1. Ffoulkes's attentions were directed towards the southern three of the hillforts atop the Clwydians, and an account of all six of them, with plans, appeared in Forde-Johnston 1965. Since then, fresh plans, more usefully contoured, have been published, as follows (stating vertical interval (vi) of the contours, in anticipation of discussion in sections 3.2 and 3.3): progressing northwards, Moel Fenlli (Gale 2011, 57 — 5m vi), Moel y Gaer, Llandedr (Gale 2011, 56 — 2m vi), Moel Arthur (Gale 2011, 55 – 2m vi), Pen y Cloddiau (Gale 2011, 54 — 5m vi), Moel y Gaer, Bodfari (Lock and Pouncett 2011, 144 — vi not stated, but presumed 2m), and Moel Hiraddug (Brassil et al. 1982, fig. 2 — c.3m vi).

2. The critical piece of 'white pottery' from Moel Fenlli is lost. Its status relative to the rampart has been debated by Simpson (1964, 211-12).

3. For a recent educated guess as to the role of Moel Fenlli in Late-Roman centuries, see Davies 2000, 89, tentatively inferring a 'wealthy establishment, perhaps the seat of a local lord', in line with certain other hillforts locally, notably Dinorben.

4. Besides those discussed here, Baring-Gould and his allies excavated at Clegyr Boia in Pembrokeshire.

5. In comparison, the overall plan of St David's Head included in Baring-Gould et al. 1899, having been prepared for publication by H.C. Mann, is an accomplished expression of the relationship between the principal archaeological features and the nature of the headland.

6. In their defence, it should be owned that Baring-Gould et al. may well have been on the right track in their broad deductions, given that the promontory-fort on St David's Head is now recognised to lie 'within the Iron Age tradition of the region', perhaps having 'origin in the Late Bronze Age' (Murphy 2001, 97-8, calling this fort 'Clawdd y Milwyr'), while Moel, or Foel, Trigarn, at least in its final form, is also now ascribed loosely to the Iron Age (e.g. Driver 2007, 134-6).

7. Hughes's plan of Braich y Dinas has been improved only by retrospective addition of 10ft form-lines (RCAHMW 1956, fig. 85).

8. Others who have concurred with Savory's retrospective (1971a) conclusions regarding the stratigraphical irrelevance of the iron axe include Hogg (1972), Davies (1977, 257) and Avery (1993, 138 and 143, note 69). As it happens, one outcome of the final, 1977-8, excavations at Dinorben offers some additional support, through decisive dating of the earliest recorded rampart to an early stage of the Iron Age (see Dinorben).

9. The word 'factoid' was coined by Norman Mailer in 1973, writing of the pernicious influence of supposed facts, or 'false truths', which some journalists chose to peddle about Marilyn Monroe but which had 'no existence before appearing in a magazine or newspaper' — he might have said much the same of some over-imaginative archaeological reporting. Closer to home, Oliver Rackham rightly railed against the influence of such misinformation in the related (in)discipline of landscape history, explaining that 'a factoid looks like a fact, is respected as a fact, and has all the properties of a fact except that it is not true', and recognising that, once accepted, they are hard to suppress, for such 'pseudo-history is not killed by publishing real history', and 'textbooks are rotten with factoids' (1994, 14, 205) — he might as well have been writing of archaeology.

10. To be fair, some excavators of those early days did recognise postholes when examining locations where their presence could reasonably be predicted, as in connection with gates at hillfort entrances, like those at Pen y Corddyn and Dinorben (Hughes 1906; Gardner 1912). Other particulars recorded by Gardner at Dinorben suggest that he might have been capable of noticing, if not necessarily understanding, possible timber-lacing, at least where charred (Gardner 1915, 203; Gardner and Savory 1964, 37, 43).

11. Some uncertainties inherent in interpreting Gardner's so-called 'land drain' as a palisade have been aired by Avery (1993, 140, note 7), also remarking a separate hint of a palisade, again recorded by Gardner, in the south-western sector (Section 19), as earlier noticed by Savory (1971a, 10), these being among possible clues in the old records that would be quite in keeping with the evidence for several lines of freestanding palisade encountered by later excavation in the south-eastern sector of Dinorben (Guilbert 1979a, 183-6; and see Dinorben).

12. Other than Hogg, few have joined the present writer in openly extolling the virtues of excavating ramparts on a scale and in a style appropriate to the intended goal of reliable recognition of both sequence and structure (Guilbert 1975a; since then, see Freeman 2000 (in effect, following Alcock 1972a, 66-8); Collis 2001, 145-8; 2010, 32; Harding 2012, 38).

13. Harding's recent discussion of excavating hillforts (2012, chapter 2) indicates a context in which the opening of a relatively narrow cutting across a rampart might now be considered acceptable, by locating it immediately adjacent to an older trench whose position can still be precisely identified and 'where the site's integrity has already been compromised'.

14. Nash-Williams was not alone in the bad habit of omitting the extent of excavation from plans, which is frustrating for those studying reports, and destined to be all the more so for any having intentions to investigate further portions of the same site, as the present writer can attest in respect of Dinorben.

15. Still less is it rational for the rudimentary plans of most hillfort entrances that were excavated early to acquire apparent credibility through direct comparison with complexities like those disentangled in one of the most recently and most extensively excavated of hillfort entrances, at Castell Henllys (Mytum 2013, figs 3.6, 12.1, 12.8, 12.11, 13.1, 14.1 and 15.2) — this cannot be regarded as a comparison of equals.

16. O'Neil's partial clearance of the entrance-passage at The Breiddin was actually more extensive and controlled than many of that and earlier eras, being sufficient to reveal a stone-packed trench crossing between large gate-postholes, each of which was apparently shown to have held two uprights, with a lesser central posthole for a stop, 'against which the two doors of the gate closed' (1937, 101-4, fig. 3). He regarded this as a single-period entrance, so demanding ad hoc explanations for sundry postholes seeming additional to basic requirements, but Musson could feel 'little confidence' in that view, recognising 'too many uncertainties … to allow detailed reinterpretation' (Musson et al. 1991, 9-10), while Avery (1993, 48-9) has thought it 'possible that some of the postholes … belonged to an earlier period of structure'. Despite the contrast given emphasis here, it cannot be claimed that the more expansive excavation of the Southern Entrance of the Inner Camp at Ffridd Faldwyn achieved unambiguous resolution of the sequence of construction represented among its palimpsest of postholes, as is evident from the diverging interpretations put forward by different commentators, with Avery (1993, 158-60, figs 41-45) presenting some seemingly sensible modifications to the three periods proposed tentatively by the excavator (O'Neil 1942, 32-9, figs 5-8), whereas Stanford (1971, 43; cf. Guilbert 1981b, 22; Cunliffe 1991, 337) plausibly envisaged a more substantial revision of O'Neil's conservative scheme, conceivably involving at least ten successive gateways. Note too that Avery's cautiously constructive approach has been applied to several other entrances selected for sweeping criticism in the main text here, viz. Llanmelin Camp (1993, 207), Pen Dinas (1993, 257-60), Tre'r Ceiri (1993, 346-7) and Pen y Corddyn (1993, 264-5).

17. Stripping of the North-West Gateway was the best thing to come out of Houlder's six weeks at Moel Hiraddug in 1954-5, with much of the remaining effort dissipated in a scatter of trenches, employing eight labourers in 'clearing debris' from narrow and partial cuttings through ramparts and in uninformative partial excavation of 'possible hut sites', all too fragmented to carry lasting value.

18. The term 'guard-chamber', or, as some prefer, 'guard-room', seems best retained, in view of both longevity of use (now more than a century — Hughes 1906, 268) and the need to avoid confusion in the literature (cf. Brassil and Guilbert 1982, 85, note 149). Whatever we may call them, and in spite of recent debate over terminology and purpose (Bowden 2006), such structures are still perhaps most readily seen as some form of porters' lodge (incidentally, the word 'lodge' has as great a pedigree as 'guardhouse' in this context — Prichard 1877, 227).

19. Avery's analysis of the published evidence for the South-East Entrance of Dinorben has produced variations on earlier interpretations, but, in maximising snippets, some of his arguments seem too rarefied for the record to support (1993, 136-7).

20. Audrey Williams had actually searched for evidence of buildings, but with less obvious success, inside other promontory forts in south Wales, where she excavated during 1938-9, so before Henllan. Starting at The Knave, a less-expansive approach left her frustrated at the paucity of recognisable postholes to substantiate structures, forcing her to fall back on banal allusions to huts with 'light framework' and even to guess that timber supports 'must have been thrust … into the joints of the rock' (1939, 215-16). Then at Bishopston, five of six recorded postholes suggested 'an oval timber structure … backing on to the rampart' (1940, 13-15, fig. 4; though it could easily have been part of a roundhouse antedating the rampart). And at High Pennard, having 'cleared' more than 100m² within the inner enclosure, she again struggled to make sense of the few apparent postholes, or 'slightly improved natural pockets in the rock', once more imagining some kind of lean-to, for which 'conjecture as to roofing structure would be unprofitable' (1941, 26, fig. 3).

21. In reviewing highlights of the Iron Age in Wales, Grimes (1949, 70-9) ignored the internal, rectilinear structures at Ffridd Faldwyn despite illustrating the adjacent excavated entrance, but he did recognise that 'Henllan has given our first picture of internal lay-out and hut-plans … in south Wales', while Bersu and Griffiths drew upon the Henllan ground-plans in analysing those subsequently excavated at Llwyn-du Bach (1949, 199-200). On the other hand, Hogg made no mention of the Henllan structures, but did refer, with laboured understatement, to Ffridd Faldwyn's 'interior set with a great number of large holes arranged in regular lines' (1965, 140). However, it was left to Stanford, some 30 years on from O'Neil's excavations at Ffridd Faldwyn, and in light of his own excavations in other Marches hillforts, to 'encourage a bolder acceptance of the lines of rectangular structures' there, meaning four-post and six-post units (initially in 1970, 108). Although some people were still not responsive to that interpretation (e.g. Spurgeon 1972, 341 — 'maze of large pits, the purpose of which is far from clear'; Savory 1976a, 263 — 'large aisled structures quite different from the small four-post structures … may provide the best explanation', with which Harding (2012, 104) seems tempted to agree), it is now widely accepted. Meantime, it is odd that this aspect of Caerau, Henllan, seems to have slipped from the reckoning for many recent commentators, though Savory claimed boldly to see 'at least four circular huts' there (Savory 1976a, 275 — his fig. 14 reproducing Williams's plan of the hillfort with outline of excavations curiously culled), while Williams and Mytum have made more cautious reference to 'roundhouses; one and possibly two examples of combinations of post-rings and wall gully' plus 'apparently isolated post-rings' (1998, 133).

22. Within a decade of digging at Castell Odo, Alcock embraced the merits of excavating hillforts on a big scale — witness his celebrated 1966-70 excavations at Cadbury Castle in Somerset (1972a), a formative experience for the present writer. Contrast Savory's final, 1965-9, foray to Dinorben (1971a), which came to be a cause of much vexation for the present writer.

23. Although it may appear unduly complex in detail, the broader distribution of trenches at Coygan Camp is explained in Wainwright 1967, 8-9.

24. The fragmented excavation of Coygan Camp is doubtless one factor contributing to a recent conclusion that the 'character of its … Romano-British occupation is far from clear', despite an 'exceptional' range of artefacts (Davies 2000, 74). Had the whole of Coygan's interior been excavated, there seems little doubt that the resulting fuller patterns of distribution of those artefacts would have added significantly to the overall archaeological value of the hillfort.

25. Reinterpretation of the peripheral structures at Walesland Rath, as first proposed by Williams (1988, 45-6, fig. 10) and, seemingly independently but more tentatively, by Mytum (1989, 68-9), has been widely adopted, including Cunliffe (1991, fig. 13.13), Davies and Lynch (2000, fig. 4.6) and Henderson (2007, fig. 6.22).

26. How long must it be before somebody excavates on a large scale within the outer enclosure(s) of such a multiple-enclosure work as that at Collfryn, for these are all too readily 'assumed to have been used as stock enclosures', a commonplace conclusion with the whiff of a factoid.

27. Besides Pen Dinas, Crawford's 'five weeks entirely occupied with survey work' in 1919 included similar contour-surveys of a small promontory fort, which he named 'Egryn Fort' (located at SH605206), and two stone circles, in total requiring 2389 spot-heights (63% of them for the Pen Dinas plan reproduced as Figure 19), recorded painstakingly with a 'spirit-level', which 'where the slopes are very steep … had to be moved frequently', and using that record to produce one-foot contours (1920, 101-2, with each plan presented on a separate, folded and unnumbered, sheet). A good impression of the monuments involved and of the quality of preservation of the multi-period landscape within which Pen Dinas sits, also of its challenging character for the sort of survey conducted by Crawford, can be gained from air-photographs taken in 1983-93, as reproduced in Crew and Musson 1996, 4-5, 7 and 15. Incidentally, there is potential for confusion through qualifying names used of the Pen Dinas, or Pen y Dinas, hillfort (SH606208), which is Tal y Bont in Crew and Musson 1996, Llanaber in Gresham 1967, and Llanaber or Hengwm in Crawford 1920.

28. For earlier generations, a simplified indication of overall slopes was obtained through hill-shading, closely related to hachuring; and it is relevant to view the discussion here against the backdrop of a debate as to the respective merits of shading and contouring, which had long been running among map-makers (whose ranks Crawford was on the verge of joining in 1919).

29. An early example is seen in Cobbold's intensive contouring of Earl's Hill Camp and Pontesford Hill Camp, not far across the border into Shropshire, recorded in 1905 and published in 1907, albeit his contours were mostly 'sketched' and doubtless simplified.

30. Hogg's total of 570-580 'hillforts' within modern Wales (1965, 116; 1975b, 11) was derived from the OS record, as published in 1962 (Ordnance Survey 1962). A comprehensive 'Atlas of Hillforts in Britain and Ireland', created by the Universities of Edinburgh and Oxford and now online, furnishes the considerably higher figure of 690.

31. For example, two hillfort plans of the same generation may be compared, with the 5ft contours seen in Figure 15 seeming well suited to depicting the simplicity of such a 'gently rounded summit' as Castell Odo (Alcock 1960, 103), whereas the 10ft spacing of those in RCAHMW 1956, fig. 72 seem inadequate, and their extent too restricted, for the irregularities underlying and surrounding the hillfort on rugged Conway Mountain. Similarly, the 5ft vertical interval adopted at Coygan Camp seems barely sufficient, especially as no attempt was made to combine these with hachuring of the artificial slopes of the earthworks (Wainwright 1967, fig. 2), which is true also of Bishopston, where anything less than the chosen 2ft interval would clearly have under-represented the earthworks (Williams 1940, fig. 2 — despite excavating on several forts where contour-mapping might have been appropriate, Williams seems to have employed it only here, perhaps because a suitably skilled labour-force was available).

32. In Driver 2013, only fig. 3.31, with its 1m contour-lines, is adequate in this respect, and how ironic that that plan portrays an unusually-sited plough-flattened enclosure. Consider how much the likes of his figs 5.17, 6.29 and 7.15 could have benefited from close contouring.

33. In fact, it would be hypocritical to do so, as I too have seen fit in certain circumstances, when speed and simplicity seemed essential, to provide only a summary diagram, lacking any contours — but this should not be taken to mean that I do not regret it.

34. A troubling explanation comes in an introductory passage of RCAHMW 1986, xxx, revealing that 'only the larger and more important sites have been contoured, as in most cases it has been felt that the amount of labour involved in undertaking this form of survey was not commensurate with the amount of archaeological information eventually portrayed' — troubling in so far as, from superficial inspection alone, who is to judge which sites are 'important' and which should be rejected summarily as unworthy of the best treatment? In reality, and hopefully, it may be that that dismissive attitude arose more from pragmatism (i.e. coping with inadequate resources) than from conviction. Suspicion that cost was an overriding consideration seems justified by the fact that the five hillforts reported to have been chosen for detailed plotting 'photogrammetrically from specially commissioned aerial photographs' are five of the six given 2m contours — surely no coincidence.

35. On the other hand, our field of study has made meagre use of stereo-plotting from vertical air-photographs (though employed by OS from the 1950s), providing a sobering illustration of how technological advances do not necessarily produce a wide benefit for archaeological practice, no matter how suited they may seem to its purposes (cf. note 34).


Internet Archaeology is an open access journal. Except where otherwise noted, content from this work may be used under the terms of the Creative Commons Attribution 3.0 (CC BY) Unported licence, which permits unrestricted use, distribution, and reproduction in any medium, provided that attribution to the author(s), the title of the work, the Internet Archaeology journal and the relevant URL/DOI are given.

Terms and Conditions | Legal Statements | Privacy Policy | Cookies Policy | Citing IA

Internet Archaeology content is preserved for the long term with the Archaeology Data Service. Help sustain and support open access publication by donating to our Open Access Archaeology Fund.