THE SPATIAL CLUSTERING OF FAINT GALAXIES

We estimate the spatial clustering of faint galaxies in an extensive new redshift survey performed using the automatic fibre positioner (Autofib) on the Anglo-Australian Telescope. The survey comprises 1100 galaxies within a magnitude range of 17 < B< 22 selected from 33 deep pencil beams. The redshift distribution extends to z almost-equal-to 0.5 and has a median redshift of z = 0.16, making the survey the most extensive and deepest of its kind for which the spatial clustering has been estimated. Over the apparent magnitude range of the survey, the galaxy number counts exceed the no-evolution prediction by a factor similar to the proportion of galaxies observed with strong [O II] emission. Adopting the hypothesis that the strong [O II] emitters can be statistically identified with the blue star-forming excess population, we investigate variation in the galaxy correlation functions between strong and weak O II populations, as well as with redshift. All the data are adequately described by a correlation function xi(r) = (r/r0))-1.7 with r0 = 6.5 +/- 0.4 h-1 Mpc. We find no evidence for evolution of the comoving correlation length with redshift: r0(z)= r0(0)(1+7)epsilon, with epsilon = 0.4 +/- 1.6. The correlation lengths for strong and weak [O II] emitters are compatible with the global value, albeit with moderately large uncertainties. A better constraint is provided by cross-correlation of the two populations which yields r0 = 6.2 +/- 0.5 h-1 Mpc. To B = 22, there is no evidence that the star-forming blue galaxies represent a new unclustered population. On the contrary, the physical association of blue galaxies with the normal population is as expected in the merger hypothesis put forward to explain the excess galaxy counts.