Dark matter in the Universe consisting of macroscopic objects such as primordial black holes may cause gravitational lensing of distant objects. The magnification associated with lensing will lead to additional scatter in the received flux from standard candles, and too small an observed scatter could rule out compact dark matter entirely. In this letter, we show how the scatter in fluxes of distant Type 1a supernovae could be used to distinguish between models with and without lensing by macroscopic dark matter. The proposed SNAP project, with $\sim 2400$ supernovae in the range $0.1\ls z\ls 1.7$, should be able to identify models at 99.9% confidence, if systematic errors are controlled. Note that this test is independent of any evolution of the mean supernova luminosity with redshift. The variances of the current Supernova Cosmology Project sample do not rule out compact lenses as dark matter: formally they favour such a population, but the significance is low, and removal of a single faint supernova from the sample reverses the conclusion.

We report the discovery of a new gravitational lens candidate Q2138-431AB, comprising two quasar images at a redshift of 1.641 separated by 4.5 arcsecs. The spectra of the two images are very similar, and the redshifts agree to better than 115 km.sec$^{-1}$. The two images have magnitudes $B_J = 19.8$ and $B_J = 21.0$, and in spite of a deep search and image subtraction procedure, no lensing galaxy has been found with $R < 23.8$. Modelling of the system configuration implies that the mass-to-light ratio of any lensing galaxy is likely to be around $1000 M_{\odot}/L_{\odot}$, with an absolute lower limit of $200 M_{\odot}/L_{\odot}$ for an Einstein-de Sitter universe. We conclude that the most likely explanation of the observations is gravitational lensing by a dark galaxy, although it is possible we are seeing a binary quasar.