1: \begin{abstract}

2: In this paper we show that to explain the observed distribution of

3: amplitudes in a large sample of quasar lightcurves, a significant

4: contribution from microlensing is required.  This implies the existence of

5: a cosmologically distributed population of stellar mass compact bodies

6: making up a large fraction of the dark matter.  Our analysis is based on

7: the lightcurves of a sample of over 1000 quasars, photometrically

8: monitored over a period of 26 years.  The intrinsic variations in quasar

9: luminosity are derived from luminous quasars where the quasar accretion

10: disc is too large to be microlensed by stellar mass bodies, and then

11: synthetic lightcurves for the whole sample are constructed with the same

12: statistical properties. We then run microlensing simulations for each

13: quasar with convergence in compact bodies appropriate to the quasar

14: redshift assuming a $\Lambda$CDM cosmology.  The synthetic lightcurve is

15: then superimposed on the amplification pattern to incorporate the effects

16: of microlensing.  The distribution of the resulting amplitudes can then be

17: compared with observation, giving a very close match.  This procedure does

18: not involve optimising parameters or fitting to the data, as all inputs

19: such as lens mass and quasar disc size come from independent observations

20: in the literature.  The overall conclusion of the paper is that to account

21: for the distribution of quasar lightcurve amplitudes it is necessary to

22: include the microlensing effects of a cosmologically distributed

23: population of stellar mass compact bodies, most plausibly identified as

24: stellar mass primordial black holes.

25: 

26: \end{abstract}

27: