Colour and Chemical Constitution. 
215 
It is probable that the free substances have merely the tautomerism of 
phenol (X about 300), viz. CCO + 2CCC, whereas the alkaline forms have 
the nitroic-acid tautomerism involving a new unit, viz. C — N = O in the 
nitroso-compounds and 
in the nitro-compounds. This would afford an explanation in the case of 
the ortho-compounds, in which there would be four tautomerisms instead 
of three, viz. CCO — CCC — CCC — CNO, but in the para-compounds 
there would still be only three, yet the wave-length is only 30 less. This 
must remain an anomaly. If the cause of the colour of quinone itself 
(as well as azobenzene, benzil and other coloured substances incapable of 
tautomerism in the ordinary sense) were known, no doubt paranitrophenol 
could be correlated better with quinone than with a tautomeric sodium 
phenate (quinone having the very high X 461), seeing that sodium para- 
nitrosophenol is almost certainly a salt of the oxime of quinone. 
We come next to the difficult problem of the azo-dyes. Their bands are 
broad and indefinite, consequently measurements of their centres have no 
exactitude. The broadness of the bands means a very perturbed vibration 
or irregular periodic-time of the moving electron in its orbit, which renders 
the additive law almost inapplicable. In addition, the grouping C — N = 
N — C is not capable of ordinary tautomerism ; hence, for example, in 
j^j-j9-dioxyazobenzene the electron cannot travel from one end to the other as 
it does in dioxybenzhydrol. Nevertheless a few regularities can be traced. 
Tuck (J.C.S., 1909) has, for example, pointed out that alkaline oxyazo- 
benzene and neutral aminoazobenzene have nearly the same spectrum, the 
figures being X 420 and X 424 respectively. This can be accounted for on 
my theory if the vibration in the group PhNNC is assigned the value 201, 
because the other part of the molecule is clearly tautomeric and has the 
value 219 for CgH^O-^ (and 223 for CgH^NHo), these giving 420 and 424 
on addition to the constant 201. 
Again, the colour of /ree oxyazobenzene can be roughly calculated if it be 
assumed to be quinone-phenylhydrazone (the salt having, on the contrary, 
the azo-formation). The tautomerism would then be 
Assuming by analogy, since the CCN tautomerism is five less than the 
CCC one, that the CNN tautomerism should have the value 93, we get CCO 
+ CCC -f CNN 22 for half-ring + 16 for phenyl, adding up to 328 for the 
calculated absorption-band in water (Baly gives A 332 [plus 455 faint] in 
alcohol). However, if this theory is the correct one, it follows logically that 
O : N NHPh. 
