We have carried out a study to simulate distant clusters of galaxies in deep ground-based optical images. We find that when model galaxies are added to deep images obtained with the William Herschel Telescope, there is considerable scatter of the recovered galaxy colours away from the model values; this scatter is larger than that expected from photometric errors and is significantly affected by confusion, due to ground-based seeing, between objects in the field. In typical conditions of $\approx$ 1-arcsec seeing, the combination of confusion and incompleteness causes a considerable underestimation of the true surface density of $z \approx 3$ galaxies. We argue that the actual surface density of $z \approx 3$ galaxies may be several times greater than that estimated by previous ground-based studies, consistent with the surface density of such objects found in the HDF.

We present deep optical images of the PC1643+4631 field obtained at the WHT. This field contains two quasars at redshifts z=3.79 & 3.83 and a cosmic microwave background (CMB) decrement detected with the Ryle Telescope. The images are in U,G,V,R and I filters, and are complete to 25th magnitude in R and G and to 25.5 in U. The isophotal galaxy counts are consistent with the results of Metcalde et al. (1996), Hogg et al. (1997), and others. We find an excess of robust high-redshift Ly-break galaxy candidates with 25.0<R<25.5 compared with the mean number found in the fields studied by Steidel et al. -we expect 7 but find 16 - but we do not find that the galaxies are concentrated in the direction of the CMB decrement. However, we are still not sure of the distance to the system causing the CMB decrement. We have also used our images to compare the commonly used object-finding algorithms of FOCAS and SExtractor: we find FOCAS the more efficient at detecting faint objects and the better at dealing with composite objects, whereas SExtractor's morphological classification is more reliable, especially for faint objects near the resolution limit. More generally, we have also compared the flux lost using isophotal apertures on a real image with that on a noise-only image: recovery of artificial galaxies from the noise-only image significantly overestimates the flux lost from the galaxies, and we find that the corrections made using this technique suffer a systematic error of some 0.4 magnitudes.

In an investigation of the cause of the cosmic microwave background decrement in the field of the z = 3.8 quasar pair PC1643+4631, we have carried out a study to photometrically estimate the redshifts of galaxies in deep multi-colour optical images of the field taken with the WHT. To examine the possibility that a massive cluster of galaxies lies in the field, we have attempted to recover simulated galaxies with intrinsic colours matching those of the model galaxies used in the photometric redshift estimation. We find that when such model galaxies are added to our images, there is considerable scatter of the recovered galaxy redshifts away from the model value; this scatter is larger than that expected from photometric errors and is the result of confusion, simply due to ground-based seeing, between objects in the field. We have also compared the likely efficiency of the photometric redshift technique against the colour criteria used to select z>3 galaxies via the strong colour signature of the Lyman-limit break. We find that these techniques may significantly underestimate the true surface density of z>3, due to confusion between the high-redshift galaxies and other objects near the line of sight. We argue that the actual surface density of z=3 galaxies may be as much as 6 times greater than that estimated by previous ground-based studies, and note that this conclusion is consistent with the surface density of high-redshift objects found in the HDF. Finally, we conclude that all ground-based deep field surveys are inevitably affected by confusion, and note that reducing the effective seeing in ground-based images will be of paramount importance in observing the distant universe.

We present the first results of a deep imaging programme to identify the system responsible for the Cosmic Microwave Background decrement in the field towards the z = 3.8 quasar pair PC1643+4631 A&B. Using the prime focus camera at the William Herschel Telescope we have carried out deep multicolour optical imaging to search for candidate cluster galaxies at extremely high redshift. Using UGR colour selection we find the surface density of z > 3 Lyman-break galaxy candidates is at least as great as that found in the field of the z = 3.1 structure discovered by Steidel et al. (1998), and may be somewhat greater.

In a companion paper (Paper I: Jones et al. 1996) we report the discovery of a cosmic microwave background decrement, indicative of a distant cluster with mass ~10^{15} solar masses, towards the quasar pair PC1643+4631A&B (z=3.79,3.83, separation 198''). To search for the cluster responsible, we have obtained R-, J- and K-band images of the field and have also carried out optical spectroscopy of selected objects in it. No such cluster is evident in these images. Assuming the cluster causing the decrement is similar to massive clusters already known, our magnitude limits imply that it must lie about or beyond z=1. This provides independent support for the X-ray-based distance argument of Paper I. The cluster must gravitationally lens objects behind it; for a cluster z around 1-2, the Einstein-ring radius for sources at z ~= 3.8 is ~ 100''. Simple modelling, producing simultaneously the S-Z effect and the lensing, shows that the source positions of quasars A and B lie within ~10'' of each other and may indeed be coincident. The two quasar spectra are found to be remarkably similar apart from their one-percent redshift difference. Assuming A and B are images of a single quasar, we present a possible explanation of this difference.

In a 15-GHz Ryle-Telescope observation of PC1643+4631 A & B, a pair of quasars at redshifts z = 3.79 and 3.83 separated by 198 arcsec on the sky, we find a decrement in the cosmic microwave background (CMB) of -380 +/- 64 uJy in a 110 arcsec x 175 arcsec beam. Assuming this to be a Sunyaev-Zel'dovich effect due to an intervening cluster, the minimum magnitude of the central temperature decrement is 560 uK. A serendipitous ROSAT observation shows that there is no X-ray-luminous cluster in the direction of the decrement at z < 1. The implied gas mass is >~ 2 x 10^{14} solar masses (assuming a temperature of ~ 5 keV), indicating a total mass of > 10^{15} solar masses. This result demonstrates the existence of a massive system too distant to be detected by its emission, but which can be found via its imprint on the CMB.