2004ApJ...608...25C

Surveys for gravitational lens systems have typically found a significantly larger fraction of lenses with four (or more) images than are predicted by standard ellipsoidal lens models (50% versus 25%-30%). We show that including the effects of smaller satellite galaxies, with an abundance normalized by the observations, significantly increases the expected number of systems with more than two images and largely explains the discrepancy. The effect is dominated by satellites with ∼20% the luminosity of the primary lens, in rough agreement with the typical luminosities of the observed satellites. We find that the lens systems with satellites cannot be dropped from estimates of the cosmological model based on gravitational lens statistics without significantly biasing the results.