The theory of Williams, however, which is, I am inclined to think, the true one in the main, seems capable of being considerably modified and improved by the hypothesis of Sir George. The hill-fort,--palpably the most primitive form of fortalice or stronghold originated in a mountainous country,--seems to constitute man's first essay towards neutralizing, by the art of fortification, the advantages of superior force on the side of an assailing enemy. It was found, on the discovery of New Zealand, that the savage inhabitants had already learned to erect exactly such hill-forts amid the fastnesses of that country as those which were erected two thousand years earlier by the Scottish aborigines amid the fastnesses of our own. Nothing seems more probable, therefore, than that the forts of eminences such as Craig Phadrig and Knock Farril, originally mere inclosures of loose, uncemented stones, may belong to a period not less ancient than that of the first barbarous wars of Scotland, when, though tribe battled with tribe in fierce warfare, like the red men of the West with their brethren ere the European had landed on their shores, navigation was yet in so immature a state in Northern Europe as to secure to them an exemption from foreign invasion. In an after age, however, when the roving Vikings had become formidable, many of the eminences originally selected, from _their inaccessibility_, as sites for hill-forts, would come to be chosen, from _their prominence in the landscape_, as stations for beacon-fires. And of course the previously erected ramparts, higher always than the inclosed areas, would furnish on such hills the conspicuous points from which the fires could be best seen. Let us suppose, then, that the rampart-crested eminence of Knock Farril, seen on every side for many miles, has become in the age of northern invasion one of the beacon-posts of the district, and that large fires, abundantly supplied with fuel by the woods of a forest-covered country, and blown at times into intense heat by the strong winds so frequent in that upper stratum of air into which the summit penetrates, have been kindled some six or eight times on some prominent point of the rampart, raised, mayhap, many centuries before. At first the heat has failed to tell on the stubborn quartz and feldspar which forms the preponderating material of the gneisses, granites, quartz rocks, and coarse conglomerate sandstones on which it has been brought to operate; but each fire throws down into the interstices a considerable amount of the fixed salt of the wood, till at length the heap has become charged with a strong flux; and then one powerful fire more, fanned to a white heat by a keen, dry breeze, reduces the whole into a semi-fluid mass. The same effects have been produced on the materials of the rampart by the beacon-fires and the alkali, that were produced, according to Pliny, by the fires and the soda of the Phoenician merchants storm-bound on the sands of the river Belus. But the state of civilization in Scotland at the time is not such as to permit of the discovery being followed up by similar results. The semi-savage guardians of the beacon wonder at the _accident_, as they well may; but those happy accidents in which the higher order of discoveries originate occur in only the ages of cultivated minds; and so they do not acquire from it the art of manufacturing glass. It could not fail being perceived, however, by intellects at all human, that the consolidation which the fires of one week, or month, or year, as the case happened, had effected on one portion of the wall, might be produced by the fires of another week, or month, or year, on another portion of it; that, in short, a loose incoherent rampart, easy of demolition, might be converted, through the newly-discovered process, into a rampart as solid and indestructible as the rock on which it rested. And so, in course of time, simply by shifting the beacon-fires, and bringing them to bear in succession on every part of the wall, Knock Farril, with many a similar eminence in the country, comes to exhibit its completely vitrified fort where there had been but a loosely-piled hill-fort before. It in no degree militates against this compound theory,--borrowed in part from Williams and in part from Sir George,--that there are detached vitrified masses to be found on eminences evidently never occupied by hill-forts; or that there are hill-forts on other eminences only partially fused, or hill-forts on many of the less commanding sites that bear about them no marks of fire at all.