The survey nowhere employs the feudal language which became familiar in the twelfth century. Compare, for instance, the records in the first volume of Hearn's Liber Niger Scaccarii. In this last we find something about knights' fees in every page. In Domesday there is not a word—[20]

it is in no spirit of captious criticism, but from the necessity of demolishing the argument, that I liken it to basing conclusions on the fact that in the census returns we find something about population in every page, while in the returns of owners of land there is not a word. As the inquest of 1166 sought solely for information on knights and their fees, the returns to it naturally contain 'something about knights' fees in every page'; on the other hand, 'the payment or nonpayment of the geld is a matter which appears in every page of the survey' [of 1086] because 'the formal immediate cause of taking the survey was to secure its full and fair assessment'.[21] Nor is this all. When the writer asserts that 'in Domesday there is not a word' about knights' fees, he greatly overstates his case, as indeed is shown by the passages he proceeds to quote. I shall be able to prove, further on, that knights' fees existed in cases where Domesday does not mention them, but even the incidental notices found in the Great Survey are quite sufficient to disprove its alleged silence on the subject. As Mr Freeman has well observed:

Its most incidental notices are sometimes the most precious. We have seen that it is to an incidental, an almost accidental notice in the Survey that we owe our knowledge of the great fact of the general redemption of lands.[22]

Here then the writer does not hesitate to base on a single accidental notice the existence of an event quite as widespread and important as the introduction of knight service.[23]

I have now endeavoured to make plain one of the chief flaws in the view at present accepted, namely, that it is mainly grounded on the negative evidence of Domesday, which evidence will not bear the construction that has been placed upon it—and further that, even if it did, we should be landed in a fresh difficulty, the gulf between Domesday and Henry's charter being only to be bridged by the assumption that Ranulf Flambard 'devised' and introduced military tenure, with its results—an assumption, we have seen, which the facts of the case not only fail to support, but even discountenance wholly.

Let us pass to a second difficulty. When we ask the advocates of the view I am discussing what determined the number of knights due to the crown from a tenant-in-chief, we obtain, I venture to assert, no definite answer. At times we are told that it was the number of his hides; at times that it was the value of his estate. Gneist, who has discussed the matter in detail, and on several occasions, has held throughout, broadly speaking, the same view: he maintains that 'since Alfred's time the general rule had been observed that a fully equipped man should be furnished for every five hidæ, but it had never been established as a rule of law as in the Carlovingian legislation':[24] consequently, he urges, 'a fixed standard for the apportionment of the soldiery was wanting' at the time of the Conquest, and this want was a serious flaw in the Anglo-Saxon polity. William resolved to make the system uniform, and

the object that the royal administration now pursued for a century was to impose upon the whole mass of old and new possessors an equal obligation to do service for reward. The standard adopted in carrying out this system was approximately that of the five hides possession of the Anglo-Saxon period; yet with a stricter rating according to the value of the produce.[25]

The difficulty encountered in ascertaining this value was a main cause of the Domesday Survey being undertaken. This is Gneist's special point on which he invariably insists: 'Domesday book laid the basis of a roll of the crown vassals';[26] upon it, 'in later times, the fee-rolls were framed'.[27] By its evidence, 'according to the extent and the nature of the productive property, could be computed how many shields were to be furnished by each estate, according to the gradually fixed proportion of a £20 ground rent'.[28] For 'the feuda militum thus computed are no knights' fees of a limited area',[29] but 'units of possession', the unit being £20 in annual value.

Dr Stubbs, on the other hand, while rejecting the view that military service, since the days of Alfred, had been practically fixed at one warrior for every five hides,[30] leans nevertheless to the belief that the knight's fee was developed out of the five-hide unit, and that the military 'service' of a tenant-in-chief was determined by the number of such units which he possessed. But, as he also recognizes the £20 unit, there will be less danger of misrepresenting his views if I append verbatim the relevant passages:

The customary service of one fully armed man for each five hides was probably the rate at which the newly endowed follower of the king would be expected to discharge his duty ... and the number of knights to be furnished by a particular feudatory would be ascertained by inquiring the number of hides that he held.[31] The value of the knight's fee must already have been fixed—twenty pounds a year.[32]