In his contribution to the present proceedings, Lutz WISOTZKI presents spectroscopic evidence for microlensing of quasars. This has important consequences (unfortunately negative) for time delay estimates of H0. It means that there are uncorrelated variations in the fluxes of images, especially in the optical continuum. The timescale for these fluctuations depends upon the size of the Einstein ring of the microlenses and upon the relative velocity of the quasar and the lensing galaxy. Taking the lenses to be a solar mass and the relevant velocities to be of order 300 km/s gives timescales of order 10 - 30 years. This ought not to matter for quasars lensed by galaxies, whose time delays are at most one year. But there have been repeated instances of uncorrelated variations in quasar lightcurves on considerably shorter timescales (e.g. [Burud et al. (2000)]), in some cases causing considerable dispute about which points ought and ought not to be included in a time delay measurement and about the nature of the fluctuations.
Seven years ago the present author prevailed upon Andrzej Udalski to monitor HE1104-1805 for a time delay with the OGLE telescope [Udalski, Kubiak, & Szymanski (1997)]. The OGLE group was at that time focussing exclusively on LMC and galactic bulge microlensing, but HE1104-1805 is at a very different right ascension. Three years of superb OGLE data were obtained in which no believable correlated variations were observed. But there were appreciable uncorrelated variations, most of which were attributable to the brighter of the two images. After three years we admitted defeat and wrote a paper about microlensing [Schechter et al. (2003)] which was, after all, OGLE's primary mission. The OGLE telescope was out of commission for upgrades during much of the following year; a few data points were obtained but were never reduced. But by then [Ofek & Maoz (2003)] had begun monitoring HE1104-1805 at the Wise Observatory. They used our data in conjunction with theirs to give a time delay that differed by a factor of two from the one used in figure 1. After Ofek and Maoz published their delay, [Wyrzykowski et al. (2003)] reduced the remaining OGLE data, confirming their value and giving the lightcurves in figure 6.
Figure 6. Lightcurves for the two components of HE1104-1805, taken with the OGLE 1.3-m telescope [Wyrzykowski et al. (2003)]. Note that while the measurement uncertainties are larger for the fainter, B image, the scatter is larger for the brighter, A image.
Notice that the error bars are smaller for the A image, yet the lightcurve is smoother for the B image. Image A appears to vary with an amplitude of 0.06 magnitudes on a timescale of one week. The interpretation of these fluctuations is somewhat speculative, but there is no question that microlensing represents an another difficulty for time delay measurements of H0.