The determination of redshifts of galaxies in distant clusters has been at a standstill since the summary paper of Humason, Mayall, and Sandage (1956) was published. In that paper redshifts up to 61,000 km s-1 or z = 0.2025 were reported. Since that time larger redshifts have been determined only for galaxies with strong emission lines and which, therefore, may not be normal. One of the most difficult accomplishments in the last 10 years was the determination of the redshift of 3C 295 by Minkowski (1960) by identifying the most obvious feature of the spectrum with the emission line 3727 of [O II]. Schmidt (1965) has determined redshifts greater than 0.2 for several radio galaxies with strong emission lines.
As far as normal galaxies are concerned, the limiting magnitude reached of V 17.4 is presumably due to the fact that such galaxies are extended objects and the sky brightness is so great that photographic spectra are completely dominated by the sky spectrum. Image-tube spectrographs decrease the time it takes to obtain a spectrum, but do not improve the contrast of the galaxy spectrum against that of the sky.
Several attempts to determine redshifts greater than 0.2 have been made. A number of these are discussed by Zwicky (1959). Of particular interest was the work of Baum (1962), who measured broad-band colors from 3730 to 9875 Å, calibrated them absolutely in terms of flux, and then determined the redshift by fitting the derived fluxes to the corresponding broad-band fluxes for nearby galaxies. He was successful in this and has published several redshifts (Baum 1962). Because the bandpasses were large, the inherent accuracy was not high. Also, the data were not sufficient to determine whether the galaxy energy distribution was normal.
Several years ago it was realized that the only practical way to determine large redshifts was to use a technique in which the sky background was subtracted accurately. A star-sky chopping system coupled with pulse-counting electronics had been used successfully in the prime-focus spectrum scanner for many years and was an obvious technique to use. With the determination of large redshifts of distant galaxies as the prime motive, the 32-channel photoelectric spectrometer was designed and built (Oke 1969). This instrument is now being used regularly for work on very faint objects; some early results are reported below. Gunn (1971) has also reported on similar work of this kind.