4.3 The Gravitationally Lensed Pairs

Statistical studies of the distribution of faint objects in the fields of A370, Cl0024+1654 and Cl2244-02 have shown marginal evidence for an excess of faint pairs having the same magnitudes and color indexes (B - R) and (B - I). These pairs are located in the central part of the lensing cluster, they have separations in the range of 2 to 10 arcsec and their angular sizes are comparable to the seeing disk (Mellier et al. 1994). Although no firm conclusions could be reached with the data available, it was suspected that some of these pairs could be magnified images of faint background galaxies with a small angular size. In fact Kneib et al. (1993) succeeded in identifying three individual lensed-pair candidates in the field of A370 that were used as local constraints for the position of the critical lines (Fig. 6). The best test for recognizing pairs (very small multiple images) is to look at their geometry. They should be almost aligned and should display a typical parity inversion between the two components (mirror-like effect). This was observed for the brightest pair B2-B3 in A370 which provides a decisive constraint for the modeling. Smail et al. (1993) also noticed a beautiful lensed pair in the HST images of AC114 from the clear parity change of the two components. Indeed the search for faint pairs or multiplets will be extremely useful in the future for providing strong additional constraints on the modeling of cluster lenses and also for probing very distant and faint galaxies. Miralda-Escudé & Fort (1993) extrapolated the galaxy counts from Tyson (1988) up to B = 29.5 and found that ~ 1000 gal/arcmin² could be expected. For typical cluster lenses about 10 close pairs brighter than B = 27 could be observed. However, the sources should be small galaxies, otherwise we would see a pair of arclets instead of almost point-like pairs. These pairs could be a new class of distant young galaxies (Miralda-Escudé & Fort 1993).

Figure 13. CCD images obtained at CFHT of a typical fold arc in Cl2244-02. The break in Cl2244-02 separates the two merging images of opposite parity. Note the bright spot which is visible between A1 and B1. This spot is observed on many frames and is real; but since it is not seen on the counter image it is a good candidate for minilensing by a small perturbation located close to the arc (see part 4.5).

Indeed a small number of randomly distributed pairs can be found in images of the sky where there is no cluster. Such fields will be referenced hereafter as blank fields according to Tyson's definition (Tyson 1988). We attempted to estimate the number of excess pairs observed in some rich cluster-lenses from a comparison with a blank field. In Table 3, we summarize our preliminary findings. We restrict our analysis to the following sub-sample:

- Maximum angular separation between the objects: < 8"

- Limiting magnitude: B 26.5

- Photometric accuracy in magnitude: B₁ = B₂ ± 0.2

- Photometric accuracy in color: [(B - R); (B - I)]₁ = [(B - R); (B - I)]₂ ± 0.35 for the two members of each pair.

This deep preliminary investigation of lensed pairs is still uncertain and it is quite clear that the deep images that will be soon obtained with the HST will clarify the situation. On one hand A370 and Cl0024+17 seem to have a real excess of pairs, whereas the dense cluster Cl2244-02 does not. Perhaps this is due to the smaller angular scale of Cl2244-02, which reduces the impact parameter for lensed pairs. In any case it is possible to predict that the study of gravitational pairs or small multiplets will be a promising new observational field both at the statistical and individual level. Such distant pairs can be found at larger distance than giant arcs (~ 100-400 h₅₀^-1 kpc), on critical lines corresponding to large z sources. Occasionally they can appear as radial pairs when they are generated by the external caustic of the lens. This new class of object may well have a decisive impact on lens modeling.