Colour and Chemical Constitution. 
71 
The calculated colour is 1380 x 065 x 0*39 = A 350. It is described in 
'Beilstein' as pale yellow, which agrees with the calculation. Dioxydihydro- 
anthracene should have a deeper colour corresponding to 1380 x (0*65) 2 = 
A 583. 
Dioxyanthraquinone is an exception : the two CO groups cause a new 
kind of vibration, and the colour is much higher — eight times as high in 
fact as theory predicts. No doubt the theoretical colour (A 76) is there also, 
but is too low to observe. 
These nineteen doubly-linked dicyclic colours have been selected so as to 
cover the whole field, so far as known (the peroxide and sulphoxide dyes 
are not yet known) : it would merely be taking up space, for example, to 
calculate erythrosine when eosine is calculated, or to calculate the fourteen 
derivatives of thionine observed by Formanek. 
There are also dicyclic dyes in which the rings are joined by chains 
longer than one element : indigo and the azo-dyes are examples, but 
the published data are insufficient — indeed almost useless — as regards a 
quantitative explanation. 
Indigo may be calculable from the hvdrated formula : 
rY NHs NH fi 
l^J — CO — C(OH) : C(OH) — CO — I y 
and the thiazole and primuline dyes similarly, viz. ; 
A/ Nff > 
[^J— S— CO— / ^>NH 2 
but the laws of the co-operation of two or more factors in a linkage are not 
yet known. The azo-dyes do not come into the scheme for the reason that 
they are coloured whether they contain ionisable groups or not ; but it is 
worth noting that the three which do contain two ionisable rings (azophenol, 
azoaniline, etc.) have colours fairly similar to those of the corresponding 
phenazines, from which they differ by two hydrogen atoms only. This 
suggests that the azo group can sometimes be written — NH — NOH — and 
the colour calculated accordingly, allowing about 0*73 instead of 0 - 78 for 
the / /NH\ factor when not replacing two hydrogen atoms of the benzene 
rings but inserted in a previously existing linkage. 
The most of the azo-dyes, however, owe their colour to the azo- linkage 
independently of all rings, since diazomethane shows colour. 
It is probably desirable to show the way in which the linkage-factors 
were ascertained from the observations, particularly as this also exhibits the 
degree of variation of each factor, and enables the reader to see that each 
factor is really a constant within the observational error. 
(1) The oxo-linkage.—(a) Eesorcin-benzene : observed A 492, The corre- 
6§ 
