The main Swabian walls have fine D shaped towers defending the bulk of the castle from the main easy approaches from the east.  Four of the towers have a diameter of 30' and 2 at the north end 20'.  They were all subsequently cut down, probably under the joint monarchs (1479-1504) and equipped for artillery.  Externally the towers still have their typically thirteenth century roll moulded projecting string courses above their impressive batters.  The roll mouldings predate the artillery works as is shown by the way in which the gun ports cut through them.  Further, within the towers, the remnants of the earlier embrasures can be picked out, some converted into gun loops.

The 6 towers have all lost at least one and possibly 2 storeys from their original Swabian form.  At the north end of the wall is the peculiar main entrance.  This is a normal Sicilian hole in the wall gateway, but this one is set between 2 boldly projecting, but different sized round towers, with a third small tower set directly behind the entrance, forcing the approach passageway into a dogleg.  This unusual arrangement is only echoed in the late thirteenth century Denbigh castle gatehouse in Wales which appears a much more sophisticated version than Milazzo.  The small ward between the inner and outer gates of the gatehouse complex is further protected by a murder chute that empties into the ward from the south - an unusual and ingenious device.  Possibly this is late fifteenth century.  The walling of this line of defence is reasonably similar, consisting of rubble stonework with much Roman tile used as infilling, but not levelling courses.  No original apertures seem to have survived from the Swabian era.  Behind the south tower to the northwest is a thin walled chamber about 20' square which is now almost totally buried, although externally the wall is over 20' high.  This has the same makeup as the south tower, but the wall running away from it to the northwest, skirting under the castle proper as its outer ward, consists of better laid rubble and much more Roman tile.  The same is probably true at the north end of the wall, but the masonry is impossible to reach due to undergrowth and the ivy covering the walls.

Outer Ward
Southwest of the Swabian entrance in the north is a small, roughly 150' triangular enclosure with the main early castle to its south side.  At the apex of the triangle to the west the overly projecting square north tower of the inner ward forms a natural gateway with a small shed and associated wall, giving access to the large outer ward on the cliff top to the west of the keep.  This ward forms a polygonal enclosure roughly 400' east to west and 300' southwest to northeast.  The walls along the cliff top are angular and there are traces of rectangular buildings to the east, northwest and west.  To the southwest are the foundations of 2 round towers, one large, 25' diameter and one smaller, 20'.  They seem slightly too far apart to form a twin towered gatehouse and in any case there is no feasible way down the cliff from where they are, unless the sea came up to the cliff base and has subsequently silted up.  In this case it may have some similarities to the way to the sea which descends down the rock face at Harlech castle in Wales.

Inner Ward
Occupying the southern portion of the large enclosure of the castle crag is the main ward of the fortress.  The entire 6' thick enceinte, which is set on top of a great cistern, is claimed by the information panels to have been added by the Swabians to a Norman keep.  There is no evidence to support such a claim and much to refute it. 

The ward consists of an awkwardly shaped rhomboid with square corner towers and mural turrets, making 8 in all, but few being of uniform plan.  Indeed, if the towers were internal instead of boldly projecting, the fortress would appear more like a playing card shaped Roman fort than a castle.  That said, projecting towers at the angles etc were common in Roman forts from the third century AD onwards.  In the Western Empire these tended towards D shaped towers, viz. in Britannia, Burgh, Cardiff, Caerwent, Pevensey, Portchester and Scarborough to name a few, while Richborough had both D shaped and square towers.  In the Eastern Empire the towers tended more towards square or rectangular, viz. Constantinople and down the African litoral, Ain Tounga, Wadi Arabah, Dush and most importantly, Qasr Bashir.  The later is so important as it is a dated structure that survived little more than a hundred years in active use, but has firm dating and building evidence in the form of its gate inscription that still graces the top of the arch of its twin towered gatehouse.  At Bashir we have a fully fledged castle with spiral stairs, rectangular corner towers and a single ward.  The build is also totally what has been called ‘Byzantine'masonry construction style.  This word has been used for caution throughout this website and describes masonry built of a usually cut rubble which is interspaced with Roman tile.  In short, without firm dating evidence any of the castles listed under Aci as Byzantine, could well be Roman.  This would make much sense and explain the total lack of Roman forts on the island.  They were there all along, its just that no one thought to look at them as such.

The 4 corner towers of the inner enceinte were all about 20' square, while the intermediate ones on 3 sides were all slightly and irregularly smaller.  The fourth tower, mid way along the northwest wall, is the much larger keep.  The corner towers all appear to have had 3 first floor loops on each external face.  Of these several survive, being built of good quality tufa blocks.  As they have no sighting slits, or top and bottom oillets, they would all appear to be pre 1150.  All the towers and turrets seem to rest upon a 2 course sloping plinth of high quality ashlar tufa, but only the keep now rises above wallwalk level.  Apparently unlike the other towers, the west tower has 2 surviving floor levels.  This would seem to be due to it standing upon a rocky ridge which also houses the keep and the possibly Swabian hall block between them.  Consequently the rock base of the west tower is some 15' higher than that of the south tower, the same also appears to be true of the west tower in comparison to the north tower.

The four intermediate towers or turrets all appear of a slightly different form.  The southwest turret is the smallest of all, projecting from the curtain a mere 10' and only being about 15' across.  It is also without any external openings.  Entrance to the inner ward is currently gained to the northeast via a hole in the wall gate of the fifteenth century, set between the 2 towers at the west end of the northwest wall.  The original entrance also appears to have been set in this segment of the castle and occupied the east face of the central northeast tower.  Here is a blocked arch that was originally about 15' high.  It is Romanesque in appearance and was made of Roman tiles.  The entrance being dog legged is an unusual feature, with few other castles/forts demonstrating this design.  The nearest version in the West would seem to be the D shaped gate tower at Caldicot in Gwent, Wales, although this entrance is at about 60 degrees and a 90 degree right angle as at Milazzo.  The entrance gates at Aci, Erice and Monreale are also dog legged, but the first 2 are set within enceinte buildings and the later in an elongated rectangular building.  The southeast tower at Milazzo is also unusual in being well off centre, like the keep to the north.  Both are nearer the western end of the enceinte.  The southeast wall of the inner ward is also curved, unlike the other 3 major walls.  Centrally in this is a 5' wide 6" deep shaft cut into the outer face of the wall, possibly as a chute for a cess pit.  The central southeast tower seems to be a garderobe tower with blocked exit chutes to east and west, but no noticeable loops.

Keep
Commanding the small courtyard made by later buildings in the northwest corner of the ward, is the keep, now known as the Mast or Saracen Tower.  That it is the biggest tower, being 34' square with walls 6'6" thick, is why it is claimed to be a keep, but it lacks many of the facilities to make it a proper Norman keep.  There is no internal stair well or external stair turret, no space for kitchens or accomodation and the host of other necessary features to make a tower individually defensible, like those found at the much larger Paterno or Rometta keeps.  Indeed the Milazzo keep is more like the principal tower of the castle, similar to those at Aci or Caccamo rather than a western great keep like those found at Ivry la Bataille in France or Dover in England. 

The Milazzo keep, with the west tower and the hall block in between, is set on the highest point of the rock.  It is uncertain whether it pre or postdates the adjoining curtains.  It certainly fits into the playing card inner ward design, but also, with the west tower and block, occupies a higher level than the rest.  The design suggests it could date back to the earliest periods rather than being Arab, for which, despite all the claims, there is no evidence.  Certainly it is difficult to point to a similar definitely Arab structure as is explained blow. 

A later 30' high plinth has been added around the foot of the keep making the base 46' square.  This batter commences at first floor height.  Presumably the original 2 course sloping plinth is encased in this.  The tower, currently some 60' high, was modified well into the sixteenth century as well as lowered in height to take artillery.  Access to the keep is gained via a modern stairway that runs in front of its southeast face and leads to an inserted pointed ogival arch in the northeast wall of the hall block.  There is also a ragged modern cut through the southest face of the keep allowing access into the basement.  Presumably in medieval times access to this room was only via a trapdoor or wooden stairs from the floor above.  In the basement is a singular lighting loop set above the modern doorway and over an internal arch which appears reasonably modern.

The first floor of the keep is currently reached by passing through the ogival doorway of the hall block and climbing some stone stairs to a Romanesque door of 2 orders in the southewest face of the tower.  This door is mirrored by one to the east that led to the northeast wallwalk and eventually the north tower.  It would seem likely that its southwest counterpart once led to a similar wallwalk that led to the south tower, but that this wall was demolished when the hall block was built further to the northwest.  Alternatively this may be the original layout and the northwest front of the castle was always on the 2 different alignments.  The first floor room in the keep has a singular low window to the southeast with 2 seats in the Romanesque embrasure of 3 orders.  This appears totally rebuilt in modern time.  Externally the window appears to have 3 relieving arches, the upper one cutting through a tufa levelling course.  This would imply that the window has been inserted in medieval times.  To the northwest is a single splayed loop 6' up in the wall.  Obviously this was solely for lighting and, like the loops at Bashir, has a rectangular and not Romanesque embrasure.  From the southwest doorway another flight of steps leads up to the summit of the hall block.  From here it is possible to double back up a modern ramp to the northeast and gain the summit of the keep.  Presumably this gun parapet was once a chamber, before it was cut down to house roof top artillery.

A modern theory has grown up that the tower must be Arab as there is a church at Casalvecchio Siculo that was built in 1172 which has similar architecture and therefore must be contemporary to the keep.  Why 1172 and one being a church would make both Arab is a moot point, but the whole theory is illogical.  The argument is that both buildings have similar decorative herringbone masonry in them towards their summits.  However, the dating of the herringbone work at the abbey of SS. Peter and Paul d'Agro is impossible.  What the Greek inscription over the doorway actually says is that the church was repaired in 1172.  Obviously this implies absolutely nothing as to which part of the church was built when.  A quick glance at the facade also shows that the doorway and masonry above it has been rebuilt, no doubt in 1172.  The herringbone decoration in this section simply copies the work on either flank which is demonstrably much older.  Quite simply the dating of this church is unknown, but there is no reason why it might not be as old as the 560 AD claimed date for the foundation of the church, or indeed that it is older or younger than this.  The herringbone decorations at both sites are therefore undated, but both would appear to be Byzantine or older.  Unfortunately the coins found in the keep basement also add nothing to the argument concerning the date of the keep.  They date from between 1072 and 1138, which proves nothing of the tower's origins other than it was used after the later date.

Internal Buildings
Between the keep and the northwest tower lies the rectangular hall block which could be Swabian in origin (1197-1250).  This is of 2 storeys, although the wooden floor is now gone.  It contains many Romanesque apertures.  There is also a rebuilt thirteenth century fireplace on the ground floor.  The building leads through to the west tower.  This is entered through a modern doorway which leads into a chamber in which a Romanesque embrasure has been blocked up.  The floor above has what seems to be an original Romanesque embrasure and an inserted vaulted roof above it.  To the southeast in the hall block was a large and elaborate limestone window.  The rest of the interior of the castle has been heavily modified by the prison cells that once lined the enceinte, so comments about its medieval form seem unhelpful.

Copyright©2019 Paul Martin Remfry

• Index
• Home