The result announced by Young and Forbes (Roy. Soc. Proc. May 17, 1881) that blue light travels in vacuo about 1.8 per cent, faster than red light, raises an interesting question as to what it is that is really determined by observations of this character. If the crest of an ordinary water wave were observed to travel at the rate of a foot per second, we should feel no hesitation in asserting that this was the velocity of the wave; and I suppose that in the ordinary language of undulationists the velocity of light means in the same way the velocity with which an individual wave travels. It is evident however that in the case of light, or even of sound, we have no means of identifying a particular wave so as to determine its rate of progress. What we really do in most cases is to impress some peculiarity, it may be of intensity, or of wave-length, or of polarisation, upon a part of an otherwise continuous train of waves, and determine the velocity at which this peculiarity travels. Thus in the experiments of Fizeau and Cornu, as well as in those of Young and Forbes, the light is rendered intermittent by the action of a toothed wheel; and the result is the velocity of the group of waves, and not necessarily the velocity of an individual wave. In a paper on “Progressive Waves” (Proc. Math. Soc. vol. IX.), reprinted as an appendix to vol. n. of my book on the Theory of Sound [Art. XLVII.], I have investigated the general relation between the group-velocity U and the wave-velocity V.