ABSTRACT The enormous cost of even relatively simple focal-plane instruments for 8 ± 2 meter telescopes is, inevitably, acting as a driver to produce very general-purpose spectrographs for each of them. Somewhat unexpectedly, the result seems to have been a convergence on very similar, cross-dispersed, R = lambda/Delta-lambda ~105 Echelles for nearly every such telescope in use, under construction, development, or consideration. Such spectrographs are well matched to studies of small numbers of relatively bright point sources (stars or QSOs) directed at understanding either the sources themselves or intervening gas responsible for absorption lines. They are much less well suited to studies of extended objects (nebulae, galaxies) and to studies of populations of stars or galaxies, including those in clusters, and to studies of very faint objects, where accurate sky subtraction is vital. Adaption of the existing and contemplated instruments to long-slit, multi-slit, and multi-object use apparently carries high prices in lost photons, increased noise, and increased exposure time.