The theory just given has been productive of an abiding controversy. It has been sanctioned by several authorities, among others by Sir Norman Lockyer[555]. But, although the subject will bear deeper investigation, there is already sufficient evidence available to prove that the theory is faulty. There exist well-established cases of variation which cannot be forced under the rule. Again, some saints have several festivals allotted to them: St Nicholas has two, St Martin three, St John four, St Peter five, St Mary the Virgin eight[556]. Suppose, moreover, the case of an ancient church dedicated, we will say, to St Barnabas, and further assume that the present-day observer examines the orientation and finds it correct for the day of that saint, June 11th. The theory would still fail, since, by an Act of Parliament passed in the year 1752, eleven days were omitted from the calendar, and the orientation would, by the hypothesis, have been taken on May 30th. In considering ancient churches, therefore, the change in reckoning must be allowed for at the start. To meet the necessities of such an example as that proposed, it must also be granted that the advice of Durandus to build at the equinox was not generally followed. But we have seen that the facts are in accord with such an assumption, so that this minor objection falls to the ground. Once more; arguing from the fact that the village feast, which is usually held on the Saint’s Day, does not always coincide with the dedication festival, Mr Airy concluded that the theory under consideration was invalid. Without prejudicing the general verdict, we notice that this contention is impaired by evidence which proves that the dates of many village feasts have, at various times, been changed to suit local convenience (cf. p. 191 _supra_).