And far more startling than the case of Somerset is that of Dorset, the 'Key to Domesday'. Mr Eyton here held that Dorchester, Bridport, and Wareham paid a full _firma unius noctis_ each, the total amount being reckoned by him at the astounding figure of £312 (p. 70)! Exeter, which affords a good comparison, paid only £18 (as render), though the king had 285 houses there: the three Dorset towns in which, says Mr Eyton, the Crown had 323 houses, paid in all, according to him, £312. The mere comparison of these figures is sufficient. But further, Mr Eyton observes (p. 93), that in 1156 'Fordington, Dorchester, and Bridport' were granted by Henry II to his uncle, 'as representing Royal Demesne to the annual value of £60'. This is an instructive commentary on his view that Dorchester and Bridport alone rendered £208 per annum. Our doubts being thus aroused, we turn to Domesday and find that it does not speak of any of these towns as paying that preposterous _firma_. The right formula for that would be 'reddit firmam unius noctis' (p. 84). Instead of that, we only have 'exceptis consuetudinibus quæ pertinent ad firmam unius noctis' (p. 70). The explanation is quite simple. Just as in Somerset, Mr Eyton admits, Langport and Ilchester, although boroughs, were 'interned' in groups of Royal demesne, paying the _firma unius noctis_, so in Dorset the boroughs were 'interned' in groups of Royal demesne. Indeed one of these groups was headed by Dorchester, and is styled by Mr Eyton the 'Dorchester group'. But he boldly assumed that 'Dorchester' must have two different meanings: [A] We assume about 100 acres to have belonged to the Domesday Burgh, and perhaps 882 acres to represent land, subinfeuded at Domesday, and annexed to Dorchester Hundred. [B] It follows that we assume about 429 acres, [to be that] ... which here figures [fo. 75] under the name Dorchester. It is not too much to say that any one, who refers to pp. 70-3, 78-101 of the _Key to Domesday_, will find that the singular misconception as to the Dorset Boroughs makes havoc of the whole calculation. But here again the point to be insisted on is not the mere mistake _per se_, but the elaborate assumptions based upon it and permeating the whole work.[204] Apart from the Manors grouped for a _firma unius noctis_, if we take the comital Manors (_mansiones de comitatu_) of Somerset, which were in the King's hands in 1086, we find their rentals given on quite a different principle to those of the Manors in private hands. (1) They are entered as renders ('reddit'), not as values ('valet'). (2) The sums rendered are 'de albo argento'. (3) In at least ten out of the fifteen cases, they are multiples of the strange unit £1 3s. As this fact seems to have escaped Mr Eyton's notice, I append a list of these Manors, showing the multiples of this unit that their renders represent: _£_ _s_ _d_