Mr. W. Ackroyd on Selective Absorption. 427 



tion. In the behaviour of mercuric iodide we have the oppo- 

 site of this, viz. decrease of absorption with increase of 

 kinetic energy, whence we infer that the change of absorption 

 which characterizes metachroinatism is not necessarily accom- 

 panied by change of kinetic energy. 



II. Expansion by heat (i. e. decrease of density) is an all 

 but universal law so far as we at present know. There are 

 several exceptions, however ; and many of these are among the 

 silicates. Their anomalous behaviour is, as a rule, pointed out by 

 the colour- change, as in the case of the zircon. But there may 

 be change of colour, as in the beryl, without alteration of den- 

 sity, i. e. without appreciable molecular approach or recession. 

 On the other hand, we have in olivine an example of change of 

 density (molecular recession) without a corresponding altera- 

 tion of colour. More facts of the same nature might readily 

 be adduced, from which we infer that molar expansion or con- 

 traction is not a necessary concomitant of alteration of struc- 

 tural absorption. We have now excluded from our list of pos- 

 sible concomitants i. and iia. Hence it would appear that ii j3 

 is the only necessary concomitant — a conclusion which receives 

 strength when we consider that the gradual increase of struc- 

 tural absorption in such bodies as HgO, PtCl 4 , and Cr 2 5 Ko 2 

 terminates with their resolution into lower compounds, or into 

 their constituent elements. 



Relation of Colour to Density. — An attempt was made in 

 1765 by Delaval to show the connexion between colour and 

 density. In a paper read before the Koyal Society, he ob- 

 serves * : — " It appears from Newton's experiments that denser 

 substances ought by their greater reflective power, in like cir- 

 cumstances, to reflect the less refrangible rays, and that sub- 

 stances of less density should reflect rays proportionably more 

 refrangible and thereby appear of several colours in the order 

 of their density." The seeming success of this attempt is per- 

 haps due to the comparative nature of his facts ; for he com- 

 pares coloured glasses of each of the following metals in the 

 order of their densities: — gold, red; lead, orange; silver, yellow; 

 copper, green; iron, blue. We can readily see now the error 

 of such a comparison ; for the colour of glass must evidently 

 depend upon the state of the metal in it; e. g. differently 

 coloured glasses may be obtained from the different oxides of 

 the same metal. To take an instance, whilst the cupric oxide 

 gives a bluish-green glass, we know that the cuprous oxide 

 gives a red one. Nevertheless it will be obvious from our 

 foregoing observations that it . is possible to establish a con- 



* Philosophical Transactions, vol. lv. p. 10. 



