In our radio-selected lenses, even without the detection of the lensing galaxy, there can be no doubt that one is dealing with gravitational lens systems, because of radio spectral index and radio/optical similarities of the images. The results we have obtained suggest strongly that lensing galaxies have and thus are not dark. Twelve out of twelve of the JVAS/CLASS lensed systems have ratios suggestive of normal luminous galaxies. Even if all the remaining 5 candidates were lensed systems, and all of these had no detectable lensing galaxy, this would still leave over 60% of lens systems with detected, and relatively normal, lensing galaxies compared to 25% reported by Hawkins (1997).
There is a possible explanation for the difference between our results and those of Hawkins. This is that the Hawkins sample has a fundamentally different distribution of image separations, as it contains only systems with separations greater than , which, since the separation s is a function of lens mass M (e.g. Schneider et al. 1992), corresponds to relatively massive lensing galaxies (, see Table 1). Could only massive galaxies be dominated by dark matter?
Let us concentrate, therefore, on radio lensed systems with separation 2 arcsec. We will further restrict our consideration to those incontrovertible lens systems in which optical counterparts to the radio components have been detected. There are two such systems with separation 2 arcsec in the JVAS/CLASS sample 2. There are two further known radio-selected systems with ; B0957+561 (from Hawkins' sample) and B2016+12 (Lawrence et al. 1984). In all four of these normal (not dark) lensing galaxies are detected. Thus amongst confirmed lens systems with separations there is no evidence for dark lenses.
There is no reason why any bias should be introduced by the use of radio selected systems; it is simply a way of securely identifying lens systems without the necessity of identifying a lensing galaxy. The discrepancy between the results for radio-selected lens systems, even for large separation lenses, and the mostly optically-selected sample of Hawkins (1997) is therefore puzzling.
We note, however, that Hawkins only considers systems with 2 images. Among the radio-selected systems there are many 4-image systems. Could the dark lenses be found only in the 2-image systems? Kochanek et al. (1997) pointed out that in general, one would expect a greater preponderance of 4-image systems where the lensing mass distribution, particularly the halo, is more elliptical (e.g. King & Browne 1996), and that one might therefore postulate that `dark lenses' have spherically symmetric mass distributions. This, however, fails to account (Kochanek et al. 1997) for why the existing radio-selected two-image systems do not have dark lenses, as well as for why we find no two-image systems with dark lenses in JVAS/CLASS.
In summary, there are two conditions that must be met in order to reconcile our observations with the idea of dark lenses. First, in order for us to find no dark lenses in the JVAS/CLASS sample, there must be a sharp cutoff such that all dark lenses must have masses of . Second, in order to explain the detection of lenses in the radio lens systems with separations 2 arcsec (mostly quads), dark lenses should have much rounder mass distributions than the luminous lenses (Kochanek et al. 1997). The alternative to these two conditions is to suppose that most, or even all, of the systems discussed by Hawkins in which no lensing galaxy is found, are indeed not lens systems. This is despite the fact that in each individual case we find the arguments persuasive that the quasar pairs are not random associations. However, if there are no dark lenses, the frequency of close physically associated pairs of quasars has hitherto been greatly underestimated.
© European Southern Observatory (ESO) 1998
Online publication: May 15, 1998