Astron. Astrophys. 336, 445-454 (1998)

7. Conclusions

We have discussed in detail the errors associated to redshift determinations in the ESO Slice Project (Vettolani et al. 1997). We have found a systematic difference between absorption and emission line velocities of km/s and we have shown that the same effect is present in the Durham/Anglo-Australian Telescope faint galaxy redshift survey (Broadhurst et al.1988). In the case of the ESP, we have excluded problems of zero-point error or calibration. Such a discrepancy has not been detected in large shallower surveys.

Shectman et al. (1996) briefly discuss a similar effect for the Las Campanas Redshift Survey, which they have corrected by using a different template for emission line galaxies.

We have generalized the suggestion by Shectman et al. (1996), who identify as the main cause of the discrepancy the systematic difference between the absorption line spectra of the standard templates and the typical emission line galaxy, particularly the blend between the Ca II H and lines, an effect already discussed by Sandage (1978). We find in fact systematic effects even from template to template, apparently due to the way each template fits the galaxy spectra; this implies that the choice of the template significantly affects redshift measurements.

For the ESP data, we decided to use the best-fit template (i.e. the one giving the smallest error), as using a different template might introduce unknown biases in the redshift measurement.

In the lack of a definitive explanation, the common assumption that the true galaxy redshift is given - on average - by the emission lines, is plausible, but not proven. It should be verified that the bias cannot be due, at least partially, to emission lines, and that the sample is not biased, for some reason, towards galaxies with outflows. One can for example speculate that other factors may contribute to the effect, such as sampling of different parts of the galaxies, the different mix of morphological types, and evolution with look-back time. A collection of high resolution data of a sample of galaxies and of different templates will be necessary to give a definitive solution.

We feel it is important to stress the existence of such an "anomaly": in view of future, large surveys, the templates used should be carefully checked and made publicly available ², as already done by the LCRS group, to discover and quantify any systematic difference. Even if the amplitude of the effect is not large, it is quantitatively more important that typical zero-point shifts, and it is significant enough to affect for example measures of velocity dispersions and galaxy peculiar velocities, or the interpretation of results for very distant galaxies, as those which are reported by Cowie et al. (1995) and Steidel et al. (1997).

© European Southern Observatory (ESO) 1998

Online publication: July 20, 1998
helpdesk.link@springer.de