The rays that we can see from some lunar craters are very spectacular and next to the lunar maria, one of the characteristic features of the lunar topography as seen from Earth. For a long time it has been recognized that these rays represent material ejected from the crater upon impact by the object that created the crater. Studies of volcanic ejecta on the Earth show that the ejecta often comes out as narrow, fan shaped plumes which lead to narrow 'fans' extending outwards from the caldera. This explains why rays exist at all, in terms of the tendency of impact and volcanic ejecta to form these narrow, vertical plumes.
If you carefully study the orientations of the rays, you will discover that they do not exactly point back to the center of the crater, however. In July 1959, Louis Giamboni of the Rand Corporation published an article in the Astrophysical Journal describing his detailed study of lunar rays, and proposed that their non-radial orientation is caused by the Moon having rotated on its axis during the time that the more distant ejects was in its ballistic trajectory over its hundred-mile journey. Material in low trajectories would fall sooner and have less deviation from radial. His model of the 68 rays in the Tycho crater indicated that the pattern of deviations could be fit only if the lunar 'day' were between 0.5 and 6.8 days long, not the 27 days that it currently is. According to Giamboni, this means that the rayed craters were created when the Moon was only about 100 million years old. This also explains why only a few craters have rays, because they are the oldest, and were formed when the crust was very very young.
Since the end of the Apollo program, however, the rays are widely interpreted as the 'splash marks' from very YOUNG craters, and the deviations seen are not considered statistically significant.