The close resemblance of the Welsh system of clustering the homesteads and trevs in groups of four and twelve or sixteen, to that prevalent in Ireland, points to the common origin of both. It confirms the inference that both in Wales and in Ireland this curious practice found its _raison d'être_ in a stage of tribal life when the families of free tribesmen did not as yet always occupy the same tyddyn, but were shifted from one to another whenever the dying out of a family rendered needful a redistribution to ensure the fair and equal division of the tribal lands among the tribesmen, 'according to their antiquity' and their rank under the tribal rules. «Redivisions and shifting of holdings.» This occasional shifting of tribal occupation within the tribe-land was still going on in Ireland under the eyes of Sir John Davies, and it seems to have survived the Roman rule in Wales, though it was there probably confined within very narrow limits. It seems, however, to have been itself a survival of the originally more or less nomad habits of pastoral tribes. «Semi-nomadic habits stopped by the Roman rule.» So, also, the frailty of the slightly constructed homesteads of the Welsh of the thirteenth century, which seemed to _Giraldus Cambrensis_ as built only to last for a year, may be a survival of a state of tribal life when the tribes were nomadic, and driven to move from place to place by the pressure of warlike neighbours, or the necessity of seeking new pastures for their flocks and herds. But the nomadic stage of Welsh tribal life had probably come to an end during the period of Roman rule. * * * * * «The grades in tribal society.» Putting together the Irish and Welsh evidence in [p237] a variety of smaller points, a clearer conception may perhaps be gained than before of the character and relations to each other of the three or four orders into which tribal life seems to have separated people--the _chiefs_, the _tribesmen_, the _taeogs_, and under all these, and classed among chattels, the _slaves_. The _chief_ evidently corresponds less with the later lord of a manor than with the modern king. He is the head and chosen chief of the tribesmen. His office is not hereditary. His successor, his _tanist_ or _edling_, is chosen in his lifetime, and is not necessarily his son.[309] The chieftains of Ireland are spoken of in mediæval records and laws as _reguli_--little kings. When Wales (or such part of it as had not been before conquered and made manorial) was conquered by Edward I. the chieftainship did not fall into the hands of manorial lords, but was vested directly in the Prince of Wales.[310] «The tribesmen.»