Eight years ago, in the introduction to a paper by Colonel J. W. Gifford,* attention was directed to the need for investigating the form of the curves of refractive dispersion by similar methods to those which had already been followed successfully in the investigation of curves of rotatory dispersion. Two attempts were made to develop a programme of research on these lines, by applying (i) to the hydrocarbons, benzene and cyclohexane, and (ii) to the optically-active base, nicotine, the extremely precise methods of measurement which Gifford had used in determining the refractive indices of glass and quartz. The conditions under which the research was carried out, however, were such that an extensive programme of work became impracticable; moreover, the data then recorded suffered from two limitations, which it was important to overcome before extending the scope of the research indefinitely. (a) The first of these limitations was due to the'common difficulty of efficient control of temperature in optical apparatus. Thus, it has already been shown that a gradient of temperature of 1° between the air and the water-jacket of a Pulfrich refractottieter produces a cooling of 0·1° in the liquid under examination. In practice, Gifford’s measurements were always made at room temperature and corrected to 15° with the help of observations at higher and lower temperatures. This is a perfectly satisfactory way of dealing with solids, which have only a small coefficient of expansion, or even with water, of which he coefficient of expansion is exceptionally small; but the much larger coefficients of expansion of organic liquids introduce a serious source of error, which makes it almost a waste of time to use the super-excellent methods (involving 36 readings for each wave-length), whereby Gifford measured the refractive indices of quartz, fluorite and amorphous silica with an accuracy that is unlikely to be surpassed during the lifetime of the present generation.