banded spectrum). 



A remarkable feature is the appearance on plates taken between 

 5 and 20 atmospheres of pronounced bands (of doubtful origin) in the 

 neighbourhood of 3914. They present a unique phenomenon in that 

 one of the two heads is displaced towards the violet. This is the first 



* Phil. Trans. Roy. Soc, 208, iii., 1908. 



t DufBeld and Eossi, A.strophysical Journal, 1908. 



