Hires spectroscopy of APM 08279+5255:: Metal abundances in the Lyα forest

We present high-S/N echelle spectra of the recently discovered ultraluminous QSO BPM 08279 + 5255 and use these data to reexamine the abundance of carbon in Ly alpha forest clouds. In agreement with previous work, we find that approximately 50% of Ly alpha clouds with hydrogen column densities of log N(H I) greater than or equal to 14.5 have associated weak C N absorption, with log N(C rv) greater than or similar to 12.0 and we derive a median N(C IV)/N(H I) = 1.4 x 10(-3). The agreement with earlier estimates of this ratio may be somewhat fortuitous, however, because we show that previous analyses have probably overestimated the number of Ly alpha clouds that should be included in this statistic. We then investigate whether there is any C Iv absorption associated with lower H I column densities by stacking 51 C Iv regions corresponding to 51 Ly alpha lines with 13.5 less than or equal to log N(H I) less than or equal to 14.0. The coadded spectrum has a S/N similar or equal to 580 but shows no composite C rv absorption. In order to understand the significance of this nondetection, we have stacked together 51 theoretical C N lambda 1548 lines with individual values of column density and velocity dispersion scaled appropriately from the values measured in the corresponding Ly alpha lines. We find that even if the typical value of N(C IV)IN(H I) = 1.4 x 10(-3) applies to these lower column density clouds, the corresponding signal in the stacked C rv region is smeared by the likely random difference in redshift between C Iv and Ly alpha absorption and becomes very difficult to recognize. This seems to be a fundamental limitation of the stacking method, which may well explain why in the past it has led to underestimates of the metallicity of the Ly alpha forest. We also analyze our spectra with the pixel-by-pixel optical depth technique recently developed by Cowie & Songaila and find evidence for net C rv absorption in Ly alpha clouds with optical depths as low as tau(Ly alpha) = 0.5-2, as these authors did. However, we show with simulations that even this method requires higher sensitivities than reached up to now to be confident that the ratio N(C IV)/N(H I) remains constant down to column densities below log N(H I) similar or equal to 14.0. We conclude that the question of whether there is a uniform degree of metal enrichment in the Ly alpha forest at all column densities has yet to be fully answered. Future progress in this area will probably require concerted efforts to push further the detection limit for C rv lines in selected bright QSOs.