6.6 Enclosing works

Excavation of rampart at Brown Caterthun fort, Angus, © RCAHMS


Palisades, walls, wall-and-fill ramparts, dump ramparts and ditches (each displaying a variety of forms and scales - e.g. with and without timber lacing) were all widely used in varying combinations or singly to enclose places in Scotland. They are constructional devices that appear to have no absolute chronological, regional, typological or functional significance; all have been identified in Late Bronze Age to Early Medieval contexts. They are not mutually exclusive; for example, palisades are recorded within dump ramparts (e.g. Harehope, Peebles. Feachem 1962), and in associations with walls in contemporary enclosure units (e.g. Brown Caterthun, Angus. Dunwell and Strachan 2007). Enclosing works are generally small in scale compared to those of other areas, such as the Wessex hillforts, but some of the largest form very imposing barriers. Variations in the character and scale of earthworks around their circuits are not uncommon. Natural topographic features (e.g. cliff edges, breaks of slope) were also commonly utilised as enclosing features, or to enhance built features.


The Hownam sequence of palisade, univallate, multivallate, first mooted by C.M.Piggott on the basis of observations during her excavations at the eponymous site in Roxburghs. is now regarded, at most, as applicable to individual sites within the eastern Cheviot valleys south and north of the Border. Survey and excavation data indicate its non-applicability to neighbouring regions such as the Lothians (Armit 1999) and south-west Scotland (RCAHMS 1997, 154-5; Banks 2000, 273-5). What seems to be becoming apparent is the sheer variety of individual site histories that do not reflect simple over-arching linear models of evolution. Excavations have demonstrated that some settlements alternated between of being enclosed and unenclosed (e.g. Braehead , Ellis 2007), or were enclosed only relatively late or briefly in a longer history of occupation. Evidence of multivallation is now to be generally regarded as evidence of multi-phase construction until proven otherwise (e.g. RCAHMS 1994).

However, some local trends in the sequence and manner of enclosing works have been suggested, however unclear, such as the extended use of timber for houses and enclosing works in Eastern Dumfriesshire  (RCAHMS 1997). Certain forms of enclosing work seem to have recurred at certain times and/or in certain places, and might be local chronological markers (such as Early Iron Age palisaded enclosures, and rectilinear enclosures in parts of southern Scotland) or even event horizons (as has been suggested for oblong forts, RCAHMS 2007, 103). How to progress in terms of understanding sequences and their chronology is a challenging question, but identifying local and regional trends seems feasible.

Occasionally encountered features

Chevaux-de-frises at Cademuir Hill, Scottish Borders © RCAHMS

In terms of occasionally encountered features, how rare they are, and why they are rare are important questions. Do survival biases account for the rarity of detection?

Breastworks have been identified in a small number of excavations e.g. Brown Caterthun (Dunwell and Strachan 2007), Balloch Hill (Peltenburg 1982), and The Dod (Smith 1982). A small number of the identified Early Iron Age twin or double palisades that define enclosures may be the remains of box ramparts (the Burnswark example being most frequently cited; Jobey 1978). Such structures are recorded more commonly in southern England, and may have appeared in Scotland as a result of an exchange of ideas (Armit 1997c, 58-9). However, it is unclear how to identify with confidence box ramparts as opposed to freestanding twin palisades, even with the benefit of excavation. The width of the spacing between the palisades has been proposed as a guide to identifying box ramparts, but with no consensus emerging consistently (RCAHMS 1997, 126 suggested 2-4m spacing was probably too wide, whereas Ralston 2006, 46-8 suggested a minimum 2m spacing was necessary). How can box ramparts be identified with any confidence? Evidence of vertical timbering within walls is uncommon, a rare example being within the entrance passage at Cullykhan.

Hedges are referred to by some commentators (e.g. Ralston 2006, 44) as a potential method of enclosure that will be difficult to detect in the archaeological record. Such suggestions sometimes arise in the context of explaining incomplete or ‘unfinished‘ enclosure circuits. It is possible that hedges were used a means of dividing and enclosing land on a much larger scale, in a way that leaves no coherent, archaeological trace (based upon comment provided by D W Harding in response to ScARF workshop). Is there any way of meaningfully addressing this issue? Might pollen or palaeobotany or molluscan analysis help?

The six chevaux-de-frises are recorded in Scotland (RCAHMS 1994, 74) do not appear functional as a barrier to mounted or foot assailants being limited in extent and, today at any rate, avoidable. They may have been more about display than defence (Harding 2004, 59). Other examples could, of course, have been removed, or buried in later defences, as at Castell Henllys, Pembrokeshire (Cunliffe 2005, 358). Evidence for timber examples suggested by Harbison (1971) and Cunliffe (2004, 358), and claimed continental examples, for example in Burgundy (I Ralston pers comm.), has yet to be found in Scotland.

Vitrified walls

The interpretation of vitrification has long been controversial, but there is now a wide consensus that vitrified walls result from the deliberate, pre-meditated destruction of timber-framed walls (Mackie 1976; Ralston 2006, 143; Cook 2010, 81-82). An understanding of the mechanics of the process is not complete but it now seems extremely unlikely to represent a constructional device and widespread vitrification, which is regularly found, is unlikely to be an outcome of accidental firing. Stratigraphic evidence has yet to be found that demonstrates the wall vitrification was a constructional device. Vitrification is evidenced in a wide range of enclosure walls of Early Iron Age to Early Medieval date (although the direct dating of vitrification by archaeomagnetism and thermoluminescence has provided controversial and contradictory results, Alexander 2002). A few forts have more than one enclosing work demonstrating vitrification or burnt walls (Clatchard Craig, Fife, Close Brooks 1986). In most cases evidence of vitrification of walls is partial, which chimes with the very mixed results achieved by those attempting to replicate vitrifaction by experimental methods where conditions, materials and time available may have been insufficient to produce the hoped-for results (e.g. Ralston 1986).

‘Unfinished‘ enclosing works

Current research is less certain than in the past about the frequency of ‘unfinished forts‘ (c.f. Feachem 1966, 68-70; 1971) - in the sense of an initial intended design being aborted and the location abandoned. There is now recognition of the possibility of palisade lines being present but invisible on the surface (e.g. Durn Hill, Aberdeen as suggested by RCAHMS 2007, 103), or apparently unfinished works as ritual tokens or deliberately segmented ramparts, or the emplacement of partial ramparts around main approaches for emphasis or display (e.g. Dunnideer, Aberdeen; Ralston 2007, 11). Many forts may have come down to us as fossilized work-in-progress, with aborted developments and modifications. For example, the outer segmented work at Brown Caterthun, Angus is probably a residue of an aborted redesign rather than an unfinished work per se. A review of the evidence with this in mind would prove valuable.


Entrance arrangements are believed to encode considerable amounts of information about the intentions of those who built, used and modified forts and enclosed settlements. There is considerable diversity. The majority of enclosed sites have one or two entrances, but some of the larger forts have multiple entrances (well beyond what might be sensible in defensive terms); yet others have no demonstrable gate (e.g. oblong forts).

Entrances generally seem to have been provided with relatively simple gate arrangements with little evidence of towers or overhead walkways (possible exceptions include Broxmouth fort western entrance at Period 3, Phase 3, Hill 1982, 156-7; and Shetland blockhouses, Ralston 2006, 84-5), although in some cases no evidence of an actual gate structure has been identified. ‘Guard chambers‘ and other traits of Atlantic roundhouse architecture are recorded in some Atlantic promontory forts and Shetland blockhouses (MacKie 1992; Carter et al. 1995). Examples are recorded of the aggrandizement of enclosing works around entrances, and the provision of additional outworks. From earthwork evidence alone the thickening of ramparts near entrances has occasionally been claimed as evidence for a gate structure. Some sites demonstrate considerable concern with the control of access along and between enclosure lines, but arguably to promote controlled (even hierarchical) access rather than to exclude. Specific questions could be addressed, e.g. are ‘drop slots‘ really evidence of portcullis-style gates (i.e. vertically opening by use of pulleys), or do they mark the foundations of sills against which gates would stop thus rendering more secure as well as impeding uncontrolled access by wheeled vehicles?

There has been no national study of the cosmological / phenomenological aspects of entrances to enclosed sites in Scotland, although there are a few accounts of fort entrances being aligned on prominent features in the surrounding landscape (e.g. Traprain Law, Kinpurney Hill). However, at face value there does not appear to be a pattern of predetermined entrance orientations for Scottish forts; a review of entrances from Peebleshire (Kokeza 2008) found most orientations followed the line of natural access routes and/or water sources. How commonly entrances were aligned on landscape features and to what end would be topics to pursue further.

Building enclosures

It is widely believed that enclosing works were locally inspired and constructed, using mobilised labour (however that was achieved - co-operation, reciprocity, social obligation, work feasts, coercion or slavery) under the direction of, or in partnership with, the occupants (in the case of settlements). Some of the larger circuits required enormous commitments of time and labour, for example to transport 2500m3 (over 4000 tons) of stone uphill to build the Brown Caterthun wall B, and even the construction of smaller enclosures required considerable resource input.

Acts of construction, maintenance and modification could have been variously episodic, seasonal, or depending upon circumstances, but assessing specific cases from archaeological evidence is not straightforward. For example at Brown Caterthun (Dunwell and Strachan 2007), there was evidence for considerable modification of ramparts, entrances, and access routes between enclosures, but given the vagaries of dating and the sample nature of investigation, it is difficult to be sure what is represented: frequent modification or a few discrete, change horizons? What can be inferred from activities described variously as ‘re-cutting‘ or ‘cleaning‘ enclosure ditches? ? To what extent can re-cutting/ cleaning episodes be disentangled before the very process becomes self-eliminative. The final Broxmouth report may be instructive in this regard.

Methods of enclosure have so far refused to show widespread sequences; identification of local and regional trends seems the best way forward.

What processes and ideas lie behind these widespread phenomena that occur sporadically but widely, such as vitrification or chevaux-de-fries?  How rare are they and why?  Do survival biases account for their rarity?

The logistics of and motives behind vitrification still offer scope for further consideration.

An audit of entrance excavation results would be beneficial.