THE ASTROPHYSICAL JOURNAL, 457:102-110, 1996 January 20

               SIMULATION ANALYSIS OF Ly-Alpha FOREST SPECTRA. I.
                        EMPIRICAL DESCRIPTION AT z ~ 3

                                ADAM DOBRZYCKI
        Harvard-Smithsonian Center for Astrophysics, 60 Garden Street,
                Cambridge, MA 02138; adobrzycki@cfa.harvard.edu

                                      AND

                                 JILL BECHTOLD
         Steward Observatory, University of Arizona, Tucson, AZ 85721;
                           jbechtold@as.arizona.edu


                                   ABSTRACT

        We present moderate-resolution (~50 km/s FWHM) spectra of the
     Ly-alpha forest for seven quasars with redshifts ranging from 2.53 to
     3.13, obtained with the Blue Spectrograph and photon-counting Reticon
     at the Multiple Mirror Telescope.  Combined with spectra of 10 other
     quasars presented elsewhere, we have characterized the distribution of
     cloud properties in a way which was designed to minimize any
     subjective part of the analysis.
        We used artificial absorption spectra, with the same resolution,
     sampling and signal-to-noise ratio as a function of wavelength as the
     actual data.  Distributions of the physical parameters of the Ly-alpha
     clouds, namely, the neutral hydrogen column density (N) and Doppler
     parameter (b), were approximated with d"N"/dN proportional to
     N^(-beta) and d"N"/db proportional to exp[-(b-<b>)^2/(2 sigma_b^2)],
     respectively.  We constructed a grid of simulated spectra with
     different input parameters.  Comparison of properties of the simulated
     spectra with the observed spectra yielded acceptable ranges of
     parameters.  Our technique differs from previous similar work in that
     we use the information contained in the distribution of the strength
     of the absorption in each resolution element and the distribution of
     separations between absorption complexes.  We derive beta = 1.4 +/-
     0.1 for N ranging from 10^13 to 10^16 cm^(-2) and <b> = 30 +/- 15
     km/s.  Most previous studies based on line lists indicated beta =
     1.7-1.9.  We attribute this difference to flattening of the column
     density distribution for low N, recently confirmed by higher
     resolution observations.  Our result for <b>, though consistent with
     values quoted in the literature, is of lower significance, since it is
     less than the resolution of our spectra.  We conclude by commenting on
     the importance of line blending in data sets of this kind.

     Subject headings: cosmology: observations -- intergalactic medium --
                       quasars: absorption lines