OPTICALLY ACTIVE COMPOUNDS ^\ 



would have all sorts of values from + infinity to — infinity. 

 From H on to V the values found for the ordinary dispersion 

 coefficient would at least remain of the same sign, which would 

 be the opposite to that for the dispersion coefficient from D to 

 A. It is clear, therefore, that so variable a quantity can be of 

 little use. If, however, the ordinary zero of rotation be disre- 

 garded, and the dispersion coefficient be calculated, for any two 

 colours of light, with reference to the point at which the T-R 

 curves for these colours intersect one another taken as a rational 

 zero, 00, then it might be that some coefficient of a more constant 

 character would be obtained ; it would at least be one which 

 would not give positive, negative, infinite, and zero values, all 

 for the same substance, in accordance with slight changes of 

 temperature. Thus, for example, the violet/red dispersion 

 coefficient at the point G would be 



a, - Zq' 



a„ and a, being the rotations at the temperature corresponding 

 to the point G, and zq being the value of the rotation at the 

 point of intersection of the curves for violet and red light. 



Values were calculated by the writer, in this way (J.C.S., 

 1916, 109, n88, 1 189), from data obtained by Pickard and 

 Kenyon for methyl-/er/butyl-carbinol and for J-i-naphthyl-;^- 

 hexyl-carbinol. The rational dispersion coefficient found for 

 violet/green was very much better than that ordinarily em- 

 ployed, and, in the latter case, remained approximately constant 

 over 200° of temperature, while the ordinary coefficient 

 assumed all values between plus and minus infinity. This 

 dispersion coefficient may therefore prove to be of considerable 

 practical value, but the examination of active substances is 

 not yet sufficiently complete, even in a few cases, to allow of 

 much extension of the idea. Nevertheless, even in the case of 

 ethyl tartrate, by using dispersion coefficients of this kind {loc. 

 cit. 1 191), a very much better constancy can be arrived at than 

 by any other method. 



A most interesting discovery was made by Armstrong and 

 Walker {Proc. R.S., 191 3, [A], 88, 392) when they found that 

 the rotation values for certain substances could be plotted upon 

 what they termed a characteristic diagram. Their procedure 

 is to plot the values of rotation for several different colours of 

 light against the value of the rotation for some other, reference, 

 colour. The original method of Armstrong and Walker was 

 modified by the present writer in a way which permitted the 

 diagram to be considerably reduced in size {loc. cit., 1 1 81 ). 



Degrees of rotation are marked along a horizontal reference 

 line, say mercury green, and then vertically above or below any 



