CHAPTER III

ANGLO-SAXON FORTIFICATIONS--_continued_

The bare list which we have given of the boroughs of Edward and Ethelfleda calls for some explanatory remarks. Let us take first the boroughs of Ethelfleda.

WORCESTER.--We have already noticed the charter of Ethelred and Ethelfleda which tells of the building of the burh at Worcester.[69] There appears to have been a small Roman settlement at Worcester, but there is no evidence that it was a fortified place.[70] This case lends some support to the conjecture of Dr Christison, that the Saxons gave the name of _chester_ to towns which they had themselves fortified.[71] The mediæval walls of Worcester were probably more extensive than Ethelfleda's borough, of which no trace remains.

CHESTER is spoken of by the _Anglo-Saxon Chronicle_ in 894 as "a waste _chester_ in Wirral." It had undoubtedly been a Roman city, and therefore the work of Ethelred and Ethelfleda here was solely one of restoration. Brompton, who wrote at the close of the 13th century "a poor compilation of little authority,"[72] was the first writer to state that the walls of Chester were enlarged by Ethelfleda so as to take in the castle, which he fancied to be Roman;[73] and this statement, being repeated by Leland, has acquired considerable vogue. It is very unlikely that any extension of the walls was made by the Mercian pair, seeing that the city was deserted at the time when it was occupied by the Danes, only fourteen years before. But it is quite certain that the Norman castle of Chester lay outside the city walls, as the manor of Gloverstone, which was not within the jurisdiction of the city, lay between the city and the castle.[74] A charter of Henry VII. shows that the civic boundary did not extend to the present south wall in his reign. Ethelfleda's borough probably followed the lines of the old Roman castrum.

BREMESBYRIG.--This place has not yet been identified. Bromborough on the Mersey has been suggested, and is not impossible, for the loss of the _s_ sometimes occurs in place-names; thus Melbury, in Wilts, was Melsburie in Domesday. Bremesbyrig was the first place restored after Chester, and as the estuary of the Dee had been secured by the repair of Chester, so an advance on Bromborough would have for its aim to secure the estuary of the Mersey. It was outside the Danish frontier of Watling Street, and could thus be fortified without breach of the peace in 911. There is a large moated work at Bromborough, enclosing an area of 10 acres, in the midst of which stands the courthouse of the manor of Bromborough. But this manor was given by the Earl of Chester to the monks of St Werburgh about 1152, and it is possible that the monks fortified it, as they did their manor of Irby in Wirral, against the incursions of the Welsh. One of the conditions of the Earl's grant was that the manor is to be maintained in a state of security and convenience for the holding of the courts appertaining to Chester Abbey.[75] Thus the fortification appears to be of manorial use, though this does not preclude the possibility of an earlier origin. On the other hand, if Bromborough is the same as Brunanburh, where Athelstan's great battle was fought (and there is much in favour of this), it cannot possibly have been Bremesbyrig in the days of Edward. Another site has been suggested by the Rev. C. S. Taylor, in a paper on _The Danes in Gloucestershire_, Bromsberrow in S. Gloucestershire, one of the last spurs of the Malvern Hills. Here the top of a small hill has been encircled with a ditch; but the ditch is so narrow that it does not suggest a defensive work, and it is remote from any Roman road or navigable river.

SCERGEAT has not yet been identified. Mr Kerslake argued with some probability that Shrewsbury is the place;[76] but the etymological considerations are adverse, and it is more likely that such an important place as Shrewsbury was fortified before Edward's time. Leland calls it Scorgate, and says it is "about Severn side."[77] It should probably be sought within the frontier of Watling Street, which Ethelfleda does not appear to have yet crossed in 911.

BRIDGENORTH is undoubtedly the Bricge of the _Anglo-Saxon Chronicle_, as Florence of Worcester identifies it with the Bridgenorth which Robert Belesme fortified against Henry I. in 1101.[78] Bridgenorth is on a natural fortification of steep rock, which would only require a stout wall to make it secure against all the military resources of the 10th century. We may therefore be quite certain that it was here Ethelfleda planted her borough, and not (as Mr Eyton unfortunately conjectured) on the mound outside the city, in the parish of Oldbury.[79] This mound was far more probably the site of the siege castle (no doubt of wood) which was erected by Henry I. when he besieged the city.[80]

TAMWORTH was an ancient city of the Mercian kings, and therefore may have been fortified before its walls were rebuilt by Ethelfleda.[81] The line of the ancient town-wall can still be traced in parts, though it is rapidly disappearing. Dugdale says the town ditch was 45 feet broad. Tamworth was a borough at the time of Domesday.

STAFFORD has a motte on which stood a Norman castle; but this is not mentioned in the table, because it stands a mile and a half from the town on the _southern_ side of the river Sowe, while we are expressly told by Florence that Ethelfleda's borough was on the _northern_ side, as the town is now. Stafford was a Domesday borough; some parts of the mediæval walls still remain. The walls are mentioned in Domesday Book.[82]

EDDISBURY, in Cheshire (Fig. 4), is the only case in which the work of Ethelfleda is preserved in a practically unaltered form, as no town or village has ever grown out of it. The _burh_ stands at the top of a hill, commanding the junction of two great Roman roads, the Watling Street from Chester to Manchester, and the branch which it sends forth to Kinderton on the east. As a very misleading plan of this work has been published in the _Journal of the British Archæological Association_ for 1906, the _burh_ has been specially surveyed for this book by Mr D. H. Montgomerie, who has also furnished the following description:--

"This plan is approximately oval, and is governed by the shape of the ground; the work lies at the end of a spur, running S.E. and terminating in abrupt slopes to the E. and S. The defences on the N. and W. consist of a ditch and a high outer bank, the proportions of these varying according to the slope of the hill. There are slight remains of a light inner rampart along the western half of this side. The remains of an original entrance (shown in Ormerod's _Cheshire_) are visible in the middle of the N.W. side, beyond which the ditch and outer bank have been partially levelled by the encroachments of the farm buildings. The defences of the S. side seem to have consisted of a long natural slope, crowned by a steeper scarp, cut back into the rock, and having traces of a bank along its crest. The S.E. end of the spur presents several interesting details, for it has been occupied in mediæval times by a small fortified enclosure, whose defences are apt to be confused with those of the older Saxon town. The rock makes a triangular projection at this end, containing the foundations of mediæval buildings,[83] and strengthened on the N.E. by a slight ditch some 7 to 10 feet below the crest; the rock on the inner side of this ditch has been cut back to a nearly vertical face, while on the outer bank are the footings of a masonry wall extending almost to the point of the spur. There are traces of another wall defending the crest on the N.E. and S.; but the base of the triangle, facing the old enclosure, does not appear to have been strengthened by a cross ditch or bank.

"It may be noted that this enclosure presents not the slightest appearance of a motte. It is at a lower level than the body of the hill, and belongs most certainly to the Edwardian period of the masonry buildings."

[Illustration: FIG. 4. Eddisbury, Cheshire. Witham, Essex.]

WARWICK Castle has a motte which has been confidently attributed to Ethelfleda, only because Dugdale copied the assertion of Thomas Rous, a very imaginative writer of the 15th century, that she was its builder. The borough which Ethelfleda fortified probably occupied a smaller area than the mediæval walls built in Edward I.'s reign; and it is probable that it did not include the site of the castle, as Domesday states that only four houses were destroyed when the castle was built.[84] The borough was doubtless erected to protect the Roman road from Bath to Lincoln, the Foss Way, which passes near it. Domesday Book, after mentioning that the king's barons have 112 houses in the borough, and the abbot of Coventry 36, goes on to say that these houses belong to the lands which the barons hold outside the city, and are rated there.[85] This is one of the passages from which the late Professor Maitland concluded that the boroughs planted by Ethelfleda and Edward were organised on a system of military defence, whereby the magnates in the country were bound to keep houses in the towns.[86]

CYRICBYRIG.--About this place we adopt the conjecture of Dugdale, who identified it with Monk's Kirby in Warwickshire, not far from the borders of Leicestershire, and therefore on the edge of Ethelfleda's dominions. It lies close to the Foss Way, and about three miles from Watling Street; like Eddisbury, it is near the junction of two Roman roads. There are remains of banks and ditches below the church. Dugdale says "there are certain apparent tokens that the Romans had some station here; for by digging the ground near the church, there have been discovered foundations of old walls and Roman bricks."[87] Possibly Ethelfleda restored a Roman castrum here. At any rate, it seems a much more likely site than Chirbury in Shropshire, which is commonly proposed, but which does not lie on any Roman road, and is not on Ethelfleda's line of advance; nor are there any earthworks there.

WEARDBYRIG has not been identified. Wednesbury was stated by Camden to be the place,[88] and but for the impossibility of the etymology, the situation would suit well enough. Weardbyrig must have been an important place, for it had a mint.[89] Warburton, on the Mersey, has been gravely suggested, but is impossible, as it takes its name from St Werburgh.

RUNCORN has not a vestige to show of Ethelfleda's borough; but local historians have preserved some rather vague accounts of a promontory fort which once existed at the point where the London and North-Western Railway bridge enters the river. A rocky headland formerly projected here into the Mersey, narrowing its course to 400 yards at high water; a ditch with a circular curve cut off this headland from the shore. This ditch, from 12 to 16 feet wide, with an inner bank 6 or 7 feet high, could still be traced in the early part of the 19th century. Eighteen feet of the headland were cut off when the Duke of Bridgewater made his canal in 1773, and the ditch was obliterated when the railway bridge was built. From the measurements which have been preserved, the area of this fort must have been very small, not exceeding 3 acres at the outside;[90] and it is unlikely that it represented Ethelfleda's borough, as the church, which was of pre-Conquest foundation, stood outside its bounds, and we should certainly have expected to find it within. As the Norman earls of Chester established a ferry at Runcorn in the 12th century, and as a castle at Runcorn is spoken of in a mediæval document,[91] it seems not impossible that there may have been a Norman castle on this site, as we constantly find such small fortifications placed to defend a ferry or ford. It is probable that Ethelfleda's borough was destroyed at an early period by the Northmen, for Runcorn was not a borough at Domesday, but was then a mere dependency of the Honour of Halton.

_The Burhs of Edward the Elder._

HERTFORD.--Two burhs were built by Edward at Hertford in 913, one on the north and the other on the south side of the river Lea. Therefore if a burh were the same thing as a motte, there ought to be two mottes at Hertford, one on each side of the river; whereas there is only one, and that forms part of the works of the Norman castle. Mr Clark, with his usual confidence, says that the northern mound has "long been laid low";[92] but there is not the slightest proof that it ever existed except in his imagination. Hertford was a borough at the time of Domesday. No earthworks remain.

WITHAM (Fig. 4).--There are some remains of a _burh_ here which are very remarkable, as they show an inner enclosure within the outer one. They have been carefully surveyed by Mr F. C. J. Spurrell, who has published a plan of them.[93] Each enclosure formed roughly a square with much-rounded corners. The ditch round the outer work was 30 feet wide; the inner work was not ditched. The area enclosed by the outer bank was 26-1/4 acres, an enclosure much too large for a castle; the area of the inner enclosure was 9-1/2 acres. As far as is at present known, Witham is the only instance we have of an Anglo-Saxon earthwork which has a double enclosure.[94] Witham is not mentioned as a borough in Domesday Book, but the fact that it had a mint in the days of Hardicanute shows that it maintained its borough rights for more than a hundred years. The name Chipping Hill points to a market within the borough.

BUCKINGHAM is another case where a _burh_ was built on both sides of the river, and as at Hertford, there was only one motte, site of the castle of the Norman Giffards, now almost obliterated. The river Ouse here makes a long narrow loop to the south-west, within which stands the town, and, without doubt, this would be the site of Edward's borough. No trace is left of the second borough on the other side of the river. Buckingham is one of the boroughs of Domesday.

BEDFORD has had a motte and a Norman castle on the north side of the Ouse; but this was not the site of Edward's borough, which the _Chronicle_ tells us was placed on the south side of that river. On the south side an ancient ditch, 10 or 12 feet broad, with some traces of an inner rampart, semicircular in plan, but with a square extension, is still visible, and fills with water at flood times.[95] This is very likely to be the ditch of Edward's borough. Both at Bedford and Buckingham the _Chronicle_ states that Edward spent four weeks in building the _burh_. Mediæval numbers must never be taken as precise; but the disproportion between four weeks and eight days, the space often given for the building of an early Norman castle, corresponds very well to the difference between the time needed to throw up the bank and stockade of a town, and that needed for the building of an earthen and wooden castle.

MALDON.--Only one angle of the earthen bank of Edward's borough remains now, but Gough states that it was an oblong camp enclosing about 22 acres.[96] It had rounded corners and a very wide ditch, with a bank on both scarp and counterscarp. Maldon was a borough at Domesday;[97] the king had a hall there, but there was never any castle, nor is there any trace of a motte.

TOWCESTER (Fig. 5).--There is a motte at Towcester, but no direct evidence has yet been found for the existence of a Norman castle there, though Leland says that he was told of "certen Ruines or Diches of a Castelle."[98] There was a mill and an oven to which the citizens owed soke,[99] and the value of the manor, which belonged to the king, had risen very greatly since the Conquest;[100] all facts which render the existence of a Norman castle extremely likely. But there can be no question as to the nature of Edward's work at Towcester, as the _Chronicle_ tells us expressly that "he wrought the burgh at Towcester with a stone wall."[101] Towcester lies on Watling Street, and is believed to have been the Roman station of Lactodorum. Baker gives a plan of the remains existing in his time, which may either be those of the Roman castrum or of Edward's borough.[102] The area is stated to be about 35 acres.

WIGINGAMERE.--This place is not yet identified, for the identification with Wigmore in Herefordshire, though accepted by many respectable writers, will not stand a moment's examination. Wigmore was entirely out of Edward's beat, and he had far too much on his hands in 918 to attempt a campaign in Herefordshire. As Wigingamere appears to have specially drawn upon itself the wrath of East Anglian and Essex Danes, it must have lain somewhere in their neighbourhood. The _mere_ which is included in the name would seem to point to that great inland water which anciently stretched southwards from the Wash into Cambridgeshire. The only approach to East Anglia from the south lay along a strip of open chalk land which lay between the great swamp and the dense forests which grew east of it.[103] Here ran the ancient road called the Icknield way. On a peninsula which now runs out into the great fens of the Cam and the Ouse there is still a village called Wicken, 6 miles west of the Roman road; and possibly, when the land surrounding this peninsula was under water, this bight may have been called Wigingamere. This suggestion of course is merely tentative, but what gives it some probability is that the Danish army which attacked "the borough at Wigingamere" came from East Anglia as well as Mercia.[104]

[Illustration: FIG. 5.--PLAN OF TOWCESTER ABOUT 1830.]

HUNTINGDON.--The borough of Huntingdon was probably first built by the Danes, as it was only repaired by Edward. In Leland's time there were still some remains of the walls "in places." Huntingdon is one of the _burgi_ of Domesday.

COLCHESTER.--This of course was a Roman site, and Edward needed only to restore the walls, as the _Chronicle_ indicates. Colchester was placed so as to defend the river Colne, just as Maldon defended the estuary of the Blackwater. As the repair of Colchester and the successful defence of Wigingamere were followed the same year by the submission of East Anglia, it seems not unlikely that Edward's various forces may have made a simultaneous advance, along the coast, and along the Roman road by the Fen country; but this of course is the merest conjecture, as the _Chronicle_ gives us no details of this very important event.

CLEDEMUTHAN.--This place is only mentioned in the Abingdon MS. of the _Chronicle_, but the year 921 is the date given for its building. This date should probably be transposed to 918, the year in which, according to Florence, Edward subjugated East Anglia. It is well known how confused the chronology of the various versions of the _Anglo-Saxon Chronicle_ is during the reign of Edward the Elder.[105] Cley, in Norfolk, would be etymologically deducible from Clede (the _d_ being frequently dropped, especially in Scandinavian districts), and the _muthan_ points to some river estuary. Cley is one of the few havens on the north coast of Norfolk, and its importance in former times was much greater than now, as is shown not only by the spaciousness of its Early English church, but by the fact that the port has jurisdiction for 30 miles along the coast.[106] It would be highly probable that Edward completed the subjugation of East Anglia by planting a borough at some important point. But as the real date of the fortification of Cledemuthan is uncertain, we must be content to leave this matter in abeyance.[107]

STAMFORD is another case where the _borough_ is clearly said to have been on the side which is opposite to the one where the Norman castle stands. Edward's borough was on the south side, the motte and other remains of the Norman castle are on the north of the Welland. It is remarkable that the part of Stamford on the south side of the Welland is still a distinct liberty; it is mentioned in Domesday as the sixth ward of the borough. The line of the earthworks can still be traced in parts. The borough on the north side of the Welland was probably first walled in by the Danes, as it was one of the Five Boroughs--Stamford, Leicester, Lincoln, Nottingham, and Derby--which appear to have formed an independent or semi-independent state in middle England.[108] Stamford is a borough in Domesday.

NOTTINGHAM.--The first mention of a fortress in connection with Nottingham seems to suggest that it owed its origin to the Danes. In 868 the Danish host which had taken possession of York in the previous year "went into Mercia to Nottingham, and there took up their winter quarters. And Burgræd king of Mercia and his Witan begged of Ethelred, king of the West Saxons, and of Alfred his brother, that they would help them, that they might fight against the army. And then they went with the West Saxon force into Mercia as far as Nottingham, and there encountered the army which was in the fortress (geweorc), and besieged them there; but there was no great battle fought, and the Mercians made peace with the army."[109] Nottingham became another of the Danish Five Boroughs. The Danish host on this occasion came from York, no doubt in ships down the Ouse and up the Trent. The site would exactly suit them, as it occupied a very strong position on St Mary's Hill, a height equal to that on which the castle stands, defended on the south front by precipitous cliffs, below which ran the river Leen, and only a very short distance from the junction of the Leen with the Trent, the great waterway of middle England.[110] Portions of the ancient ditch were uncovered in 1890, and its outline appears to have been roughly rectangular, like the Danish camp at Shoebury. The ditch was about 20 feet wide. The area enclosed was about 39 acres.

This borough was captured by Edward the Elder in 919, when after the death of his sister Ethelfleda he advanced into Danish Mercia, taking up the work which she had left unfinished.[111] The _Chronicle_ tells us that he repaired the borough (burh), and garrisoned it with both English and Danes. Two years later, he evidently felt the necessity of fortifying the Trent itself, for he built another borough on the south side of the river, and connected the two boroughs by a bridge, which must have included a causeway or a wooden stage across the marshes of the Leen. It is not surprising that the frequent floods of the Trent have carried away all trace of this second borough.[112] The important position of Nottingham was maintained in subsequent times, and it was still a borough at Domesday.

THELWALL.--According to Camden, Thelwall explains by its name the kind of work which was set up here, a wall composed of the trunks of trees. This was another attempt to defend the course of the Mersey, which was once tidal as far as Thelwall. No remains of any fortifications can now be seen at Thelwall, which was not one of the boroughs which took root. But the Mersey has changed its course very much at this point, even before the making of the Ship Canal effected a more complete alteration.[113]

MANCHESTER.--The _burh_ repaired by Edward the Elder was no doubt the Roman castrum, which was built on the triangle of land between the Irwell and the Medlock. Large portions of the walls were still remaining in Stukeley's time, about 1700, and some fragments have recently been unearthed by the Manchester Classical Association. It was one of the smaller kind of Roman stations, its area being only 5 acres. Manchester is not mentioned as a borough in Domesday, but the old Saxon town was long known as Aldportton, which literally means "the town of the old city." This is its title in mediæval deeds, and it is still preserved in _Alport_ Street, a street near the remains of the _castrum_.[114] The later borough of Manchester, which existed at least as early as the 13th century, appears to have grown up round the Norman castle, about a mile from the Roman castrum.[115]

BAKEWELL.--The vagueness of the indication in the _Chronicle_, "nigh to Bakewell," leaves us in some doubt where we are to look for this _burh_, which Florence calls an _urbs_. Just outside the village of Bakewell there are the remains of a motte and bailey castle (a small motte and bailey of 2 acres), which are always assumed to be the _burh_ of Edward. But the enclosure is far too small for a borough, and Edward's burh would certainly have enclosed the church; for though the present church contains no Saxon architecture, the ancient cross in the graveyard shows that it stands on a Saxon site. It is more reasonable to suppose that Edward's borough, if it was at Bakewell, has disappeared as completely as those of Runcorn, Buckingham, and Thelwall, and that the motte and bailey belong to one of the many Norman castles whose names never appear in history. There is no conclusive evidence for the existence of a Norman castle at Bakewell, but the names Castle Field, Warden Field, and Court Yard are at least suggestive.[116] Bakewell was the seat of jurisdiction for the High Peak Hundred in mediæval times.[117]

##