A search for signatures of quasar evolution: Comparison of the shapes of the rest-frame optical/ultraviolet continua of quasars at z > 3 and z ∼ 0.1

For 15 bright (V < 17.5), high-redshift (z > 3) quasars, we have obtained infrared spectra and photometry, and optical spectrophotometry and photometry, which we use to construct their spectral energy distributions (SEDs) from λ_rest ∼ 1285-5100 Å. High-resolution spectroscopy for seven enable measurements of their continua shortward of Lyα, and L′ detections of four of these extend their SEDs redward to λ_rest ∼ 7500 Å. We examine the optical/UV continuum shapes and compare these to those of a set of 27 well-studied low-redshift (z ∼ 0.1) quasars which are matched to the high-redshift ones in evolved luminosity. Single power-law fits to the average fluxes within a set of narrow, line-free, windows between 1285 and 5100 Å, but excluding the 2000-4000 Å region of the Fe II + BaC "small bump," are adequate for most of the objects. For both the high- and low-redshift samples, the distributions of spectral indices, α_ouv (F_v ∼ v^αouv) span a wide range, with δα_ouv ∼ 1. The cause of such diversity is investigated, and our analysis is consistent with the conclusion of Rowan-Robinson: that it arises from differences in both the emitted continua themselves and in the amounts of intrinsic extinction undergone. The mean (median) optical/UV spectral indices for the high- and low-redshift samples are -0.32 (-0.29) and -0.38 (-0.40), respectively. A Student's t-test indicates that these do not differ significantly, and a K-S test shows likewise for the distributions. Assuming the optical/UV continuum derives from accretion, the similarity of the spectral indices at high and low redshift is inconsistent with models which interpret the statistical evolution as resulting from a single generation of slowly dimming quasars and instead favors those involving multiple generations of short-lived quasars formed at successively lower luminosities. A clear difference between the high- and low-redshift samples occurs in the region of "small bump." The power-law fit residuals for the low-redshift sample show a systematic excess from ∼2200 to 3000 Å; but this feature is weak or absent in the high-redshift sample. Further study is needed to determine what is responsible for this contrast, but it could reflect differences in iron abundance or Fe II energy source, or alternatively, an intrinsic turnover in the continuum itself which is present at low but not at high redshift.