Another serious problem, already broached, concerns the frailties of narrow cuttings excavated through the walls, banks and ramparts that give definition and greatest distinction to hillforts — frailties felt particularly in terms of revealing and comprehending such complexities of structure and sequence as can sometimes be found embodied within them. By referring once more to Dinorben, the structural issue can be exemplified at a simple level by comparing separate episodes of work in the south-east quarter of the hillfort, where another of Gardner's narrow trenches (Section 10, opened in 1914) was encompassed by a much wider excavation in 1978. This is one of very few instances of extensive area-excavation of a hillfort's ramparts yet undertaken in Britain, the old trench being re-emptied before an adjoining stretch of bank was unravelled stratigraphically. Where the 1914 trench sliced through the multi-build inner bank and into the underlying deposits, it was shown in 1978 to have been less than 1m wide at the crucial level of the buried ground-surface, there passing, fortuitously and with uncanny accuracy, clean between the paired postholes of a timber-frame now known to have been a vital constituent of the earliest rampart (Figure 7; and compare Gardner and Savory 1964, 33, fig. 9, with Guilbert 1979a, 184-5). Measured along the line of the rampart, those postholes averaged little more than 1.5m apart centre-to-centre, which means that, while it was merely a matter of chance whether or not the 1914 trench intersected any of them, there was obviously never any prospect that it could disclose the repeated pattern of postholes that establishes the validity of the 1978 interpretation. It may seem reasonable to argue that, carefully excavated, a trench of, say, 2m width could not have failed to reveal at least one pair of those postholes, while a 3m trench would be unlucky not to encounter at least two pairs. However, this makes no allowance for evidence elsewhere that apparently significant components of rampart construction are known to have been much more widely spaced than the primary postholes at Dinorben. In general terms and for various reasons, it can be deduced that, even if appropriately excavated, a cutting of 10m width may well be inadequate in respect of certain earthworks (Guilbert 1975a, 116-17, including comparison of 4m and 15m cuttings at Moel y Gaer, Rhosesmor). Setting such hypothetical caveats aside, the clear-cut case at Dinorben, encapsulated in Figure 7, eloquently expresses one of the reasons — viz. structural form — not to trust results derived from rampart excavations of restricted width, irrespective of whether executed roughly by early excavators or with seemingly improved sensitivity by some archaeologists of more recent generations.
Incidentally, it is debatable whether Gardner, or indeed most of his contemporaries, would actually have appreciated the potential relevance of those postholes even had his cramped trench 10 happened to hit one or two of them, and even had the digging in the bottom of that trench been of a quality proper for the purpose.10 Anyhow, if recognised at all, why would Gardner not have regarded a couple of postholes underlying a rampart as indicative of some pre-rampart building, just as he did in a different trench (1920, 263)? In much the same vein, at the base of the deepest of his cuttings through Dinorben's 'huge main rampart' in the central sector of its southern earthworks (Section 1 of 1920, where that bank stood as much as 7m tall — see Dinorben, Figure L), Gardner may well have laid eyes upon one or more palisade-slots, unwittingly recording what may be seen retrospectively as our first evidence anywhere in Wales for a palisaded precursor to the earthworks of a fully-fledged hillfort (i.e. including 'the mysterious land drain': Gardner and Savory 1964, 29-32, fig. 7).11
It should not be imagined that the sort of shortcoming outlined above is confined to excavations of Gardner's era, for it was not until the 1970s that rampart cuttings of sensible width were opened on hillforts in Wales, those at The Breiddin (Musson et al. 1991) and at Moel y Gaer, Rhosesmor (Guilbert 1975a) preceding that of 1978 at Dinorben. These aimed to implement excavations with genuine open-area attributes, conducted manually, so as to dismantle an uninterrupted stretch of a well-preserved earthwork in such a way as to allow study of the make-up of each stratigraphical-cum-structural unit in the horizontal as well as the vertical plane. Only thus can some kinds of structural element be detected, particularly traces of timbering that decayed undisturbed and unburnt; and only thus can consistent, and hence crucial, structural patterns be distinguished from localised detail. This will apply even to a well-preserved rampart of single-phase construction, still more to a bank comprising successive ramparts piled one upon the other. The resulting revelations of those 1970s excavations were enough to persuade Hogg, who witnessed them on the ground and who was formerly a proponent of relatively small-scale sampling, that 'an old-style rampart section is probably a waste of labour' (1976, 16).12
It is regrettable that so few excavators have heeded that conclusion with positive action, presumably because the major commitment of resources (time, effort and money), which is its unavoidable counterpart, has deterred most from tackling ramparts on a sufficiently ambitious scale. The plain truth is that narrow cuttings allow ramparts to be examined purposefully in vertical section only, there seeming to testify to the sequence of construction, but unlikely to achieve even that if not excavated manually and with suitable care and technique, and then only if a trench is wide enough for intricacies of stratigraphy to be viewed to good effect. What is more, even this limited objective will remain unfulfilled unless luck dictates that every important element of each successive construction is epitomised proportionally on the exact line elected for cutting across the circuit of the earthwork. And what is worse, the record derived from a narrow cutting can deceive the unwary into believing that the simplistic impression and interpretations thus acquired, and which are all that can be expected of it, are reliable facts rather than potential factoids. In short, narrow cuttings through ramparts are to be discouraged.
Paradoxically, the Dinorben experience might be taken to suggest that we should be grateful for the small mercy that both Gardner in 1912-22 and Savory in 1956-69 were generally satisfied with narrow bank-cuttings, for the narrower were the trenches opened in any early campaigns of excavation, the better will be the longer-term prospects for credible investigation of the same bank, simply because cuttings of restricted extent do at least reduce the impact upon the earthwork, so lessening the degree to which its coherence will be compromised for later excavators.13 Nevertheless, it is inevitable that opportunities for area-excavating substantial unbroken stretches of bank, which, it is here argued, forms a prerequisite for reaching a useful understanding of any hillfort's structural development, will be reduced by earlier narrow cuttings, particularly where these were located at close intervals.
Internet Archaeology is an open access journal based in the Department of Archaeology, University of York. 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.
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.