![]() | Annu. Rev. Astron. Astrophys. 2000. 38:
667-715 Copyright © 2000 by Annual Reviews. All rights reserved |
The HDFs are exquisitely deep and
sharp images, detecting thousands of objects distributed throughout
the observable universe, but they are also very small fields of
view, and each is only one sightline. Figure 2
illustrates some parameters relevant for studying galaxy evolution with
a single HDF. The total co-moving volume out to redshift z has
been scaled (top panel) by the present-day normalization of the
galaxy luminosity function
*.
This gives a rough measure of the number of
"L* volumes" out to z,
i.e. approximately the number of L*
galaxies expected in that volume, or at high redshift the number of
L* galaxies that the "proto-galaxies" found
there will someday become. At z < 1, this number is ~ 10 - 30
depending on the cosmology: very few high-luminosity galaxies
are expected (or found, for that matter), and even purely
Poissonian variations introduce large uncertainties in any
statistical conclusions that can be derived from them.
Given real galaxy clustering, these uncertainties are still
greater. As an example, ~ 24% of the total rest-frame
6500Å luminosity summed over all HDF-N galaxies out to
z = 1.1 comes from just four galaxies: two in a redshift
"spike" at z = 0.96 and two in another spike at z = 1.02.
The safest use (statistically) for the HDF at z < 1 is thus
to study the vastly more numerous, low-luminosity galaxies.
This caution similarly applies to clustering studies. The angular
correlation functions derived from the z
1 sample
primarily refer to low-luminosity galaxies, whereas those at higher
z refer to higher luminosities. Clustering variation with
luminosity (or mass) can mimic evolution.
At high redshift, the HDF volume is large (especially for the open and
cosmologies), and
thus is more likely to provide a fair sample of objects.
2
For (
M,
) =
(0.3, 0.7) there is ~ 20 × more volume at
2 < z < 10 than at 0 < z < 1, and the
likelihood of finding even moderately rare objects becomes
significant. There is, however, little time out at high redshift:
Most of cosmic history takes place at z < 1
(Figure 2, bottom).
2 Even at high redshift, however, uncertainties due to clustering (cf. [Adelberger et al. 1998]) need to be kept in mind when analyzing one or two sightlines like the HDFs. Back.