The main result of the present study is the detection of a red fuzzy object located between the two components of HE 1104-1805. This result is one more very strong argument in favour of the lensed nature of this double quasar.
Wisotzki et al (1993) do not detect the lensing galaxy in R down to a limiting magnitude of 23-24. In Fig. 6 the tracks in the vs colour-colour diagramme of two galaxy types, an elliptical and an Sa galaxy, are plotted. They are plotted with and without evolution and for the redshift range . Also plotted is the range allowed by the observations in this paper and the optical observations of Wisotzki et al. (1993).
If we include the effects of evolution (thick lines in the figures), the IR colours are compatible with an elliptical galaxy (as shown by Fig. 5) between and (Fig. 4). The IR-optical colours are less compatible with this; however, the expected R magnitude of the lensing galaxy is , and this may have been difficult to see 1:001 away from the QSO which is 5 to 6 magnitudes brighter. In fact, a preliminary detection of the lensing galaxy by Grundahl, Hjorth & Sorensen (1995) allowed to measure an I -band magnitude of 20.6, in better agreement with our findings.
One of the two metallic absorption line systems found at and (Smette et al. 1995) could be produced by the lensing galaxy. In particular, the absorption system at is seen almost only in QSO A. Since the angular distance from the lens to QSO A is much smaller than to QSO B it seems reasonable to think that the lensing galaxy we detect is more likely to be at rather than 1.320.
Despite the depth of our IR images, which enables us to detect galaxies up to the redshift of the QSO, we do not detect any obvious overdensity of galaxies which could contribute significantly to the total gravitational potential involved in this system. However, two faint galaxies (G1 and G2) are detected close to the line of sight to the QSO. G1 has a colour of 1.1, while G2 has a colour close to that of the lens galaxy. These two objects could constitute an external source of shear, for example responsible for the misalignment between the lensing galaxy, QSO A and QSO B. However, one cannot exclude the simpler (but unlikely) explanation that the light and mass centroids of the lensing galaxy do not coincide.
We can infer from our deconvolutions that QSO A is not exactly compatible with a single point source. The deconvolution leaves significant residuals at the location of QSO A, even in J where the PSF is rather stable across the field. The signal-to-noise ratio and resolution of our observations do not allow to draw definite conclusions, but we suspect that image A is either not single or is super-imposed on a fuzzy faint light distribution.
From the geometry of the lensed system, given in Table 1, and assuming we see a galaxy at , the time delay we can expect between the two images of HE 1104-1805 is of the order of 3.5 years. We have assumed that the lens can be modeled as a Singular Isothermal Sphere (SIS) and that km s-1 Mpc-1. This large delay means that one measurement every second week would be enough to derive good light curves.
Assuming that an SIS is appropriate for the lensing galaxy, we derive a mass of , not too far above the masses expected for big elliptical galaxies. The apparent magnitude suggests that the lens is many times more luminous than a normal galaxy if the lens is actually at .
Finally, the magnitude difference between the lensed images is and , where the magnitude difference is taken as . The magnitude difference expected from the SIS model is 0.75 magnitude, but with the deflector angularly closer to the faint image than to the brighter one.
This could indicate that component B is reddened relative to A, or that component A is preferentially amplified (e.g. slight image splitting) relative to B and that this preferential amplification is more efficient in the blue. The latter hypothesis is more likely since the lens galaxy is angularly closer to QSO A than to QSO B and would therefore redden A more than B, assuming the reddening is due to the lens galaxy. On the other hand, the lensing potential might be more complex than a SIS (for example elliptical + core), in particular if G1 and G2 introduce a significant source of shear.
Wisotzki et al. (1995) showed that microlensing was acting on QSO A and that it was more efficient in the blue than in the red. The magnitude difference they observed in B in 1994 was , larger than our present values in J and K (although the quasar has probably varied between 1994 November and 1997 April). This suggests that microlensing is less efficient in the IR than in the visible. If the source quasar is found to be variable in the IR domain, an IR photometric monitoring of HE 1104-1805 may then minimize contamination by microlensing events and allow a better determination of the time delay than with optical data.
© European Southern Observatory (ESO) 1998
Online publication: January 8, 1998