12 Pi'of. Pliicker on the Spectra in Rm'efed Gases 



vacuum, and accordingly no light is then produced. We are led 

 therefore to the conclusion that the light of the discharge-cur- 

 rent, and the consequent corresponding spectrum of such gas- 

 vacua, entirely depend upon the residual traces of gas, and are 

 consequently characteristic of the gas employed on each occasion. 

 For the determination of the special spectra, I chose, to begin 

 with, hydrogen, oxygen, and carbonic acid ; and I obtained for 

 these three gases perfectly definite, constant, and distinct spectra. 

 It is true that it is impossible in a graphic representation to re- 

 produce even approximately the original colours of such spectra. 

 Such representation is also rendered the more difficult by certain 

 parts of the spectra being characterized by an unusual intensity 

 of light. But such a representation furnishes a true picture of 

 the phsenomenon for those who have once seen it, and it suffices 

 completely for subsequent comparison. 



98. Graphic representation is also rendered less exact by the 

 fact that the circumstances conditioning the subjective judg- 

 ment of colours are of a very complex kind, and that such judg- 

 ment loses all value vifhen weakly illuminated portions are 

 bounded by tints of a bright colour. The impression made 

 upon the eye by the beautiful spectra of many gases stands in 

 contradiction to the fundamental law of the theory of colours, 

 namely, that the colour of the light depends upon the length of 

 the undulation, according to which, for the same piism, light 

 broken to the same extent must have the same colour under 

 a/l conditions. I may mention in illusti-ation of this, the (im- 

 pure) spectrum of fluoride of boron already provisionally de- 

 scribed (71), in which a beautiful violet colour follows immedi- 

 ately upon a (certainly somewhat faded) green colour. If the 

 blue were entirely wanting, as is the case with other gases, the 

 violet would be bounded by a black space towards the side of the 

 red. It seems most natural to assume that a faint blue is pre- 

 sent in the specti'um, and that this is converted apparently into 

 a green by a subjective yellow, the complementary colour to the 

 neighbouring and unusually bright violet. Nevertheless it 

 seemed to me desirable for our gas-spectra, to support the above 

 la A' by a direct experiment. 



99. In the formation of such spectra I employed the method 

 already described (69), with some modifications. The luminous 

 electrical discharge-current was concentrated in thermometer- 

 tubes, whose internal diameters were nearly the same for the 

 different gases examined, being about 0'6 millim. (A column 

 of mercury, 155 millims. long, in sucli a tube weighed 0"6677 

 gr., which gives this diameter for the internal circular section 

 of the tube.) Fig. 2 shows the form of tlie perfect separate gas- 

 tubes, as well as the manner in which they inayl)c connected on 



