4.4. Multiple lenses

Another problem associated with modeling lensing galaxies is illustrated by the system HE0230-2130. There is not one lensing galaxy, but two, as seen in figure 3. The second, fainter galaxy provides the quadrupole moment that makes this a 4-image system and not a 2-image system. When we see two galaxies separated by only a few kpc, we must wonder how the non-baryonic matter is distributed. Are the halos for the two galaxies distinct? Or have they merged? How many components should we model? And where should their centers be? In such a case something with elements of the [Williams & Saha (2000)] form-free approach might be preferable to a straightforward parameterized model.

Figure 3. The quadruple system HE0230-2130, observed with the Baade 6.5-m telescope through g' (left) and i' (right) filters. The two red objects are lensing galaxies. The image between the two galaxies is a saddlepoint of the time delay function.

The system with the best time delays is CLASS1608+656. [Fassnacht, Xanthopoulos, Koopmans, & Rusin (2002)] have measured four beautiful radio lightcurves. All four time series faithfully reproduce the many bumps and wiggles. The multiple delays in this case are good enough to provide meaningful additional constraints to the deflections and distortions. But modeling the system is not straightforward. As with HE0230-2130, there are two lensing galaxies (figure 4) that appear to be interacting. A dust lane encircles them both. Again the question of how the dark matter might be distributed looms as crucial. [Koopmans et al. (2003)] have worked exceedingly hard to constrain this system, measuring positions for the images, the shape of the ring, and measuring the velocity dispersion of the more massive lens. We can nonetheless imagine a devil's advocate coming up with a plausible dark matter distribution uncorrelated with the observed galaxies that gives a very different value for the Hubble constant.

Figure 4. The quadruple system CLASS1608+656, observed with the Hubble Space Telescope through the F814W filter. The lensing galaxies are marked G1 and G2. A dust lane can be seen encircling them.

The problem of multiple lenses can be serious even when one of the galaxies is very much smaller than the principal lens. Consider the case of RXJ0911+0554, shown in figure 5, where the primary lensing galaxy has a faint companion. Given its faintness and our aversion to adding additional parameters, we might be tempted to ignore it, as did [Schechter (2000)] in the model used by Courbin to construct figure 1. But allowing for a mass at the position of this dwarf companion changes the predicted time delay by 10%. Though the smaller galaxy is a factor of 10 fainter than the primary lensing galaxy, its effect is to move the center of mass closer to the midpoint of the images, decreasing the differences in path length.

Figure 5. The quadruple system RXJ0911+0554, observed by the CASTLES consortium with the Hubble Space Telescope through the F160W filter. The primary lensing galaxy G1 has a dwarf satellite G2. Allowing for the mass of the dwarf changes the derived value of H0 by 10%.