The word _castellum_, it is true, appears occasionally in Anglo-Saxon charters, but when it is used it clearly means a town. Thus Egbert of Kent says in 765: "Trado terram intra castelli moenia supranominati, id est Hrofescestri, unum viculum cum duobus jugeribus, etc.," where _castellum_ is evidently the city of Rochester.[57] Offa calls Wermund "episcopus castelli quod nominatur Hroffeceastre."[58] These instances can easily be multiplied. Mr W. H. Stevenson remarks that "in Old-English glosses, from the 8th century Corpus Glossary downwards, _castellum_ is glossed by _wic_, that is town."[59] In this sense no doubt we must interpret Asser's "castellum quod dicitur Werham."[60] Henry of Huntingdon probably meant a town when he says that Edward the Elder built at Hertford "castrum non immensum sed pulcherrimum." He generally translates the _burh_ of the _Chronicle_ by _burgus_, and he shows that he had a correct idea of Edward's work when he says that at Buckingham Edward "fecit _vallum_ ex utraque parte aquæ"--where _vallum_ is a translation of _burh_. The difference between a _burh_ and a castle is very clearly expressed by the _Chronicle_ in 1092, when it says concerning the restoration of Carlisle on its conquest by William Rufus, "He repaired the borough (burh) and ordered the castle to be built." The following is a table of the thirty boroughs built by Ethelfleda and Edward, arranged chronologically, which will show that we never find a _motte_, that is a moated mound, on the site of one of these boroughs unless a Norman castle-builder has been at work there subsequently. The weak point in Mr Clark's argument was that when he found a motte on a site which had once been Saxon, he did not stop to inquire what any subsequent builders might have done there, but at once assumed that the motte was Saxon. Of course, if we invariably found a motte at _every_ place where Edward or Ethelfleda are said to have built a _burh_, it would raise a strong presumption that mottes and burhs were the same thing. But out of the twenty-five burhs which can be identified, in only ten is there a motte on the same site; and in every case where a motte is found, except at Bakewell and Towcester, there is recorded proof of the existence of a Norman castle. In this list, the _burhs_ on both sides of the river at Hertford, Buckingham, and Nottingham are counted as two, because the very precise indications given in the _Anglo-Saxon Chronicle_ show that each _burh_ was a separate construction. _Burhs of Ethelfleda._