Before leaving the subject of this levy, there is one point on which I would touch. When we find, as we often do, that the sum paid in 1168 in respect of a fief does not tally with the number of fees recorded in the _cartae_, we must remember that in the _Liber Niger_ and _Liber Rubeus_ we have not the original _cartae_, but only transcripts liable to clerical error. Checking the _cartae_ by these payments, we constantly find cases in which the number of fees should be slightly greater than is recorded in the _carta_.[175] I suspect that the transcriber, in these cases, has omitted entries in the original _carta_, and this suspicion is strongly confirmed by the fact that where the original return enables us to test the transcript, we find in the great _carta_ for the honour of Clare that the original transcriber has omitted half a fee of William de Hastinges, has left out altogether the entry 'Reginaldus de Cruce, _dimidium militem_', and has changed the quarter fee of Geoffrey fitz Piers into half a fee; while in that of the Bishop of Chichester, Robert de Denton's half fee is converted into a whole one. The later (Red Book) transcriber has made a further omission. Another source of discrepancy may be found in the dangerous resemblance of formulae. Thus the _carta_ of Ranulf fitz Walter records three and three-quarter fees duly accounted for. Yet his payment in 1168 is not £2 10s but £2 4s 5d. The explanation is that the holding was really three and one-third fees,[176] but the transcriber read 'iij[^{a.}] pars' (one-third) as 'iij. partes' (three-quarters). How easily such errors arose may be seen in the elaborate entries on Simon de Beauchamp's fief. Here the formula 'decem denarios quando Rex accipit marcam de milite', correctly reproduced in the Black Book, becomes 'x. denarius', etc., in the Red Book. The former expression means '_tenpence_ in the marc' (_i.e._ one-sixteenth of a fee); whereas the latter is equivalent to '_the tenth penny_ in the marc' (_i.e._ one-tenth of a fee), and upsets the whole reckoning. The correct formula is a not uncommon one and should be compared with the 'de xx. solidis viii. denarios' (eightpence in the pound) which is given as the holding of two knights of the honour of Clare, and represents the thirtieth of a fee.[177] Lastly, I think that, on further examination, there are three fiefs of which the _servitia debita_, though at first sight irregular,[178] may fairly be brought into line as multiples of the _constabularia_. That of Bohun, though implied by the _carta_ to be thirty and a half knights, paid in the fifth and eighth years on exactly thirty; that of Malet, though similarly given as twenty and one-sixth in the _carta_, is returned in the _Testa de Nevill_ as exactly twenty;[179] that of Beauchamp of Hacche, though distinctly given as seventeen in the _carta_, will be found, on careful collation of the rolls for 7 and 8 Hen. II, to be claimed by the exchequer as 17 + 3, _i.e._ 20.