The fact that gratings give better resolving power than prisms in spectrometry in instances where available source energy or available detector sensitivity, and not diffraction blurring of spectrum details, sets the limitation, is shown to be because of the greater dispersion of gratings. The available energy for a given band pass, Δν0, for either prisms or gratings, is shown to be proportional to the product of efficiency and dispersion. This product is estimated for gratings and prisms. The best grating spacing is indicated to be of the order of one wave-length which, for the visible spectrum, would mean a grating of some 37,400 lines per inch. The importance of obtaining true echelette facets is emphasized and a means of obtaining them is suggested.