Ly-
and metal line systems are
clearly distinguished in terms of N(HI). Since
HI is the only species detected in the
Ly- systems, and it is
ubiquitous in metal
line systems, N(HI) is the only column density that can be used for a
comparative
study of the different narrow lined systems. As we saw above, this is
unfortunate
since it would be much more useful to classify systems in terms of their
total column densities, or better still, their masses.
The first determination of the distribution of N(HI) in the
Ly- systems by
Carswell et al. (1984)
revealed a roughly power law distribution. This has now been
extended to include a hopefully representative small sample of all
narrow lined systems
(Tytler 1987a).
The most impressive feature of the column density distribution is the fact that
the narrow lined systems cover an enormous range of logN(HI), from below
12.8 for
the weakest Ly- systems, up to
21.9 for metal systems
(Smith et al. 1986).
This 9 order of magnitude spread is comparable to the difference in
luminosity between
the white dwarf stars and the supergiants. Here then is the clearest
evidence for
diversity amongst the absorption systems, a diversity which encourages
one to search for subclasses of systems.
It then came as a surprise to find that to first order the column
density distribution
could be represented by a single power law. It appears that the majority
of systems
with logN(HI) 
17 are
Ly- systems, while those with
larger N(HI) are predominantly
metal line systems. If the power law fit to the
Ly- systems is
extrapolated up to the
larger column densities of the metal line systems, one obtains a good
estimate of the
frequency of occurrence of the latter. These preliminary results suggest
that there is some commonality between the
Ly- and metal line systems, as
has been noted by
Bergeron and
Boissé (1984).
New data now show that the power law slope flattens at both low column
densities
below logN(HI) 
14.35
(Carswell et al. 1987),
and also above
logN(HI) 20
(Smith et al. 1986).
In addition, the distribution remains very poorly constrained at other
N(HI). However a number of interesting questions are suggested. Why are
there so
few Ly- systems with logN(HI)
17? Why are metal line systems
with logN(HI)
17 extremely rare, and could possible connections between the
Ly- and
metal line systems account for these observations?
The simplest response is that the
Ly- are really metal line systems
which have too
low a total column density to show metal lines. For all reasonable
levels of ionization,
metal lines would be observed if total column densities exceed
1018.3 for abundances greater than 0.01 solar
(Chaffee et al. 1986).
For a typical logN(HI) = 14, this limits
ionization to n(H+) / n(H)
104.3. Those
Ly-
systems which have larger N(HI) must
have lower ionization or lower metal abundances.