EVOLUTION IN THE CLUSTERING OF GALAXIES TO R=26

We present results for the two-point angular correlation function of galaxies to a limiting magnitude of r=26. The final sample is 97% complete to r=26.0, yielding 5730 galaxies over a 90.1 sq. arcmin field. The correlation function for our faint galaxies can be parameterised by a power law, $A \theta^{-0.8}$, in agreement with the clustering statistics of shallower catalogues. The derived amplitude, $A$, is small, but non-zero. We combine this measurement with the latest statistical constraints on faint galaxy redshifts from gravitational lensing studies, which imply that the bulk of the r<26 field galaxies should be at redshifts of order 1. Our derived $A$ is significantly lower than that predicted from the local bright galaxy correlation function using the lensing-determined galaxy redshift distribution and modest growth of clustering. This simplistic model does not include the variation in observed morphological mix as a function of redshift and apparent magnitude in our sample. At our faintest limits we reach sufficiently high redshifts that differential $K$ corrections will result in the observed galaxy mix being dominated by star bursting dwarf and low surface brightness irregulars, rather than the early-type systems used to define the local bright galaxy correlation function. Adopting the correlation function measured locally for these low surface brightness galaxies and assuming modest clustering evolution, we obtain reasonable agreement between our model and observations. This model supports the scenario in which the high number density of faint galaxies is produced by normally clustered star forming dwarf galaxies at modest redshifts.