The Four Stages of Harlech Castle
extracted and augmented from: Harlech Castle and its True Origins (ISBN 1-899376-88-7)

For over a hundred years it has been asserted that Harlech castle was one of the wondrous new castles constructed for Edward I by his great engineer, Master James St George.  A close examination of the contemporary records would show that this assertion does not stand up to scrutiny.  Harlech rock had a long history before Edward I arrived there and there should be little doubt from the evidence that a castle stood there before this time.  It is the purpose of this paper to briefly review one or two aspects of these statements.  The evidence as a whole has been studied in the above book in far greater detail.

Early in the twentieth century it was decided that Harlech castle was founded by Otto Grandison (c.1238-1328), a captain of Edward I (1272-1307), who led 560 infantry led to the virgin site in April 1283.  This assertion has been fostered to the extent that it is now claimed that ‘recognition of the strategic importance of the site must be credited to Sir Otto Grandison, marching north from his capture of Castell y Bere'.  The assumption that the militarily unimportant Otto and his relatives of Savoy then dominated military developments in Wales has been held as a truism ever since, but as ever there is little evidence to back the grandiose claims.  The evidence of Otto's foundation of Harlech castle is based upon a single document held in London which states:

Payment to Lord Otto Grandison for the sustenance of 560 foot with him advancing from Castell y Bere to Harlech, £20 through tally.

All this actually tells us is that there was a place called Harlech during the Welsh war of 1282-83 and that it was important enough for Otto to go there with troops after the fall of Castell y Bere.  We know that on 22 April 1283, Earl William Valence of Pembroke, captain of the king's army of South Wales, and Roger Lestrange, captain of the king's army of the Marches, were besieging Castell y Bere and that the fortress surrendered under terms three days later.  It is therefore likely that Otto reached Harlech - some sixteen miles from Castell y Bere - about the end of April 1283, with perhaps Otto and his small force striking through the mountains while Roger Lestrange took the easier route to Cymer abbey and Abereithon where he looted Prince Llywelyn's goods.  Whatever the case, Otto did not stay long at Harlech, for by 10 May 1283 he was with Edward I at Llanrwst.

According to the twentieth century ‘historical' accounts about Harlech castle these entries refer to the beginnings of major fortifications at the site.  Yet this view simply does not stand up to a scrutiny of the evidence.  At all the rest of the virgin Edwardian castle sites a very different picture is drawn.  At Caernarfon the first thing done was the sending of brattishing to enclose the site.  The same is true of
Builth, Conway, Flint and Rhuddlan.  So why was this not thought necessary at Harlech?  Instead a few stoneworkers and sawyers were sent certainly after Otto had quit the site.  This hardly implies a major scheme of fortification.  Further, stonecutters and the quarrymen would appear to have been paid between 1s 3d and 2s 8d per day as shown from detailed receipts surviving at Harlech from 1286.  The men sent in 1283 seem to have received 6d per day, much more than the general 2d per day of a stonecutter, but way less than the shilling a mason might have expected.  This strongly suggests that they were either experienced stoneworkers or quite lowly masons or that there was one mason and nine stonecutters or some such breakdown of wages.

When we compare what was going on at Conway and Caernarfon at this time, the implication that a castle was being built at Harlech grows even weaker.  Just looking at ditchers and wheelbarrow men working in the ditch and quay at Caernarfon in the year 1283-84 we find at least several hundred men at work daily.  Where are these workers at Harlech?  Even worse, where are the payments for iron and steel, charcoal, cord, bellows, wheelbarrows, carthorses, and tools?  How could work progress without these?  Similarly the delivery of stone by sea to Caernarfon from diverse places from 15 January 1285 to 14 October 1285 cost £440.  Apparently land based deliveries of stone, brushwood and sand cost a further £140 5s 4½d.  If Harlech castle was really being built in 1283-85 where was the stone to build it brought from and why was it never accounted for or even paid for?  This is an especially important question remembering that Harlech castle ditch was not even begun until 1285.  It has been argued that the records of these deliveries to Harlech have simply been lost: ‘of their extent and cost we have no precise knowledge'.  If this were the case then where are the entries for the total expenditure in the pipe rolls?  They have certainly not been lost and great effort was expended to make sure that all costs were accounted for; as has been seen by the account of the deceased William Luda made up in 1291 many years after his death.  The idea that thousands of pounds of expenditure and payments to works have simply disappeared does not make sense, especially without even a mention of a roll of particulars that occurs again and again for other castle sites in this building period.

To get around this insuperable problem of the missing building records for Harlech, which could only have occurred in 1283 and 1284 as we have the records from 1285 to 1289, it has been suggested that the allegedly lost expenses were covered in the monies that had been issued by William Luda, for the wages of workmen at
Caernarfon, Chester, Conway, Criccieth, Harlech, Hope, Rhuddlan, West Wales and elsewhere, between March 1282 and November 1284.  The document quoted is a long one, an entire rotulus, versus and dorso, from the pipe roll of 1291 and copied in part into the chronicle of John Oxenedes.  Thankfully though, this is quite legible and close study of the original shows that this claim does not stand up.  Not because the translation is inaccurate, but simply because it does not appear likely that the workmen were used for castle building.  It is therefore worth printing the relevant parts of this document in full.  It is titled:

The account of Master William Luda, custodian of the king's wardrobe, of the receipts and takings in the expedition of the same king into the parts of Wales against Llywelyn ap Gruffydd, then prince of Wales and David his brother, from Sunday 22 March 1282 up to 20 November 1284 and for some part of 1284 by means of the counter roll of Thomas Gunneys produced through the hands of the said Walter Langton after his death.

The mere heading of this document shows the lengths the Exchequer went to insuring that their accounts were correct.  Deep in the main part of the document comes the sentence:

And in advances and wages to various stonecutters, carpenters, ditchers, woodcutters, charcoal burners, overseers and their masters for all the afore mentioned time at Chester, Hope, Rhuddlan, Conway, Caernarfon, Criccieth, Harlech, West Wales and diverse locations in the aforesaid war, £9,414 4s 11d, just as contained in the roll of particulars which the same Master William delivered to the treasury.

This sentence does not state that these workmen were employed on castle works.  Quite the opposite, it states, by the lack of the word castri attached to the place names, that these men were used for other purposes than castle building on the route from Chester around the coast of Wales and elsewhere during the war.  When looking at this sort of document it is readily apparent that the compilers were quite precise in what they meant and again and again you find reference to ‘work at xxx castle' or suchlike.  All this Harlech entry actually says is that this money was spent on the workers mentioned at diverse locations which included seven named places where we know that there happen to be castles.  However, where else would workmen have formed, but at royal camps?  The castles are clearly incidental to the placenames.  It therefore seems possible that such a force would more likely have been used for road clearance and repair, upon which we know from other records that the king was spending much time and effort.  Therefore the massive sum previously thought to have been divided amongst these fortresses, and other supposedly unnamed ones, consequently has to be removed from the pot of castle building funds.

Having delved briefly into original records it is necessary to look at what was actually happening at Harlech rather than at the Edwardian castles in general during the first months after the conquest of North Wales.  Between 17 and 20 August 1283, King Edward I actually stayed at Harlech.  It is inconceivable that since June ten stoneworkers and six sawyers could have constructed a castle at Harlech and it seems most unlikely that they would simply have built an undefended house for the king's stay in August.  Yet when the king left on 20 August he left five household squires of the king's household remaining in the garrison at Harlech (scutiferi de hospicio regis morantes in munciione apud Harlegh).  These men were John Cosyn of Grandison, Ebulo Montibus, John Gayton, John Scaccario and Peter Cornhill.  These five squires, with Scaccario dropping out soon after Christmas 1283, remained in charge of Harlech until the appointment of the first constable in October 1284.  Traditionally these men are said to have undertaken the ‘affairs' of the castle as well as building it with money sent them by the king.

This immediately begs the question of what these men ‘of the garrison at Harlech' were in charge of?  It has been alleged that from August 1283 until October 1284 they undertook the building of the first castle.  These buildings are first said to be encountered in the documentary evidence during 1289 when they are described as ‘old works' and by implication were in existence by 1285.  This scenario obviously needs examining closely.  The best way to do this, of course, is to print the original evidence and then discuss it.  The initial sentence of interest to us occurs between entries dated 2 October and 11 October 1283, which places our entry somewhere between those two dates. 

And Ralph Scot, squire of the king, and Gaylord Morlan in safely taking £100 from Acton Burnell to Harlech for the works there, for that charge for the expenses of their bodies for six days going and returning 4s 6d.
Monday, 11 October... Also Gaylord Morlan and Ralph Scot, king's squire, recently taking £100 from Kymor to Harlech for the works there for the expenses of their bodies for four days, 3s.  And William Marshall for his wages for one hackney in carrying £60 from Acton Burnell to Harlech, 2s.

It is quite surprising that this sending of £260 from the wardrobe to Harlech ‘for the works there' has been taken as evidence that the ‘old walls' (veteris muri) of the castle were constructed then, especially as there is no mention of castle building at Harlech, merely of work being carried out.  The question is what was that work and what would £260 have built, without a workforce other than the sixteen men and five squires that we have already seen?  The answer is next to nothing when it is considered that the real work at Harlech cost some £8,000 between 1286 and 1289.  In other words finishing Harlech cost more than thirty times the money that was sent to Harlech between 1283 and 1285.

The nature of the work that was occurring at Harlech starts to become clear from evidence contained in a roll of provisions.  This states that quantities of iron, steel and nails were sent from Chester by William Perton ‘for the works of the bridge of Bangor and the castles of Criccieth, Harlech and Caernarfon' during 1283 and 1284.  This consisted of 450 pecks (5.625 tons) of iron being delivered to Hugh Leominster for the castles of Criccieth and Harlech.  Another 400 pecks (5 tons) were to be delivered to the two castles to be made into sixteen sheaves of arrows, while 200 pecks (2½ tons) were to be converted into eight sheaves of steel arrows at Caernarfon.  A full twenty thousand nails of various sorts were sent to all four places, while the castles of Criccieth and Harlech received eight barrels full of quarrels for ‘the garrison'.  Presumably they were divided four to each.

This again is most interesting as Harlech and Criccieth were listed together as two functioning castle garrisons before the end of 1284 and by implication much earlier.  The document ends with lances being sent to several castles.  Again Criccieth and Harlech castles are mentioned together and forty lances are sent to them.  Caernarfon is not mentioned as a castle and only eleven lances are sent there.  The similarities between Criccieth and Harlech as Welsh castles existing before 1283 are again apparent and will remain so until Criccieth's destruction in the early fifteenth century.  There can be little doubt that Harlech was a fully functional castle with a garrison before the end of 1284 and by implication this fortress was standing when Otto Grandison arrived there with his troops at the end of April 1283.  Further this deduction is clearly backed up by chronicle evidence.

Not long after [the death of Prince Llywelyn], the king... subjugated almost all Wales; vills and lands, which were in the landlocked middle of Wales, distributing them to his own nobles, but retaining the maritime castles in his own hand; which brought forth great tranquilly in the following era.
The king of England, with the bridge now completed, crossed over with his army into Snowdonia, all the castles of which, without notable resistance, were captured and burned.  Truly the earl of Pembroke took Castell y Bere, which once belong to Prince Llywelyn; and soon afterwards all of Wales, with all its castles, voluntarily submitted to the king.

This is interesting that the castles of Wales were thought by an English monk to have submitted to Edward after the fall of Castell y Bere.  We know that Dolwyddelan fell on 18 January 1283 and that Dolbadarn was still in Dafydd's hands on the first of May.  It would therefore appear likely that this fortress and those of Caernarfon, Criccieth and Harlech fell soon after Castell y Bere on 25 April.  Thus chronicle and royal documentation are in harmony.

Further work was noted at King Edward's other castles which did not take place at Harlech.  Significantly, no brattishing is recorded as being erected at Harlech before operations got underway.  This is in direct contrast to the works at Conway, Caernarfon, Beaumaris, Aberystwyth and Buellt.  Similarly wooden enclosures or claustura were built at Rhuddlan and Flint in 1277 as well as at Conway and Caernarfon in 1283, before building work got underway.  At Harlech and Criccieth these are conspicuously absent.  In other words the evidence suggests that there was already a structure of some description at Harlech in August 1283 when Edward I arrived there and that this was already masonry.  Otherwise why send a handful of stonecutters and sawyers to the site?  Why too should Edward leave five of his squires there if there was only a platform by the seaside with a handful of tents and shacks on it?  Also, why would you send just £260 to the castle, enough for the materials, but not the labour, to build just one tower?   The documents do not state that this was for castle works, merely works.  It would therefore seem logical that some of the October money was sent to upgrade the castle after Edward's visit in August, as well as to pay the workers and the small garrison deployed there.  Certainly iron, steel and nails being sent to the castle works as well as military equipment for the garrison on the same scale as those sent to Criccieth castle indicates that Harlech and Criccieth were in a similar state of defence and repair in 1283-84.

Other than the suggestions made above, it is not possible to give any definitive answer as to the purpose of the £260 sent from the Shropshire border in October, or for the deliveries of iron at this time.  Considering all of this evidence the logical conclusion is that Harlech castle was standing in April 1283.  Therefore, as Edward I did not found Harlech castle, merely refortify it, we are left with the basis of the castle being earlier Welsh work.  The first ‘castle' was probably built in the mid-twelfth century on the site of an ancient llys.  These are the ‘old walls' that were found at the castle in 1285 and built upon.  The idea that such ‘old walls' were built in the previous two years is quite simply at odds with all the surviving evidence.

It is now difficult to decide what the first Harlech castle built by the Welsh princes might have looked like.  The best guess can only be based upon what we can see today and the hope that one day a proper excavation will be undertaken on this so misunderstood world heritage site.  The first castle would seem to have consisted of an irregular rectangular enclosure which now forms the basis of the inner ward.  This would have stood about twenty feet high with battlements and looked very much like Criccieth castle inner ward curtain now does.  The main buildings would probably have lain to the north and west as they do today and the entrance may have been around where the gatehouse is today or at the unflanked north entrance into the inner ward.  This castle was further defended on its northern side by the outer ward which currently survives from the north-eastern bastion to the water gate.  This fortress might have dated back to the time of Hywel ab Owain (d.1170) when a castle was mentioned in this district.  Indeed it could even date to as late as the time of Maredudd ap Llywelyn Fawr (bef.1225-1255), the last of the princes of Meirionydd.  It is also eminently possible that this was the llys described at Harlech in the tale of Branwen from the Mabinogion which was certainly recorded before 1250 and possibly as early as 1100 in the earliest forms we have.

Harlech castle, as originally built, seems to have lacked flanking and had relatively low walls.  This is probably because the fortress was mainly defended from the battlements by a spear equipped garrison, the main weapon of the soldiers of Meirionydd as recorded by Giraldus in the late twelfth century.  The same would be true of the refurbished Welsh castle which was abandoned to Otto Grandison in 1283.  This fortress, before 1283, had been substantially upgraded, quite possibly by Prince Llywelyn ap Gruffydd of Wales (bef.1215-82) and it is possible that he was influenced in this by what he and his men had seen at Caerphilly castle in Glamorgan.  The Harlech work consisted of a new great gatehouse which has been described as the most powerful gatehouse built in the British Isles before Beaumaris.  Beaumaris itself was simply a larger copy of Harlech and was supervised by Master James St George, who was of course the ex constable of Harlech.  Instead of Beaumaris being his military masterpiece, it comes across from this view of history as merely an enhanced copy of Harlech without the rock!  The Harlech gatehouse was flanked by two round towers which never seem to have progressed above the height of the original curtain wall.  At the same time as this was built, a middle ward was constructed to the north which shadowed the inner ward in a similar manner to the low ward at Caerphilly.  Of this work at Harlech only the north gatehouse and north-eastern bastion survive for sure.  It is also possible that these were added to an earlier middle ward.  The southern and eastern portions of the middle ward were probably swept away when the great ditch was dug in 1285.

The style of this early Harlech castle was typically Welsh.  There was no quick entrance for cavalry.  Anyone wanting to reach the hall before 1324 had to enter the east gate of the outer ward at the bottom of the ditch, rise up through the north gatehouse of the middle ward, pass around the main enceinte to the east and then through the great gatehouse.  All of this had to be done without the necessity of crossing that most essential of Norman imports into Great Britain - the drawbridge.  Pivoting bridges were only added to the path from the sea in 1286-89.  The eastern drawbridges were only added as late as 1324.  To put it quite simply, King Edward did not build any of his other fortresses without drawbridges, so why did he at Harlech?  The answer is obvious, he did not.  All King Edward I actually did was make use of the castle of the princes of Wales that he found there.

Similarly Rhuddlan and Flint castles, which were fortified for Edward I between 1278 and 1281, literally bristle with drawbridges and crossbow loops for an archery based defence.  The same is true of Conway and Caernarfon (1283-86), and also Beaumaris, although here St George copied Harlech and consequently forgot drawbridges for his great gatehouses.   Harlech castle, alleged in the twentieth century to have been built by Edward between 1283 and 1289, has only two possible archery loops and these were based within the great gatehouse gate passageway and in any case probably date back to the tower's inception before 1283.  The lower loops in the corner towers were not designed for archery, nor were the embrasures or windows higher up in the gatehouse.  In short, the early castle was designed for defence from the wallwalks - an idea absurd in the age of Edward I and his crossbows.  Such an idea, of course, suits the method of warfare of the Welsh of Meirionydd, but not that of the king of England.

In 1285 King Edward, after two visits, began the conversion of Harlech into an Edwardian fortress, but all he actually did in 1286-89 was build upon the foundations that Prince Llywelyn and his predecessors had left for him.  Even then he did not materially alter a great deal of the castle's design.  He raised the walls to make them more majestic and unscalable, but he did not lay on triple layers of crossbow loops as he did at Rhuddlan or a fortified town as he did at Conway and Caernarfon.  His main addition instead were two militarily irrelevant apartment towers with wonderful views over the coast and a host of garderobes.  This was hardly a battle castle of the first order, but a sea-side residence, made quite different in appearance from its old sister castle across Tremadog bay in Criccieth.  Perhaps the greatest shame is that Edward failed to make use of his seaside retreat after 1284.

Finally, in the reign of Edward II (1307-27), the castle was modified for the last time with the eastern entrance added and possibly the northern approaches blocked up.  Even with this work Harlech remained an infantry castle.  No cavalry component was kept there - cavalry the backbone of English armies.  There are no stables at Harlech and no way of getting horses on to the rock.  Harlech began its life as a Welsh infantry castle and never actually progressed beyond that point.  The £8,000 refortification of the site by Edward I was no more than a white elephant.  The unreconstructed Criccieth castle did Edward and his grandchildren just as well, at a fraction of the cost.  Castell y Bere was abandoned after 1295 when it was found to be so vulnerable to being isolated due to the lack of a navigable approach.  Criccieth was abandoned at some point between 1404 and 1410.  Harlech survived and lived on as the white elephant it had always been since the days of independent Meirionydd.  As such this isolated but extensive castle found itself supporting the lost causes of bygone ages, Dafydd ap Gruffydd in 1283, Owain Glyndwr in 1409, Henry VI in 1464, Richard III in 1485 and finally King Charles I in 1649.  Even in death Harlech was not worth demolishing and so was allowed to moulder on, almost alone in impenetrable Meirionydd, still maintained and upheld by its almost hereditary constables, the Wynns of Gwydir.

In conclusion we are left with the impression of Harlech castle, not as a single masterpiece of King Edward I and his overrated mason, Master James St George, but as a living fortress that was begun some time in the lost mysterious past of the Mabinogion, which evolved through the age of the princes, maturing into what could have been a great castle of Prince Llywelyn ap Gruffydd (d.1282).  However, that plan was cut short by the collapse of Gwynedd and Meirionydd in 1283 and Harlech became a major and largely irrelevant stronghold of King Edward I and his descendants, guarding a wild and impopulous country which not even a military king like Edward IV could be bothered to conquer due to its isolation.  This is the real Harlech castle.
A Timeline of Harlech Castle based on original documents
Date Lord/Constable Event
1147-70 Hywel ab Owain a castle in Meirionydd
Mar 1188 Gruffydd ap Cynan a new castle, Deudraeth, building in Eifionydd
bef. 1200 ‘Mabinogion' a llys at Harlech
1221 Llywelyn ab Iorwerth a castle built in either Meirionydd or Ardudwy
Apr 1283 Dafydd ap Gruffydd Harlech briefly occupied by Otto Grandison
Jun 1283 King Edward I 10 stonecutters and 6 sawyers sent to Harlech
17-20 Aug 1283 King Edward I Edward I stays at Harlech
20 Aug 1283 King Edward I English garrison of Harlech inaugurated under 5 squires
Oct 1283 King Edward I £260 to Harlech for works there
1283-84 King Edward I Hugh Leominster in charge of the castle works at Harlech
10-23? May 1284 King Edward I Edward I at Harlech
21 Oct 1284 Hugh Wlonkeslowe formal garrison of 30 men established
27-28 Oct 1284 Hugh Wlonkeslowe Edward I at Harlech
1285 Hugh Wlonkeslowe great ditch dug for £205
3 Oct 1285 John Bonvillars new constable
1286 John Bonvillars £1,500 spent on works with 1,826 tons of stone delivered on site
Jan 1286 John Bonvillars chimney added to gatehouse
Mar 1286 John Bonvillars wheelbarrows bought for works at Harlech
Mar 1286 John Bonvillars castle houses roofed with wattle and slates
Apr 1286 John Bonvillars scaffolding bought
Aug 1287 Agnes Bonvillars John Bonvillars drowned
1286-89 John/Agnes Bonvillars western towers built with garrets
1286-89 John/Agnes Bonvillars prison tower raised 23 feet to the height of the new curtain wall and then a further 14? feet
1286-89 John/Agnes Bonvillars outer ward repaired
1286-89 John/Agnes Bonvillars gatehouse stair turret finished
1286-89 John/Agnes Bonvillars topmost northern window of west wall of gatehouse repaired
1286-89 John/Agnes Bonvillars fireplace & window finished in top south room of gatehouse
1286-89 John/Agnes Bonvillars gatehouse and 2 (eastern?) towers roofed in lead
1286-89 John/Agnes Bonvillars west wall of inner ward doubled in thickness
1286-89 John/Agnes Bonvillars new hall, chapel, pantry and kitchen built
1286-89 John/Agnes Bonvillars 2 towers towards the sea roofed in lead
1286-89 John/Agnes Bonvillars kitchen, hall, chapel and pantry roofed in lead
1286-89 John/Agnes Bonvillars wall from the great ditch to the middle gate built?
1286-89 John/Agnes Bonvillars way to the sea with wall built
1286-89 John/Agnes Bonvillars south garderobe turret built
late 1295 Robert Staundon a wall built from the new tower to the sea
1306 Edward II new bakery built (on site of kitchen?)
1306 Edward II chapel and bridge beyond the castle gate repaired
1308 Edward II new buttery and pantry built within the great hall
1321 Edward II storeroom and cellar mentioned
1323-4 Edward II east bridge, drawbridges and gatehouse built

Essay extracted, without footnotes and transcipts of original documents, and augmented from Harlech Castle and its True Origins (ISBN 1-899376-88-7)

Available for £39.95.

Why not join me at Harlech and other Great British Castles this October?  Please see the information on tours at Scholarly Sojourns.


Copyright©2016 Paul Martin Remfry