Morgan, Keenan, and others have found that low dispersions (80-130 Å mm-1) are ideal for obtaining spectral types and luminosity classes, especially for early-type stars. This is because at those dispersions certain blends are unresolved and remain useful, the hydrogen line wings are seen most clearly, and differences in rotational velocities are not so important. Such dispersions are also ideal or adequate for detecting some peculiarities, such as Am and Be spectra and some extreme Ap spectra, but they are inadequate for detecting abnormalities involving weak lines (as in Ap(Hg, Mn) stars) and line profiles (such as in shell spectra).