In the case of Somerset one need not dwell on his giving its amount for the Williton group as £105 16s 6-1/2d, when the sum named is £105 17s 4-1/2d, although absolute accuracy is, in these matters, essential. We will pass at once to the bottom of the page (ii. 2), and collate his rendering of Domesday with the original: 'T.R.E. reddebat dimidiam 'Reddebat T.R.E. dimidiam firmam noctis et quadrantem' noctis firmam et unum quadrantem' (Domesday). (Eyton). Domesday gives the payment (in a characteristic phrase), as _three-quarters_ [a half and a quarter] of a _firma noctis_. Mr Eyton first interpolates a 'unum', and then overlooks the 'quadrantem', with the result that he represents the due T.R.E. as a _firma dimidiæ noctis_ (i. 77). So far, this is only a matter of error _per se_. But Domesday records the commutation of the due T.R.W. at £79 10s 7d. This proves to be _three-quarters_ of the commutation, in two other cases, for a whole _firma noctis_ (£106 0s 10d). Mr Eyton, however, imagining the due to have been only _half a firma_ set himself to account for its commutation at so high a figure (i. 77-8). This he found no difficulty in doing. He explained that 'this was not a mere commutation', but 'was doubtless a change which took into consideration the extra means and enhanced value of Meleborne'. The probability is, then, that what we have called the _enhanced ferm_, was enhanced by something less than the gross profits we have instanced; that is, that a part of those profits, say the Burgage rents, or some of them, had contributed to the _dimidia firma noctis_ before the commutation. All these ready assumptions, we must remember, are introduced to account for a discrepancy which does not exist. Great masses of Mr Eyton's work consist of similar guesses and assumptions. Now, if these were kept scrupulously apart from the facts, they would not much matter; but they are so inextricably confused with the real facts of Domesday that, virtually, one can never be sure if one is dealing with facts or fancies.