When we come to place side by side the returns of 1166 and the payments made upon those returns in 1168, we find (at least, on the lay fiefs) the same distinction in both between 'the old feoffment' and 'the new'. But while the _returns_, as we saw, were made under three heads,[52] the _payments_ were made under two, namely, under the two feoffments. The reason of this difference can be established beyond dispute: the exchequer clerks had, in every instance, added the returns under the _third_ head to those under the _first_, and classed them together as 'old feoffment'. This is one of the points which, I think, have never been hitherto explained. Plenty of examples might be given, but these two will suffice. Walter de Aincurt returns 24 fees _de veteri_, 5 _de novo_, and 11 _super dominium_. The exchequer, in 1168, records him as paying on 35 fees _de veteri_, and on 5 _de novo_.[53] Richard de Haie returns 11 fees _de veteri_, 4 _de novo_, and 5 _super dominium_. The exchequer records him as paying on 16 _de veteri_, and 4 _de novo_. The main point, however, on which I propose to insist, is that these returns were intended to provide, and, as a matter of fact, did provide a new feudal assessment, wholly superseding the old one, in no case to the advantage of the tenant, but in many to the advantage of the crown. The _modus operandi_ was as follows. Instead of either adhering to the old assessment (_servitium debitum_), or uniformly substituting a new one based on the fees actually created, the crown selected in every case whichever of these two systems told in its own favour and against the tenant of the fief. If he had enfeoffed fewer knights than his _servitium debitum_ required, the crown retained that _servitium_ as the irreducible minimum of his assessment; but if he had created an excess of fees, the crown added that excess to his pre-existing assessment and increased the 'service' due from him _pro tanto_. This discovery is no conjecture, but is capable of arithmetical demonstration.