432 MEDITERRANEAN ISLANDS. Chap. XI.

enemy, because most of them are inland. They could not be made to serve for the protection of the rich during insurrections or civil wars, nor to enable robbers to plunder in security the peaceful inhabitants of the plain. In short, unless the ancient Sardinians lived in a state of society of which we have no know- ledge elsewhere, these Nurhags were certainly not military works, When we turn to the second hypothesis and try to consider them as temples, we are met by very much the same difficulties as beset the fortification theory. If temples, they are unlike the temples of any other people. Generally it is assumed that they were fire temples, from their name Nur — in the Semitic languages signifying fire — but more from their construction. The little circular chambers in their interiors are admirably suited for pre- serving the sacred fire, and the external platforms as well adapted for that Sabean worship of the planets which is generally under- stood to be associated with fire-worship. But assuming this to be the case, why so numerous ? We can count on our fingers all the fire-temples that exist, or were ever known to exist, in fire- worshipping Persia ; and if a dozen satisfied her sjiiritual wants, what necessity was there for three thousand, or probably twice that number, in the small and sparsely inhabited island of Sar- dinia? Had every family, or little village community its own separate temple on the nearest high place ? and did each perform its own worship separately from the rest? So far as we know, there is no subordination among them, nothing corresj)onding to cathedrals, or parish churches or chapels. Some are smaller, or some form more extensive groups than others, but a singularly republican equality reigns throughout, very unlike the hierarchical feeling we find in most religions. In one other respect, too, they are unlike the temples of other nations. None of them are situated in towns or villages, or near the centres of population in the island. Must we then adopt the third hypothesis, that they v.ere tombs ? Here again tho same difficulties meet us. If they were tombs, they are unlike those of any other peojile with whom we are acquainted. Tlieir numbers in this instance is, however, no difficulty. It is in the nature of the case that sepulchres should accumulate, and their number is consequently one of the strongest arguments in favour of this destination. Nor does their situation militate