2004ApJ...617..867D

We have formed ``composite spectra'' by combining the integrated-light spectra of individual galaxies in eight intermediate-redshift and 12 low-redshift clusters of galaxies. Because these composite spectra have much higher signal-to-noise ratios than individual galaxy spectra, they are particularly useful in quantifying general trends in star formation for galaxy populations in distant clusters, z>0.3. By measuring diagnostic features that represent stellar populations of very different ages, a grand composite spectrum can reflect the fractions of those populations as accurately as if excellent spectral measurements were available for each galaxy. Such composite spectra can also be useful in the study of finer spectral signatures, for example, spectral indices that break the age-metallicity degeneracy, and the shape of the Hδ absorption line as an indicator of the age and duration of an epoch of starbursting galaxies in a cluster. Measuring the equivalent widths of spectral features in composite spectra is especially well suited for comparing the cosmic variance of star formation in clusters at a given redshift or comparing clusters over a range of redshifts. When we do this we find that [O II] emission and especially Balmer absorption is strong in each of our intermediate-redshift clusters and completely separable from a sample of 12 present-epoch clusters, where these features are weak. Cluster-to-cluster variations at a given epoch seem to be smaller than the strong trend in redshift, which suggests that cosmic evolution is the major factor in the star formation histories of cluster galaxies. Specifically, we show by comparison with the Hδ strengths of present-epoch populations of continuously star-forming galaxies that the higher redshift samples must contain a much higher fraction of starburst galaxies than are found today in any environment.