As a first guess, heavy-element systems have been identified on the basis
of the line list presented by GCFT. Eventually, other element lines have
been added to the previously found systems, making use of a list of the
lines most frequently observed in QSO absorption
spectra, derived from Morton (1991).
The oscillator strengths of Si II 
and Si II
have been corrected according to Verner et al. (1996).
As for the Ly
lines, the operation of fitting has been carried out within
the context FITLYMAN of the reduction package MIDAS. In the case of the
complex profiles observed for the heavy-element lines, a minimum number of
components needed to obtain a 
and a good fit was deduced from the
C IV profiles (or, in one case at low redshift, from the Mg II one).
Then an identification programme, based on the method of Young et al. (1979), has been applied to the newly observed lines that seemed not to belong to existing systems.
Besides the systems previously found at lower resolutions (Young et al. 1979;
Sargent et al. 1988; GCFT), we show two systems identified on the
basis of the C IV doublet, whose column densities and equivalent widths were too
weak to be detected in previous observations. We do not confirm the existence
of the system at z=2.33 found by GCFT, whose lines were
all inside the Ly forest. Six of the twelve systems show a
multicomponent substructure, with a velocity extent up to 
km
s-1. We have not adopted any velocity window, or other strict rules for
the classification into systems versus sub-systems. We have simply
considered as one system those groups of lines for which a simultaneous fit
was required because of the superposition of the various line profiles.
Each system has been assigned a letter, from A to L, in order of increasing redshift.
The two strongest Ca II doublets of this system have been already observed
by GCFT. Furthermore, two other components have been fitted, together with the
respective Mg II doublets. The two components with higher column density
values show also Mg I 
.
The total spread in velocity of the system is
.
The existence of this system was formerly suggested by Young et al. and subsequently confirmed by Sargent et al. (1990) and by GCFT.
The C IV 
line at 
is blended with the
Si IV 
of the complex system H, but the other line of the
doublet has a very clean profile that guarantees the reliability of the fit.
We have found another component by means of the C IV doublet, which is
slightly separated (
km s-1) from the others. The maximum
separation among the three is 
km s-1.
Corresponding to two lower redshift components we have fitted the Si IV
doublets and at also Si II
, Si II 
and Al II 
.
This system, identified by Sargent and collaborators
(1988), shows two C IV doublets. Relative to the stronger component, we
observe an uncertain Fe II 
and the Ly, both in a
region with low s/n ratio. The Si IV doublets, if present, are blended
with the stronger C II complex of system E.
We have identified this low column density C IV doublet in a region with high s/n ratio and absence of lines. This new system does not show any other metallic line associated with it.
This multicomponent system shows ten sub-features, three more
than in GCFT, with a total spread in velocity of 
km s-1.
The number
of components and the relative redshifts have been derived from the
observation of the C IV lines which lie outside the Ly forest.
Lines shortward of the Ly emission are difficult to examine because
they are blended with H I lines. In these cases (e.g. C II
) we
have identified and fitted any components in coincidence with C IV
components. Nevertheless, the b and N values of these lines, and even
their existence, are to be considered doubtful.
In addition to the elements found by GCFT, we have fitted Si III
, Si II and 
, all in a
region of the spectrum with low s/n ratio.
The corresponding Ly line has been fitted separately. As a matter
of fact, it was so strongly saturated that even the observation of the
Ly
did not help in the determination of its profile.
A clear C IV doublet, first identified by Sargent et al. (1988), is
observed
outside the Ly forest; all the other lines of this system lie inside
it. At the same redshift of C IV, Si IV doublet and N V doublet
have been found.
Si IV doublet has been fitted using Si IV 
only, since
Si IV is blended with a wide Ly line. Besides, it
shows a possible second component separated by
from
that identified by the C IV. A Si II is observed at this
last redshift, while the existence of the other lines found by GCFT is not
confirmed.
This system has been identified by the C IV doublet, which presents
two components separated by
.
Unlike in GCFT, Ly and Si IV doublet, even if extremely weak, have
been fitted. Other lines, whose identification is uncertain, are: Si III
blended with the multicomponent Si II of system H and
Fe II blended with a Ly line.
This is the most complex system found in our spectrum. It
appears as a blended feature of 
km s-1.
We observe seven C IV doublets and four possible very weak lines of
C IV . Four more components have been identified using
low-ionization lines falling outside the Ly forest, and all the lines
unambiguously identified in the Ly forest have been fitted.
Besides the lines fitted by GCFT, we have added those of Si II
, , 
, 
and the
fine-structure excited level C II .
No fine-structure line of Si II has been observed.
As for the other complex system, E, the Ly line has been observed but
not fitted together with the other elements. The presence of the
corresponding Ly line does not help much because of the strong
saturation.
Figure 6: The shape of the ultraviolet background at the epoch
z=2.9. Continuous line: as computed by
Haardt & Madau (1997); dashed
line: with an increased (of a factor 10) jump around 4 Ryd. The
ionization potentials of a few ions are shown
Meyer & York (1987) identified this system as a single C IV
doublet. We have added the identifications of O I 
, Si III
, Ly, Ly and Ly
, while the existence
of Si II , 
is uncertain due to severe blending.
None of the low-ionization lines reported by GCFT have been observed, apart
from Al II that has become a C IV at z=3.216542.
- J
This C IV system (Meyer & York 1987) shows also Si III
, Si IV 
, Ly Ly
and Ly
.
This is a Lyman-limit system identified by Sargent et al. (1990).
The Ly line of this system probably has a complex structure which
has not been possible to disentangle even examining the corresponding
Ly, still badly saturated.
At high resolution, the C IV identification is doubtful (a wavelength
discrepancy of about 0.5 Å is observed between the two components of
the doublet). We have identified and fitted a very weak Si IV
and Si III (possibly contaminated by Ly).
The existence of the N V doublet is not verifiable since a wide
H I line is present at that wavelength.
This possible system, previously unknown, shows a C IV
doublet (the C IV line was identified as Al II
at z=2.81959 by GCFT) and a Si IV doublet outside the Lyman forest.
We have fitted also the Ly line and Si III in
the forest.
For the systems I and J the reliability of the fit of the Lyman series
allows a measure of the metallicity. In system K it is possible
to put an upper limit to it on the assumption of 
.
We have used the standard photoionization code CLOUDY (Ferland 1996),
following the same approach described in detail in Savaglio et al.
(1997).
In the analysis of system J, if the standard UV background at
z=2.9 (Haardt & Madau 1996) is adopted and the size of the cloud is
assumed not smaller than 20 kpc, we end up with an unplausibly high
overdensity of silicon with respect to carbon, 
.
To obtain a more realistic 
the cloud size
should decrease substantially below 10 kpc.
Incongruities of this type are not uncommon,
as reported by Songaila & Cowie (1996) and
Savaglio et al. (1997).
Several processes may increase the
Si IV/C IV ratio: a photoionizing background dominated by local sources,
non-equilibrium temperatures and non-uniform radiation fields
(Giroux & Shull 1997). In
Savaglio et al. (1997) and here we have
explored the possibility of an enhanced break at 4 Ryd in the
metagalactic background radiation, as could be originated by
a contribution of primeval galaxies in addition to the standard QSO background.
With a jump increased of a factor 10 at 4 Ryd (Fig. 6 (click here)),
the observed line ratios turn out to be compatible, assuming a cloud
size of 30 kpc, with a metallicity -2.5 dex solar
and an overabundance of silicon with respect to carbon of a factor 5-6.
The analysis of systems I and K, under the assumption of the same
UV background with an enhanced jump at 4 Ryd and cloud sizes 
kpc, provides metallicities of 
and 
,
respectively. System I, again, requires an overabundance of silicon with
respect to carbon of a factor 4 that would become much larger
if the standard UV background is assumed.