HS 1216+5032 was discovered in the course of the Hamburg Quasar Survey (Hagen et al. 1995). It was later confirmed as a double QSO at by Hagen et al. (1996) through direct images and low-resolution (FWHM Å) spectroscopy in the optical range. The B magnitudes of the bright and faint QSO images (hereafter "A" and "B", respectively) are and . The separation angle between A and B is , and the emission redshifts are and . From the available optical data of HS 1216+5032 AB there appears to be some evidence favoring its physical-pair nature instead of a gravitational lens origin (this subject will be discussed in Sect. 6.1), so the projected distance 1 between LOSs as a function of redshift is known ( kpc at ).
This paper presents ultraviolet (UV) spectra of QSO A and B taken with the Faint Object Spectrograph (FOS) on-board the Hubble Space Telescope (HST ). The prime goal of these observations was originally to investigate the Ly forest in the little explored redshift range between and . In particular, the projected separation between the LOSs to HS 1216+5032 A and B is convenient because it samples well the expected Ly cloud sizes of hundreds of kpc.
For the redshift range accessible from the ground, a pioneering work on cloud sizes was made by Smette et al. (1992) using the gravitationally lensed double QSO UM 673 () with known lens geometry. All observed Ly lines were observed to be common to both spectra and a statistical lower limit of kpc could be set for the transverse sizes (spherical clouds, no evolution). A similar result (Smette et al. 1995) was found for HE 1104-1805 AB () but here the position of the lens was unknown. A transverse size of kpc was estimated for .
Surprisingly, QSO pairs at very large projected distances seem to still show Ly lines common to both spectra. In the spectra of LB 9605 and LB 9612 (), Dinshaw et al. (1998) find 5 such lines within 400 km s-1 and derive a most probable diameter of 1520 kpc at . If the lines arise in different absorbers, however, an upper limit of kpc is derived. Further studies in the optical range have been carried out by Dinshaw et al. (1994; , kpc at ) and Crotts et al. (1994; same results).
At low redshift (), Ly clouds show a very flat redshift distribution: only Ly lines with Å are expected in the wavelength interval covered by the FOS on the HST (Weymann et al. 1998). Therefore, the size-estimate uncertainties for low-redshift Ly absorbers are large. Studies using FOS spectra of QSO pairs have been made by Dinshaw et al. (1997; , kpc at ), and more recently by Petitjean et al. (1998; , kpc at ).
All the aforementioned limits on Ly cloud sizes result from rather crude models which assume non-evolving and uniform-sized absorbers. Such simple models might be unrealistic, as suggested for example by a trend of larger size estimates with larger LOS separation (Fang et al. 1996) or by the claim (Fang et al. 1996; Dinshaw et al. 1998) that the cloud size increased with cosmic time. However, we note that none of the above claims is confirmed by D'Odorico et al. (1998), using a larger database. Clearly, the issues of geometry and size of Ly absorbers remain open due partly to a lack of size estimates at low redshift.
In this paper we use a likelihood analysis (Sect. 6) that attempts to constrain the size of Ly clouds in front of HS 1216+5032 A and B based on simple considerations on shape and evolution. We model the absorbers either as non-evolving spheres or as filaments. The technique takes advantage of the information provided by line pairs observed in both spectra at the same redshift, hereafter referred to as "coincidences", and Ly lines observed only in one spectrum, referred to as "anti-coincidences". A likelihood function is then constructed using an analytic expression for the probability of getting the observed number of coincidences and anti-coincidences in each of the two geometries.
In general, modelling the geometry of Ly absorbers at low redshift on the basis of line counting is made difficult because the line samples are intrinsically small, implying a poor statistics. In addition, HS 1216+5032 B shows broad absorption lines (BALs) caused by several ions with transitions in the UV, so the effective redshift path for the detection of Ly lines is even shorter.
Intervening metal absorption lines of interest are presented in Sect. 4. The BAL systems are described in Sect. 5. The conclusions are outlined in Sect. 7.
© European Southern Observatory (ESO) 2000
Online publication: May 3, 2000