1881. J o/i thr, Orl'jia and Idcuilhj of Hi>(xh'a, GOO 



tho liiiuinifcrouH otluir ; an'l to traco Katisfiu-iorily tlir; prociHO con- 

 iioctioii ])(;t\v(joii tlio occurroiloo of tlio varioiiH vihratioriH and tlio 

 circuriiHtaiicoH under vvliich tlioy occur, will rcrj[uiro an extended 

 soricH of obHcrvationH. 



On the Spectrum of Water. 



In our observations " On tho Spectrum of tlio Compounds of Car- 

 bon," wo noticed that a remarkable sericH of lines, extendinf^ over tho 

 region between the lines S and K of the solar spectrum, wero 

 developed in the flame of coal-gas })iirnlng in oxygen.* The arrange- 

 ment of lines and bands, of which this spectrum consists, is shown 

 in the PI. II., Fig. 3. It begins at the more refrangible end with 

 two strong bunds, with wave-Iengtbs about 30G2, 3008, and extends 

 up to about the wave-length 3210. It is well developed in tho 

 flame of hydrogen as well as of hydrocarbons, burning in oxygen, 

 and less strongly in the flames of non-hydrogenous gases, such as 

 carbonic oxide and cyanogen, if burnt in moist oxygen. The same 

 Rpectrum is given by the electric spark taken, without condenser, 

 in moist hydrogen, oxygen, nitrogen, and carbonic acid gas, but it 

 disappears if the gas and apparatus be thoroughly dried. We are 

 led to the conclusion that tlie sjieetrum is that of water. The plate. 

 Fig. 3, is a general view of this sj>ectrum. It was necessary to pass 

 a current of dry gas for fully an hour through the warmed sj)arking 

 apparatus before the moisture was sufficiently absorbed by tho 

 dehydrating agents. When this was done, pliotographs of the spark 

 showcf] either no trace, or only the faintest traces, of the spectrum 

 above described. On introducing a drop of water, and letting it 

 spread over a plug of asbestos placed in the current of gas, tho 

 spectrum above described at once imprinted itself on the photo- 

 graphic phite. The efiect was the same, whetlier the gas used was 

 hydrogen, oxygen, nitrogen, or carbonic acid. In the case of 

 nitrogen, some of the channelled bands due to that gas overlap tho 

 water spectrum, and partly obscure it, but not so much but that it can 

 be still very distinctly recognised. When a condenser is used, tho 

 water spectrum disappears. The same spectrum appears in tho 

 De Meritens arc, but is less fully developed. The spectrum wo have 

 figured does not by any means exhaust the ultra-violet spectra 

 of the flames we have observed. In writing of this and other spectra 

 which wc have traced to compounds, we abstain from speculating 

 upon the particular moh^cular condition or stage of combination 

 of decomposition, which may give rise to such spectra. The fact of an 

 ultra-violet spectrum of water occurring in spectra of flames opens up 



♦ This we recorded in a Noto of date June 8, 1880, see 'Proc. Rr>y. Roc* 

 No. 20.5, p. .5. Dr. Huggin.s diwcovered the wiinc BpectnJin independently, and 

 communicated the sanne on June l^j, 1880. Our paper on thiH 8p(;cial Bpcctrum 

 bears date 17th June. Loth paperii wore read at tiic same meeting of the 

 Society. 



