In both the C and D configuration maps, an extended, negative source was immediately apparent, about 25 from the image phase center. At full 6 resolution, this feature was at the -5.5 level. At lower resolution, where it was less resolved, it was more prominent: it had a peak amplitude of or at resolution, for instance. An image at 30 resolution is shown as Fig.2. We performed a variety of tests to determine whether such a feature could be produced by instrumental effects. These are fully described in Richards et al. (1997). No instrumental explanation was found. Monte Carlo calculations were used to see if such a feature could be produced by the chance superposition of the side lobes of the positive sources in the map: the answer is no. The amplitude and angular scale of this feature () are consistent with the explanation that it is the Sunyaev--Zel'dovich signal from a cluster of galaxies. If this explanation is correct, then, as shown by Fig.2, we need to account for the asymmetry of the profile of the decrement.
Figure 2: The ``Black Cloud'' at 30 arcsec resolution. Radio contours at , times the rms noise are overlaid on the radio image (grey scale).
The HST exposure on this field are not particularly deep: for instance, if we assume a cluster at a redshift of 2.56, only galaxies with at a rest frame wavelength of 2300Å will be visible. The MDS did reveal two quasars separated by Mpc with and . We suggest that these quasars may be associated with a cluster of galaxies, otherwise invisible in the MDS frame. Limits on the X-ray flux can be set from a deep Rosat HRI image of an adjacent field taken by Hu & Cowie (1996). The 3 detection limit is . A cluster with the properties suggested by the amplitude of the microwave decrement and reasonable assumptions about physical parameters like temperature and core radius would produce a flux below this limit if the cluster were at . Thus our failure to detect an X-ray source in the vicinity of the microwave decrement does not rule out the S--Z assumption. What we can say is that a cluster at a redshift would have been visible in the Rosat image.
As noted, the Cambridge group will provide evidence here of a rather similar decrement in the CMBR, again in the vicinity of a pair of high redshift quasars. If these two ``holes'' in the CMBR are interpreted as high redshift clusters, there are some interesting consequences for models of structure formation. In the general class of CDM models, where large structure forms late, it would be surprising to find a thermally relaxed cluster, with or without galaxies, at a redshift as high as 2.5. Detailed comparisons with models of secondary fluctuations produced by the S--Z effect in anonymous clusters appear in Richards et al. (1997). Here, I wish simply to remark that the observations described in this brief paper set two kinds of constraints on such models: first, we appear to have robustly detected an S--Z signal from a high redshift cluster; and, on the other hand, we are able to set rather stringent upper limits on the overall, rms, fluctuations introduced by a foreground of such clusters. The conclusions here and in Richards et al. (1997) must remain tentative until both the observations and the explanation are tested. We plan further observations at radio, mm and ir wavelengths to test the model given here. Nevertheless, our interpretation is strengthened by the rather similar results reported by the Ryle group, Jones et al. (1997) and Saunders et al. (1997). True, two detections do not make a population, but the two decrements look remarkably similar. Are we discovering a new and interesting population of high redshift objects? Perhaps we are. If so, given the attention devoted to them at this meeting held in Cambridge, I would propose that we honor a famous figure in astronomy at Cambridge, Sir Fred Hoyle, by naming these dark absorbing/scattering sources ``Black Clouds,'' in honor of his very fine and very entertaining science fiction novel published forty years ago.