A much more serious difficulty lies in the discrepancy between the three older calendars in which we have notes for this day and the almanacs of the later Empire, viz. that of Philocalus (A.D. 354) and the rustic calendars. The former tell us of a Mars-festival, with a horse-race; the latter know nothing of these, but note a festival of Mamurius, a name which, as we saw, occurred in the Saliare Carmen apparently as a variant of Mars, and came to be affixed to the legendary smith who made the eleven copies of the ancile. How are we to account for the change of Mars into Mamurius, and of feriae Marti into Mamuralia? And are we to suppose that the later calendars here indicate a late growth of legend, based on the name Mamurius as occurring in the Carmen Saliare, or that they have preserved the shadow of an earlier and popular side of the March rites, which the State-calendars left out of account? Apparently Mommsen holds the former opinion[107]. In his note on this day he says that it is easy to understand how the second Equirria came to be known to the _vulgus_ as Mamuralia (i.e. so distinguished from the first Equirria on Feb. 27), seeing that Mamurius who made the ancilia belongs wholly to the cult of Mars, and that this day was one of those on which the Salii and the ancilia were familiar sights in the streets of Rome. In other words, the Salian songs gave rise to the legend of Mamurius, and this in its turn gave a new name to the second Equirria or feriae Marti. And this I believe to be the most rational explanation of our difficulty, seeing that we have no mention of a feast of Mamurius earlier than the calendar of Philocalus in the fourth century A.D., which cannot be regarded as in any sense representing learning or research[108]. But of recent years much has been written in favour of the other view, that the late calendars have here preserved for us a trace of very ancient Roman belief and ritual[109]. This view rests almost entirely on a statement of a still later writer, Laurentius Lydus of Apamea, who wrote a work, _de Mensibus_, in the first half of the sixth century A.D., preserved in part in the form of two summaries or collections of extracts. Lydus was no doubt a man of learning, as is shown by his other work, _de Magistratibus_; but he does not give us his authority for particular statements, and his second- or third-hand knowledge must always be cautiously used.