Graeland de Thani of Essex owed seven and a half knights (the half of fifteen), and Roger de Berkeley probably the same. Those who owed a servitium of five knights were Robert fitz Harding, Baldwin Buelot, Simon de Cancy, Nigel de Lovetot (of the honour of Tickhill), Amfry de Cancy, Hugh de Dover (of the honour of Brunne),[92] Walter de Bolebec (Northumberland), Robert de Brus, Roger Bertram, and probably Stephen de Bulmer,[93] and Herbert 'de Castello'.
The cases in which the servitium can be shown not to have been a multiple of five are comparatively few. That of Simon de Beauchamp of Bedford was 54, of William Fossard 33½, of Humphrey de Bohun 30½, of William Malet 20⅙, of Robert de Beauchamp (of Somerset) 17, of William fitz John (of Harptree) 13¾, of William Blund 12, of Hugh Wac 10⅛, of William de Ros, William fitz John (of Weston) and William de Beauchamp (of Worcestershire) 7, of John de Bidun and Jocelin de Lovaine 5½.[94] But these, it will be seen, are quite insufficient to overthrow the accumulated array of evidence on the other side, and some of them are, doubtless, capable of explanation. The Bohun fief, for instance, in 1162 paid on exactly 30 fees.
It is impossible to resist the inference, from such evidence as we have, that the amount of the servitium debitum was a matter of custom and tradition, and could not usually be determined by reference to written grants or charters. On this point the returns of three Essex tenants are most instructive, while their similarity is so striking, that, as in the case of the Shropshire formulæ, it can scarcely be due to accident. The Earl of Essex closes with the words: 'et homines mei dicunt mihi quod debeo Domino Regi lx. milites'. Walter fitz Robert, who follows him, writes: 'et hoc mihi homines mei intelligere faciunt, quod debeo inde Regi servitium de l. militibus'. William de Montfichet ends thus: 'et hoc faciunt homines mei mihi intelligere—quod pater meus deserviebat per xl. milites'. With these expressions we may compare those of William fitz Alan's tenants, who assert that his Norfolk fief 'non debet domino Regi nisi i. militem ... ut antiqui testantur'; that his Shropshire fief 'non debet Regi nisi x. milites in exercitu ... sicut antiqui testantur'; and that, as to his Wiltshire fief, 'non sumus certi quod servitium debeat Regi de hoc tenemento'. The Abbot of Chertsey, also, states his servitium debitum with the proviso 'secundum quod scire possumus'. These expressions explain the uncertainty as to the servitium debitum in such cases as the See of Worcester and Ramsey Abbey.[95]
The same principle applies to the relation between the tenant-in-chief and his under-tenant. Thus the very first entry in the cartae runs as follows:
Willelmus de Wokindone iiij. milites et dimidium; et praeter hoc, ex testimonio curiae meae, dimidium exigo, quem ipse se non debere defendit.
Of another tenant on the same fief we read: 'praeter hoc, ex testimonio curiae meae, adhuc j. militem exigo'. Here, we see, appeal is made not to record evidence, but to oral testimony. So, too, the Bishop of Exeter adds this clause to his return:
Et praeter hos omnes, sicut a multis audivi, comes Gloucestriæ, et comes Hugo, et comes de Clare debent tenere de Exoniensi Episcopo; sed nullum ei servitium faciunt vel recognoscunt.
Surely in all such cases as these the obvious inference is that the tenant had been enfeoffed sine carta, or in the very words of the Provisions of the Barons (1259) 'feofatus sine carta a tempore conquestus vel alio antiquo feofamento' (§ 1).
And now for my theory. No one can have even glanced at the lists I have compiled without being instantly struck by the fact that the 'service' is reckoned in round numbers, and is almost invariably a multiple of 5, if not of 10.[96] This discovery, of course, is absolutely destructive of the view that it always represented the number of five-hide (or £20) units contained in the fief. Further, the number of differing fiefs assessed at precisely the same figure proves that the assessment was wholly arbitrary and cannot have been even the round sum which approximated most nearly the number of such units.[97] What then was the true determinant in the light of these conclusions? I reply—the unit of the feudal host.