ON OPTICAL ROTATORY DISPERSION. 
269 
agreement was much, better than with the linear formula, as was inevitable since there 
was exact agreement at three points instead of two ; in the more concentrated solutions, 
also, the parabolic formula gave rather better results than the linear law ; but in the 
more dilute solutions the errors were increased very considerably by the introduction 
of a third term, with the result that outside the limit of 55 to 85 per cent, of water, 
the parabolic was distinctly worse than the linear formula. 
The real character of the deviations was only revealed when the errors of the various 
formulae were plotted against the concentrations as in figs. 2 and 3. As in the case of 
