3 Discussion

3.1 The clustering of dusty-SF EROs

The clustering of dusty-SF EROs is small, and maybe consistent with the values of $1\lesssim r_0/($h^-1 Mpc $) \lesssim 2.5$ measured for star-forming galaxies at $z\sim1$ (Le Fevre et al. 1996; Carlberg et al. 1997; Hogg et al. 2000). This would suggest the former to be a subclass of the latter, but with stronger dust extinction. Locally, dusty galaxies detected by IRAS are also known to have a relatively weak clustering (e.g. Saunders et al. 1992).

The low level of clustering seems also to be at odds with the idea that the dusty-SF EROs are in a starburst phase following a major merger event, eventually expected to produce an elliptical galaxy, as in this case one would expect to find a correlation length somewhat lower than, but similar, to that of the ellipticals at the same redshift.

SCUBA sub-mm selected sources are also thought to be dusty objects at high redshift detected by virtue of the emission from dust warmed by star-formation or AGN activity. This population is expected (Magliocchetti et al. 2001) and tentatively observed (Scott et al. 2001) to show strong angular clustering at the level of $A(1^{\rm o})\sim0.01$ (see also Ivison et al. 2000; Chapman et al. 2001). Therefore our result suggest the dusty-SF EROs are a different population with respect to SCUBA galaxies, with small overlap, in agreement with the latter being typically fainter and more distant (K>20, median redshift $z\sim2.5$-3, Smail et al. 2000, see also C02, Mohan et al. 2001, and Dannerbauer et al. 2002 for MAMBO sources).

Finally, we note that the class of dusty-SF EROs could be internally inhomogeneous: some of them maybe spirals with moderate extinction (e.g. van Dokkum & Stanford 2001) and there could also be a mixture of dust-enshrouded AGNs and starburst galaxies (C02).

3.2 Field $\mathsfsl{z\sim 1}$ early-type galaxies

Observations of samples of faint ERO galaxies have led to three key conclusions regarding bright early-type galaxies with $L\gtrsim L_*$, at $z\sim1$. Firstly, their space density is consistent with that of local luminous early-type galaxies, when account is made of minimal pure luminosity evolution (PLE) (C02). Secondly, spectroscopy implies age $\gtrsim$3 Gyr for their stellar populations (assuming solar metalicity, C02); and thirdly, a comoving correlation length $r_0\gtrsim 12$ h^-1Mpc (this paper and D01) has been measured comparable with the local value for luminous early-type galaxies.

A large correlation length, $r_0 \gtrsim 10$ h^-1 Mpc, is anticipated theoretically for the hierarchical merging paradigm for which a rapidly increasing bias is predicted for massive galaxies by $z\sim1$(e.g. Mo & White 1996; Moscardini et al. 1998). Such a large correlation length is not expected for the PLE (galaxy conservation) scenario (D01). However, also current semi-analytical renditions of the hierarchical models seem to be at odds with the observed results. For example, the Cole et al. (2000) model predicts a comoving density (Fig. 1 of Benson et al. 2001) of all the $z\sim1$ galaxies with $10^{11}~M_\odot$(consistent with our $K\leq 19.2$ selection) which is a full order of magnitude below the density of just the old EROs observed by C02. Similarly, the Kauffmann et al. (1999) model predicts a comoving density of $z\sim1$ EROs ($R-K\geq5$, $K\leq 19.2$) that is 3(6) times lower than observed by C02 for old(all) EROs. In addition, in these models $z\sim1$ galaxies qualified as field early-types appear to have experienced recent star-formation, while the present sample of old EROs is dominated by an old stellar population. We conclude that to our knowledge no semianalytical rendition of the hierarchical merging models can yet account for all the 3 key observed properties of $z\sim1$ field early type galaxies described above.

Acknowledgements

We thank the referee, Rob Ivison, for constructive remarks, and Huub Röttgering and Sperello di Serego Alighieri for useful comments to the manuscript.