20 Proceedings of the Royal Society of Edinburgh. [Sess. 
form, but the changes in rotation produced by alteration of temperature 
become greater as the refrangibility of the light increases (J.G.S., 1913, 
103, 165; 1916, 109, 1145, ,1147). 
6. In certain cases the T-R curves for different colours of light inter- 
sect, but no instance appears to be known in which the intersections occur 
all at one point. On the contrary, as for ethyl tartrate ( J.G.S . , 1916, 109, 
1145), the various pairs of curves intersect over a considerable range of 
temperature. This gives rise to what is known as anomalous rotation- 
dispersion, although it is quite probable that this behaviour will prove to 
be the rule rather than the exception. 
7. It has been shown that this region of intersection shifts about as 
the result of change of solvent (J.G.S . , 1916, 109, 1147, 1155), change of 
concentration (J.C.S., 1913, 103, 167), or of change of constitution (ibid., 165, 
166), in much the same manner as does the maximum rotation. Thus, since 
both the maximum and the region of intersection are displaced in a similar 
manner, it seems that the curves are probably displaced as a whole. 
8. By piecing together, as before, evidence obtained in various ways, it 
has been shown that intersection of the T-R curves does not necessarily 
take place on each side of a maximum or minimum. The general behaviour 
so far as it has hitherto been examined, is represented by those parts of 
the graphs in fig. 1 marked abcdefghiklmn. Thus in the case 
of ethyl tartrate below the ordinary temperature, the rotation values are 
numerically small and the rotation-dispersion is negative — that is, the 
absolute values of the rotation diminish as the refrangibility of the light 
increases, at least in the visible spectrum. With rise of temperature the 
rotation values increase in such a manner that in the neighbourhood of 
e / intersection takes place, the rotation-dispersion becoming visibly 
anomalous. On further increase of temperature the curves reach maxi- 
mum values at slightly different temperatures (region g h), after which 
they fall, and passing through points of inflection but without inter- 
section, reach minimum values likewise at somewhat different temperatures 
(region k l m), subsequently rising again, also without intersection, to some 
maximum which it has not yet been found possible to investigate (J.G.S., 
1916, 109, 1153). 
9. Further it is to be noticed very particularly, that although the 
temperature at which any two T-R curves intersect — say, for example, 
those for red and violet light in the case of some substance, A — may be 
very different from the temperature of intersection of the corresponding- 
curves for a derivative or related compound, B, the rotation values at the 
intersection do not differ much. Thus, for example, in homogeneous ethyl 
