Excavations undertaken at the end of last century showed that the site of Carew castle has been fortified since Roman times.  This enclosure, possibly about 130'x190', seems the regular size of a Roman fortlet, or a Welsh llys.  In the Norman era the site seems to have been occupied by Gerald Windsor (d.1126+), who held Pembroke castle against the Welsh during the great siege of 1098.  Gerald was also lord of Cilgerran.  Despite this, the history of the castle is blank until 1210 when King John confiscated the house of Carew (Domum suam de Carrio) as William Carew, the great grandson of Gerald Windsor, had gone to Ireland to help the Braose rebellion.  It was on 21 May 1212, that Faulkes Breaute, the royal baillif of Glamorgan and founder of Builth Wells castle, was ordered to return the house to William Carew.  This order gives little indication of what Carew castle may have looked like at the time, although the implication is that it was not heavily fortified.  William Carew seems to have died before 1213.  It is his grandson, Nicholas Carew (d.1297), who is credited, yet again without substantive evidence, with turning 'the house' into a castle.

The Carew family seem to have sold Carew castle shortly before the battle of Bosworth in 1485 and the new owner, Rhys ap Thomas (d.1525), a partisan of Henry VII (d.1509), began to convert the fortress into a palace.  In 1507 he held a great tournament there, 'the most magnificent entertainment in the history Wales'.  Rhys' grandson, Rhys
ap Gruffydd (d.1531), was one of the many executed by Henry VIII (d.1547).  The Crown then seized the castle before it was granted in 1558 to Sir John Perrot (d.1592), the son of Thomas Perrot (d.1531) and Mary Berkeley.  It was claimed by John's grandson, Robert Naunton, who had never met him, that he was the illegitimate son of Henry VIII (d.1547).  There appears no substance at all to the claim.  John built the great Elizabethan range on the north side of the castle.  The castle was much used and abused in the Civil War, but survived 3 sieges, still being inhabited by George Carew in 1686, though it seems to have soon fallen into decay.

The oldest part of the castle is the old rectangular gatetower built with small pieces of coursed rubble.  This is similar to the gatetowers found at Hay on Wye, Llanstephan and Manorbier.  Its basement is barrel vaulted and above that are 2 original storeys.  When the gate arch was blocked in the thirteenth century a loop was set centrally in the blocking wall.  To the south a latrine turret was added to service the upper rooms.  When the tower was raised in height an oriel window was added into the upper floor.  The tower still does not rise as high as the 2 surrounding curtains, while their junctions show that much rebuilding and alteration has gone on here.  As this was a gatetower, presumably there was once more to the castle, namely the 'house' that was in existence by 1210.  

When the 'Edwardian castle' was built in the thirteenth century the gatehouse was blocked and its foreward wall used as part of the east curtain of the new east range.  The new work used larger rubble blocks and had long crossbow loops.  Of this only the east front survives externally, for the north and west fronts have been encased by later work.  Possibly this range is all there ever was and this may have been part of 'the house' of 1210.  It consists of a block of domestic quarters running north from the old gatetower.  The north-east corner of this is at right angles and this may well have been the lakeside end of the castle, it terminating in a rectangular tower.  Whatever the case, this section has been much altered.  West of this is the great undercroft which runs down to end against the old gatetower.  South-east of the gatetower was a gate which was later converted into an internal gatetower.

South-east of the new entrance was a D shaped tower of a vaguely Welsh design which appears unaligned with the east front it should have defended.  This was three storeys high and was defended by numerous crossbow loops.  It has a fine plinth and a peculiar flat projection to the east.  The rear of this projection led out onto the outer ward wallwalk, though the rounded doorway has now been partially blocked and the outer wall reduced to half its original height.
 The junction of the tower battlements over this are mismatched and some modification has obviously gone on here.  This has led to the claim that the north backing wall is a later addition.  However, there is no indication of this in the external masonry lower down.  A low wall ran from here to the chapel tower covering this side of the east range.  There is a break in this wall opposite the new gateway.

East of the old gatetower is a boldly projecting polygonal chapel tower which abuts the north-east rectangular tower on its north side and is meshed into a chamfering off of the corner on the south side.  This contains a spiral stair.  The chapel tower has a pronounced plinth which is lacking on the rest of the front and appears to be a later addition.  Certainly its battlements do not align with those of the adjoining curtains and the curtain parapet was at least 6' higher than that of the projecting chapel tower.  This whole east range is most unusual and difficult to decipher.  At the same time as this was built, an outer ward may have been added of which only the north-east wall still survives.  Obviously this side always had been the entrance.

Unfortunately the next phase of the castle is not connected to the earlier remains.  This consists largely of the west side of the castle and two great round towers with massive Goodrich style spur buttresses.  These also appear at Newport castle some 25 miles north and at Cardigan keep some 30 miles away.  The north tower is positioned at a most awkward angle and there seems to be no logical reason for this.  Entrance to both towers was gained via a 3 storey great hall set between the two.  Apparently, as part of the same build, the hole in the wall gateway south of the old gatetower was converted into a proper internal gatehouse.  However, the backing to the old gatetower and undercroft appears to be more early sixteenth century.  Around the same time as the great hall was built in the thirteenth century, a narrow outer curtain was added to the south and east sides of the outer ward.

In the inner ward the curtain which joined the west tower to the south tower has lost its eastern half.  The great hall, with its normal plinth to the south, could well predate the west tower with its fine spur buttresses.  It certainly predates the plinthless section of south curtain which terminates against the heavily ruined supposedly sixteenth century turret.  It would seem that the outer gatehouse was added beyond the thirteenth century outer curtain probably by Rhys ap Thomas (d.1525).  This
structure partially fills the great rock cut ditch of the outer ward.

On the north side of the site, the gap between the two castle phases is joined by Sir John Perrot's impressive Elizabethan range.  It appears likely that the original castle thirteenth century wall still exists in part in the south-west portion of the inner wall of Perrot's work against the great hall.  If this is correct, all of Perrot's work is therefore outside the original castle.

This description and measly history of the castle of course leaves the perenial problem of when were these various phases of masonry built?  As ever there is no easy answer, but some comments can be made.  The gatetower as a design seems early and so could be eleventh or twelfth century.  Polygonal towers like the chapel tower are said to be late thirteenth century.  Justification for this is usually given by quoting the equally undated chapel tower at Kidwelly, or Denbigh's second phase castle.  None of these sites have secure dating, although Denbigh's polygonal towers are almost certainly post 1282.  What is forgotten in this is that polygonal towers at Dyserth castle in Clwyd date to the period 1241 to 1246 and there is no way of telling whether these were an early or a late style.  The same is true of the great spur buttresses.  Possibly the most securely dated similar feature are the towers at Goodrich castle which seem to date from the 1260s to the 1280s.  Marten's tower at Chepstow is somewhat later, but there is uncertainty over the dates of those at Cardigan and Newport.  In short, it might make more sense for the Carews to have built their 'Edwardian' castle before the overthrow of independent Wales in the 1280s, rather than afterwards.

There is also the problem of the south tower.  This has the appearance of a south facing mural tower, not a corner tower.  Perhaps originally this is what it was on the vulnerable southern front and that originally the outer ward was part of a main ward which was divided into two by the east range in the thirteenth century.  Whatever the case, Carew castle still has many mysteries to solve.

Why not join me at Carew and other Lost Welsh Castles next Spring?  Please see the information on tours at 
Scholarly Sojourns.


Copyright©2019 Paul Martin Remfry