MYTHMAKERS OF MAIDEN CASTLE: BREAKING THE SIEGE MENTALITY OF AN IRON AGE HILLFORT

The idea that an invading Roman army brought about the end of hillforts in south-west Britain, using artillery fire to demoralize and defeat their enemy, is one of the most powerful narratives in British archaeology, being a constant element in both academic literature and public discourse. At the heart of the debate is the evidence recovered by Mortimer Wheeler during fieldwork conducted at Maiden Castle, in Dorset, between 1936–37. Wheeler interpreted a series of burials found in the east gate of the hillfort as a ‘war cemetery’, residue of an ultimately futile defence of the site, in the face of Roman aggression, by the local Durotriges tribe. A recent survey of hillforts in Dorset has, however, cast significant doubt on Wheeler’s hypothesis, suggesting that not only is the widely accepted battle-theory unsupported by the archaeological evidence, but also that the Durotriges themselves were unconnected to any fortification or defence of Maiden Castle. This paper explores the conclusions of that survey, examining how the dramatic story of a siege first took shape in the late 1930s and why it became so immediately popular with the public. The problems of linking material remains to postulated historical events are outlined and the beginnings of a new model for Late Iron Age settlement at Maiden Castle are presented for the first time.


INTRODUCTION
Belief that the Iron Age hillforts of south-west Britain formed the focus of native resistance to the invading Roman army in AD 43-44 is not only widespread but strongly held. Since the first full archaeological report on Maiden Castle appeared (Wheeler 1943), countless books, papers and television documentaries have treated a speculative Roman assault upon the hillfort as definitive fact. The account of a furious but futile defence of property, family and land by the local Iron Age tribe of the Durotriges, leading eventually to their slaughter or enslavement, is undeniably powerful and remains one of the more potent stories relating to the demise of British prehistory. A recent study of Dorset hillforts has, however, called this narrative into question (Stewart and Russell 2017, 90-101, 106-13, 155-70), reassessing the evidence provided for the so-called Great Battle of Maiden Castle. The discrediting of the siege scenario has serious repercussions, not only as to how we interpret evidence revealed through excavation, especially in relation to perceived historical events, but also how archaeological discoveries are reported in the press. battle, so persuasively described by Mortimer Wheeler (1943, 61-4), may actually be nothing of the sort (Sharples 1991a, 91-101;1991b, 118-25;Stewart and Russell 2017, 158-62). In 1991, following a new phase of excavation, Niall Sharples noted that the bodies found by Wheeler in the east gate could plausibly have been 'brought from elsewhere to this cemetery and do not indicate death in the immediate vicinity of the graves ' (1991b, 125). None of the individuals found by Wheeler, of course, can ultimately be shown to have died in defence of the hillfort; we know only that they were laid to rest within the east gate in what was, by the first century AD, already a wellestablished burial ground. The reuse (or hijacking) of existing earthworks for burial appears to have been a defining aspect of Durotrigian inhumation, the deposition of human remains effectively rewriting the significance and meaning of earlier monuments (Russell et al. 2014, 220-1;Russell et al. 2017, 108-9).
More considered examination of the graves unearthed in 1936 and 1937 has demonstrated that, contrary to the view that burial had been 'carried out anxiously and hastily without order' (Wheeler 1943, 62), the funerary rite followed was, in reality, careful and well-ordered with regard to body position and the selection of grave goods (Fig. 3: Sharples 1991b, 119-25;Redfern and Chamberlain 2011, 69-70). Even more damaging for the war cemetery hypothesis, although it

MILES RUSSELL
OXFORD JOURNAL OF ARCHAEOLOGY 327 was apparent that over 74% of individuals represented had died violently, with evidence for targeted blows to the head and upper torso (Redfern 2011, 131-3), specific episodes were many and extremely varied in date, suggesting that the population had lived through multiple periods of stress, competition and conflict from the first century BC on until at least the mid first century AD (King 2014;Western and Hurst 2014). Since the completion of Wheeler's fieldwork at Maiden Castle, the excavation of other hillforts across Britain has shown evidence for 'defensive activity, signs of burning at the gates, charnel pits and ammunition dumps', apparently reflecting major periods of social instability and acts of aggression throughout the Iron Age (Cunliffe 2003, 77). At Danebury, in Hampshire, the remains of over 100 individuals were recorded, the largest group of deposits relating to skulls or fragments of skull (Cunliffe 1993, 105;Sharples 2010, 261). Evidence for injury, where detected, came in the form of spear thrusts and sword blows to the head, decapitation and knife cuts 'of a kind that might have resulted from scalping' (Cunliffe 1993, 53). Whether such trauma ultimately derived from skirmishes, raids, single combat, sacrifice, execution or endemic warfare (Hooper in Cunliffe 1984, 465-73;Cunliffe 2003, 74-7, 149-56), it is clear those living in the southern British Iron Age were quite able to inflict multiple forms of body horror upon one another without the assistance of Rome. Evidence for a Roman artillery barrage at Maiden Castle hillfort, famously presented by Wheeler in the excavation report (1943,62) and thereafter (Wheeler 1956, 105), is also ambiguous. The famous ballista bolt, found in the spine of one of the cemetery skeletons (Wheeler 1943, 63, 352), arguably one of the most famous discoveries from later British prehistory (Fig. 4), is actually a simple projectile weapon, such as a spear or javelin, and is possibly even of native manufacture (Smith 2017, 149-50;Stewart and Russell 2017, 160-2). Of the other bodies recorded from the area of the east gate, only one possessed an injury consistent with a Roman projectile (a square-shaped perforation in the left temporal bone: Wheeler 1943, 352-3;Redfern 2009, 410-11;Redfern and Chamberlain 2011, 70), but this is perhaps more likely to have been caused by the head of a javelin than an iron-tipped ballista bolt (Stewart and Russell 2017, 160).
Another aspect of the Battle of Maiden Castle, as related by Wheeler (1943, 63-4, 118) concerned the destruction of roundhouses in the area of the east gate during the initial phase of the Roman assault. It has long been apparent, however, that evidence for burning here derived from sustained (non-war related) ironworking (Sharples 1991b, 125). The area of secondary forging and welding established within the eastern entrance during the Later Iron Age is, to date, one of the most Close up of skeleton P7A from the 'war cemetery' in the east entrance of Maiden Caste, showing an iron projectile, a spear or javelin rather than the more usually stated ballista bolt, embedded in the vertebra (© the Society of Antiquaries of London and Dorset Natural History and Archaeological Society -reproduced with permission). significant centres of prehistoric iron artefact production recorded from the British Isles (Sharples 1991b, 118), ore having been brought in from as far afield as the Sussex/Surrey/Kent Weald, north Wales and the south-west of England. Metal working, in many pre-modern cultures, was a process with strong ritual and religious associations (Hingley 2006, 217-18), smithing and smelting activities being set away from more normal forms of occupation. The establishment of iron manufacturing workshops within the eastern entrance of Maiden Castle would appear to indicate that, immediately prior to the arrival of a Roman army, this particular part of the hillfort no longer served as a point of access. The undeniable fact is that the presence of both a cemetery and area of metalworking in the eastern entrance to Maiden Castle represents a significant period of change in the function and overall use of the hillfort. Most damning of all, with regard to the theory that Maiden Castle was savagely attacked by a Roman legion, it would seem that by the early first century AD, the hillfort had been largely abandoned, no longer operating as a political or economic centre, nor indeed a state-of-the-art battle-fortress. The clearest evidence of a change in settlement focus came during Niall Sharples' investigations which suggested that sometime after 100 BC, 'many of the outwork ditches were infilled and that settlement spread out of the entrance between the banks and surviving areas of the outworks ' (1991b, 117). Maiden Castle was, by the first century AD, not a defensive emplacement, but a site whose ramparts were largely neglected and whose interior system of occupation had irretrievably broken down (Papworth 2008, 132-3).
Despite the battery of evidence that has, since the mid-1980s, been accumulating against the theory that a single, violent siege brought settlement at Maiden Castle to an end, the battle story persists (e.g. Field 1992, 26-9;Manning 2002, 32;Mattingly 2006, 98-9;Brown 2009, 177;Harding 2012, 179-85;Powell 2013;Sparey-Green 2015, 119). This is due, in part, not only to the emotive nature of the evidence revealed in 1936-37, but also to the dramatic and extremely convincing way in which the conflict was presented by Mortimer Wheeler to the wider world. To effectively break from this 'siege mentality', we need to see the Maiden Castle evidence for what it is, shorn from a rather simplistic, one-size-fits-all archaeo-historical interpretation. We also must also better document the context and circumstances under which the potent myth of the war cemetery first came into being. Only then can we more fully understand the significance of Maiden Castle within the later Iron Age and early Roman period. As Wheeler himself observed: 'so much for the story; now for its basis' (Wheeler 1943, 62) METHODOLOGY AND MOTIVE That Maiden Castle hillfort has such a hold on the public imagination, not to say a prominent place in the archaeo-historical literature, is due in no small part to the two co-directors of the 1934-37 project, the husband and wife team of R.E.M. and T.V. Wheeler. (Notefor sake of clarity, in the discussion that follows, when referred to independently, Robert Eric Mortimer Wheeler (Mortimer to the public and Rik to those who knew him well) is simply Wheeler, whereas Tessa (following the convention set by Carr 2012, 22-3) is Verney Wheeler. The Wheelers were the premier archaeological partnership of the day, having, prior to Maiden Castle, already completed and published fieldwork at the predominantly Roman sites of Caernarfon, Brecon Gaer, Caerleon, Lydney Park and St Albans. They were active, not only in museum and society circles, but also within the developing world of professional archaeology, establishing the first Institute of Archaeology in London, to better teach and mould the discipline. They had, by the mid-1930s, 'never been more powerful, or more active, in the small, widely influential world of British archaeology' (Carr 2012, 238).
In 1933, Wheeler's eyes were starting to turn away from the Roman town of Verulamium (St Albans) where he and Verney Wheeler had been working since 1930, and to a wholly new archaeological objective. 'The mechanical, predictable quality of Roman craftsmanship, the advertised humanitas of Roman civilization, which lay always so near to brutality and corruption, fatigued and disgusted me' Wheeler was to later complain, noting that, as the Verulamium project came to an end, there was an opportunity 'to break away from the pretentious Roman machine and to transfer our large and experienced following to other aspects of that pre-Roman Iron Age' (1956,100). The target acquired was Maiden Castle, Dorset, one of the largest hillforts in the country, whose immense, sevenfold ramparts dominated the southern approaches to Dorchester.
Wheeler outlined his key fieldwork objectives for Maiden Castle early in the final published excavation report. Simply put, these were: to investigate the structural history of the great fortifications; to identify and correlate the associated cultures; to explore the possibility of recovering some part of the prehistoric town plan (Wheeler 1943, 3-4). These objectives were commendable enough, although, with the benefit of hindsight, a fourth and perhaps ultimately more significant motive for investigating Maiden Castle hillfort can be identified.
The Wheelers were both firm adherents of what today may be termed historical archaeology; linking excavated material remains with an established sequence of events. In his preface to Archaeology from the Earth, Wheeler reminded his readers that 'the archaeologist is digging up, not things, but people' and that archaeology itself 'is a science that must be lived, must be seasoned with humanity ' (1954, 13). Many of the excavation reports produced by Team Wheeler, from the work conducted at Caerleon (1928) to those of Verulamium (1936), were infused with a distinct archaeo-historical perspective, drawing multiple connections between recorded physical remains and documented characters and events. Maiden Castle, one of the most impressive hillforts in Britain, undoubtedly provided a wealth of potential opportunities for the Wheelers to link archaeological remains with an apparently well-established historical narrative.
For the Wheelers, there was only one event in Late Iron Age Dorset that really mattered: the arrival, in late AD 43 or early 44, of an all-conquering Roman legion, the II Augusta, led by the general (and later emperor) Titus Flavius Vespasianus. According to the second century Roman historian Gaius Suetonius Tranquillus, Vespasian had, following the primary stages of the Roman invasion, led the II Augusta into battle against the Britons, reducing to submission 'two powerful tribes, more than twenty towns, and the island of Vectis' (Suetonius, Vespasian 4). Vectis being the Isle of Wight, Wheeler felt that, of the unnamed tribes mentioned by Suetonius 'it is a good guess to affirm that one was the Durotriges of Dorset', whilst of the twenty towns it would have been 'an unthinkable insult to our most famous earthwork to exclude Maiden Castle' (Wheeler 1956, 104).
For all his protestations that Roman humanitas 'fatigued and disgusted' him, it is hard to believe that the prospect of finding definitive evidence to support a military campaign against the southern British tribes was not an exciting proposition for Wheeler, especially as he was to later claim that Maiden Castle had been an objective on which he had 'secretly meditated for several years ' (1956, 100). In retrospect, he was perhaps not being entirely truthful when stating that his motive for examining the hillfort was the desire to completely avoid anything Roman. In any case the hill had, prior to his arrival in 1934, already produced evidence for at least one Roman structure together with 'finds of the type Verney Wheeler loved bestcoins and mosaics' (Carr 2012, 211-12). Wheeler's citation of three specific fieldwork objectives, none of which directly mentions Roman archaeology, appears, at best then, somewhat disingenuous.

DEALING WITH THE PRESS
The Wheelers were adept at utilizing the press, working closely with newspapers not only to educate and interest the wider public, but also to help alleviate the severe financial burden of fieldwork and post excavation. 'Our more conventional archaeological friends' Wheeler was to later note, 'sometimes raised their eyebrows and sniffed a little plaintively at 'all this publicity of the Wheelers"! But we were not deterred' adding that, aside from the academic aspect of a particular project, any publicity generated was vital 'because this same public was incidentally contributing in gifts no small part of our considerable funds ' (1956, 102). In the final published report for Maiden Castle, Wheeler provided more specific information on the dig finances and the nature of the public contribution, recording that, of the total project expenditure of £5,363, £3,307 was obtained 'as the result of the printed appeal' whilst £1,266 'was received in gifts or in the form of profits on publications' from visitors to the site (1943,3). So large a financial contribution, Wheeler observed, 'speaks eloquently for the increasing interest of the general public in archaeological discovery' adding, rather tellingly, that 'the local and national newspaper press deserve special praise'. Jacquetta Hawkes put Wheeler's gift for capturing the imagination of both press and public more succinctly when she noted that 'it can hardly be doubted either that the interest and occasional excitement built up by what the critics regarded as a ballyhoo stimulated a generous response to the private appeals ' (1982, 97).
As a result of the press attention surrounding their fieldwork at both Verulamium and Maiden Castle, the co-directing husband and wife team were fast becoming media personalities, Wheeler apparently with great enthusiasm, Verney Wheeler less so (Carr 2012, 199). Wheeler's father, Robert, had worked for the Yorkshire Observer and much of his early life had been coloured by journalism, learning from a young age 'to write for and attract ordinary readers' (Carr 2012, 199-200). Indeed, Wheeler's skill at getting to the heart of a good story, and telling it well, helped him throughout all aspects of his professional career, especially when, following the arrival of television in the 1950s, he became the public face of archaeology (Hawkes 1982, 298-308). From the excavation of the Roman amphitheatre at Caerleon in 1926, largely funded by the Daily Mail, to the 1930s British Movietone newsreels covering work at Verulamium, Wheeler was acutely aware of the benefits of good publicity, he and Verney Wheeler arguably being the first archaeologists 'to fully, efficiently exploit the press' (Carr 2012, 199) .
Wheeler seems to have relished the opportunity to popularize archaeology and, in particular, those projects that he and Verney Wheeler were engaged in. 'Anyone' Jacquetta Hawkes was later to note 'who has a powerful personality, energy, independent ideas and a forceful way of expressing them will attract hungry journalists or other newsmen ' (1982, 98). Whilst Wheeler himself seemed never happier than when in front of the camera, Verney Wheeler appeared to have 'hated the personal focus it placed upon her' (Carr 2012, 201-2), being an extremely private person and 'the last person to enjoy the drama of the press'. Increasingly, it seemed that it was the more public side of the Maiden Castle project that engaged Wheeler, with Verney Wheeler, in the first two years of the excavation, concentrating more on the developing site strategy and the day-today management of works, particularly recording.
An increasing dependency on 'the story' to generate financial interest did have a downside, however, for the 'omnipresence of the press, and of the public via the press' increased the 'pressure to make dramatic finds' and it is perhaps no wonder that the Wheelers appear to have sometimes 'discreetly 'saved' something until they felt a site needed attention' (Carr 2012, 201). Although it could be argued that this was really no more than playing the game, sleight of hand and certain subtle archaeological deceptions never really affecting the nature of final report, it is worth being aware that not everything presented to journalists during the course of the Maiden Castle dig was exactly how it first appeared. 'The press' Wheeler was to later note 'is not always accurate and does not always emphasize those aspects of an excavation which are scientifically the most important', adding that 'sympathetic help from the directors of excavation is the best corrective of these failings' (1943,3).
Today, archaeologists sometimes attempt to control the reporting of a particular site or discovery thorough use of specially formulated press releases, tempering the potentially more extreme aspects of a story prior to any journalistic mangling, whilst simultaneously trying to excite a prospective readership with, what can only be described as, the 'wow-factor'. Both the archaeological community and, to some extent, the greater public, are more savvy now than they were in the 1930s about the inconsistencies of journalism, aware that there is often a degree of artistic licence in the style of reporting, content and overall conclusions. When at last the final statement concerning an archaeological excavation appears in print, the more sensational of claims, often made at the point of first reporting, have usually been softened, following more detailed finds analysis, background research and considered debate. Unfortunately, in the case of Maiden Castle, the press release ultimately seems to have become, in no small part, the definitive conclusions for the site, as presented by Wheeler in the published report.

THE LOSS OF TESSA VERNEY WHEELER
A key factor in the way that the Iron Age cemetery at Maiden Castle was ultimately reported and interpreted, especially in the final publication (Wheeler 1943), related to the absence of the co-director, Mortimer Wheeler's wife and archaeological partner, Tessa Verney Wheeler. Verney Wheeler had unexpectedly died, following a minor medical procedure, in 1936, just before what was planned to have been the final season of fieldwork at Maiden Castle. As a consequence, she contributed nothing to the recovery, presentation or subsequent interpretation of the so-called war cemetery, an endeavour that was left to Wheeler.
We can, of course, never know how different the final conclusions, as presented by Wheeler in 1943, would have been had Verney Wheeler been alive to co-author and edit the text or indeed to temper some of the more spectacular claims. Certainly Wheeler, with his military and journalistic experience, seemed to prefer the grand sweep of history to help illustrate and explain phases of archaeological activity and, with post excavation and the compilation of the first draft occurring at the very start of the second world war 'it is not hard to see how he could violently magnify events' (Carr 2012, 235). Wheeler himself acknowledged how the shadow of global conflict affected his writing of the Maiden Castle report when, in the dedication (to Tessa), written in August 1941, he observed that 'the wreckage of the present has in these days been more instant to my mind than the wreckage of the past, and inter arma I have no heart for studentship' Would things have been any different with regard to the understanding and interpretation of Maiden Castle, especially the cemetery in the east gate, had Verney Wheeler lived to see the publication and would Wheeler have gone quite as far, with his full-blooded recreation of the hillfort's final moments, if the publication had been co-authored? Lydia Carr, in her biography of Verney Wheeler, thought that the answer to this second hypothetical question was an undoubted 'yes, if Verulamium is any guide' (Carr 2012, 235). Whilst it is undoubtedly true that both members of Team Wheeler adhered to a more historical form of archaeological interpretation, whereby contexts, sequences, features and constructional phases could be linked to documented characters and events, it is perhaps unlikely that Verney Wheeler would have gone quite as far as her husband in interpreting the violent demise of the hillfort. Arguably, she too would have likely linked the trauma evident on so many of the skeletons in the east gate cemetery with injuries sustained during a futile defence of the fortress, but the dramatic and lurid descriptions, together with the punchy style of journalistic delivery that appear in the 1943 report belong to Wheeler alone.
If the roles had been reversed, and Wheeler had died in 1936, with Verney Wheeler living and working on alone, the final publication, given her evident unhappiness at dealing with the press, might well have played down the significance of a putative storming of the hillfort by the II Augusta Legion. Of course, Verney Wheeler did not have the experience of war that Wheeler himself had acquired (and would again in 1939-45) and certainly did not possess his eye (or desire) for identifying the dramatic, some might say more sensational, aspects of a particular discovery. In any case, any assault upon Maiden Castle by a Roman army would, however savage, have been swiftly resolved, representing overall the shortest aspect of the site's history (Carr 2012, 235). A report on the excavations by Verney Wheeler alone would probably have concentrated more upon the lengthy sequence of prehistoric occupation inside the hillfort, together with the later Roman temple building, rather than a speculative violent and fiery demise.

THE MENTALITY OF WAR
Perhaps the most important aspect in both the final interpretation of the Iron Age cemetery at Maiden Castle and also its wider, more emotional, impact on the public, was, as already noted, the time and circumstances of Wheeler's writing and publication of the final report. Completion of postexcavation work had been delayed, first by Wheeler's archaeological expedition to France, investigating hillforts there, and then, in 1939, by the commencement of hostilities. Despite his attention being elsewhere, Wheeler having enlisted and already made significant preparations for war by August 1939, the Maiden Castle typescript was essentially complete by 1941, the finished volume appearing in 1943 'by which time the author had been through the Battle of El Alamein, had been promoted to Brigadier and was so completely a soldier that Maiden Castle appeared as a distant dream' (Hawkes 1982, 175).
It goes without saying that Wheeler's interpretation of the dataset gathered during the final season at Maiden Castle was influenced by his own extensive military experience. During the First War, he had served as a Major in the 76th Army Field Artillery Brigade, seeing action on a number of occasions, most notably at the Battle of Passchendaele, later winning the Military Cross. His exploits, documented in Still Digging (1956, and also by Jacquetta Hawkes in Mortimer Wheeler: Adventurer in Archaeology (1982,, demonstrate not only a keen, journalistic eye for colourful detail, but also an evident skill at recounting the fast-paced nature of combat with all its exhilaration and body horror. With that in mind, two examples of Wheeler's writings are worth noting here, the first relating to events occurring in preparation for an attack, late in the evening of Wednesday 21 st of August, 1918: 'A blinding flash, with a halo of flying sparks, lit up for an instant the long column of straining horses and flooded wagons. A team swerved, and the white eyeballs of the startled horses gleamed for a moment in the glare. The night closed its fist once more upon its prey, and the parting shell from the roadside gun added another note to the squelching hoofs and jolting wheels. For the second night in succession, the teams plodded slowly up the last shell-eaten hill-side towards the jagged tooth that had once been a large and prosperous farm. From this landmark the terrain sloped down towards a valley where, a mile away, the British front line wound like a rivulet amongst the foothills. On the opposite side of the valley, some six hundred yards further on, lay the German line with its outpost trenches, machine-gun posts and wire' (Wheeler 1956, 30-1).
Later, during action on the 23 rd August, Wheeler found himself advancing with his gun battery towards the enemy: 'As the battery dropped into action, a squadron of cavalry flitted by like shadows in the mist and a tank waddled up uneasily from the front and finally broke down beside the guns. Leaving the battery to settle in its new home, I rode forward with my staff of signallers in search of news and observation. No Man's Land in a fog is no place for an equestrian exercise, and the tangled masses of wire, the treacherous shell-holes and other obstacles soon brought us to our feet. The horses were sent back to such shelter as could be found, and with my staff I picked a tortuous way forward amongst the debris. In the fog the only sure guide was the broken road, and the Hun was fully awake to the fact. His barrage was now falling noisily on and near it. Dodging from bank to shell-hole, our little party crept forward until at last, with grateful hearts, we stumbled upon the old German main trench where it cut the road. The trench marked the second stage of our journey; in its comparative security we paused for breath' (Wheeler 1956, 35).
Compare now the interpretation of archaeological remains recovered from within the area of the east gate at Maiden Castle during the final season, as recounted by Wheeler in the final report of 1943: 'First, the regiment of artillery, which normally accompanied a legion on campaign, was ordered into action, and put down a barrage or iron-shod ballista-arrows over the eastern part of the site. Following this barrage, the infantry advanced up the slope, cutting its way from rampart to rampart, tower to tower. In the innermost bay of the entrance, close outside the actual gates, a number of huts had recently been built; these were now set alight, and under the rising clouds of smoke the gates were stormed and the position carried. But resistance had been obstinate and the fury of the attackers had been roused. For a space, confusion and massacre dominated the scene. Men and women, young and old, were savagely cut down, before the legionaries were called to heel and the work of systematic destruction began … That night, when the fires of the legion shone out (we may imagine) in orderly lines across the valley, the survivors crept forth from their broken stronghold and, in the darkness, buried their dead as nearly as might be outside their tumbled gates, in that place where the ashes of their burned huts lay warm and thick upon the ground' (Wheeler 1943, 61-2).
There is no doubting the pace and immediacy of the Maiden Castle narrative, which reads like a dispatch from the frontline and which also at once vividly sears itself into the mind of the reader. Nor is there any quarter given to those who may dare to question the primary evidence or, worse, doubt the overall interpretation, Wheeler characteristically prefacing his comments with the words 'what happened there is plain to see' (Wheeler 1943, 62), as if to silence all future debate. So powerful was his interpretation, that it is now almost impossible to think of the burial ground at Maiden Castle without recalling the 'confusion and massacre' in which Wheeler affirmed that 'men and women, young and old, were savagely cut down' by a brutal and unforgiving Roman legion.
The masterstroke, of course, certainly as far as Wheeler was concerned, was the identification of residue relating to a Roman artillery barrage. During the war of 1914-18, Wheeler had seen at first hand the importance of artillery to soften up an enemy: the discovery of bodies buried in the Iron Age cemetery of Maiden Castle with injuries apparently consistent with catapult fire seemed, on the face of it, too good to be true. 'One skull', Wheeler remarked 'showed the square piercing of a quadrangular Roman ballista-bolt, whilst another skeletonmost vivid relic of allhad an iron arrow-head embedded deeply in a vertebra' (Wheeler 1956, 105). These sad remains, together with the ash of 'burnt huts', naturally led him to propose that an initial phase of attack upon the Iron Age stronghold had been heralded by ballistae, putting down 'a barrage across the gateway, causing casualties at the outset' (Wheeler 1956, 105). This is, of course, exactly how the assault upon a well-defended enemy position would have been conducted if led by an experienced artillery officer such as Wheeler, and his harrowing narrative certainly had an impact upon subsequent archaeological research. At Hod Hill, also in Dorset, excavations conducted between 1951 and 1958 were very much influenced by the Maiden Castle dig, site director Ian Richmond suggesting that Iron Age roundhouses inside this particular hillfort had been the target of Roman artillery during the initial phases of a putative attack (Richmond 1968, 33). Richmond's interpretation of a short siege ending with Roman catapult fire at Hod Hill has largely gone unquestioned since first proposed, and yet here too the archaeological evidence is ambiguous, ballista bolts recovered having just as likely been fired from the Roman fort built in the north-west corner of the hillfort (during target practice) as from outside during a siege (Maxfield 1989, 25;Stewart and Russell 2017, 167-8). Unfortunately, certainty in the nature of the Maiden Castle war cemetery supported the belief that Hod Hill had also been attacked by Roman artillery, whilst the evidence from Hod Hill further lent credence to the idea that Maiden Castle had been besieged. Both sites were locked in an endless cycle of mutually assured interpretation.

STICKS AND STONES
It is fair to say that Mortimer Wheeler polarized opinion, especially in the years immediately after the death of Verney Wheeler in 1936. Some treated him as a hero, describing him 'with unqualified awe and admiration' (Hawkes 1982, 1). There were others, particularly (although not exclusively) in the field of academia, who were jealous of his success or who disliked his personality and increasingly promiscuous lifestyle (as recorded in some detail by Hawkes: 1982, 10-13, 135-7, 182-6, 189-92). In the preface to Archaeology from the Earth, Wheeler noted the natural, albeit unjust, principle whereby 'every generation is liable to belittle the achievements of its predecessors ' (1954, 15). As a prominent, successful and academically divisive archaeologist, Wheeler had, perhaps quite naturally, already experienced aspects of this.
In 1938, a review of Verulamium: A Belgic and Two Roman Cities, written by Nowell Myres, was extremely critical of the limited nature of the Wheelers' archaeological investigations, failing to be won over by the compressed nature of the text and the over-confident interpretation of features. 'The greater part of this review' Myres noted 'has been deliberately taken up not so much with the positive achievements of the authors' work at Verulamium as with what may be termed its negative results, the old questions which they have not answered, and the new problems which they have created but not solved ' (1938, 25). The effect on Wheeler was, by all accounts, electric: 'no one had ever ventured to criticize the master in print', something which because 'the attack was severe, came from a young man and concerned an important excavation made it the more heinous' (Hawkes 1982, 160).
Following the publication of the Maiden Castle report, Wheeler again came under some criticism for making assertions based on what seemed to be a limited and selective excavation sample, Christopher Hawkes observing that the published conclusions were 'at times apt to assume a background of evidence which they really have not got ' (1944, 156). Further dangers surrounding the use of historical sources and the over-interpretation of excavation datasets were repeatedly raised by a new wave of post-war archaeologists, primarily, as Jacquetta Hawkes was later to complain, 'as a platform from which they could sling stones' at Wheeler (Hawkes 1982, 176).
In later life, particularly as a regular contributor to radio and television, Wheeler became 'one of the best-known figures of our time' not only in archaeological circles but 'also in fields as diverse as those of the administrator, of the author, of the entertainer (in the best sense of the word) and of the soldier' (Rivet 1971, 1). Throughout this period, Maiden Castle, and especially the final two seasons which had focused more upon the war cemetery, arguably represented Wheeler's most famous, and certainly most popular, archaeological endeavours. It would be wrong to say that his position, in his final years, as elder statesman and paternal figure standing over British archaeology, in some way insulated his early work from detailed critical analysis, but there certainly appears to have been a sense of the untouchable surrounding his pioneering pieces of fieldwork, especially those projects conducted with Verney Wheeler. It was not until the start of fresh excavations at Maiden Castle in 1985, nine years after Wheeler's death, that some of his more flamboyant interpretations could be challenged (Sharples 1991a;1991b). The four seasons of excavations at Maiden Castle, from 1934-37, can be judged an unqualified success, despite Wheeler's later assertion that, of his three objectives (originally set out in the Maiden Castle report : 1943, 3-4), only the first two were achieved, the third, 'exploring the possibility of recovering some part of the town plan' lying beyond the reach of the dig team. In retrospect, however, the hope of understanding anything close to a coherent sense of the internal structure of Maiden Castle in the days before non-destructive geophysical survey, seems wildly optimistic. Despite this, the Wheelers' fieldwork effectively established the importance of Maiden

MYTHMAKERS OF MAIDEN CASTLE
OXFORD JOURNAL OF ARCHAEOLOGY 338 Castle, recording, for one of the first times in the history of archaeological fieldwork, complex stratigraphy 'in sufficient detail to assess the chronological history of a site continuously occupied for over 400 years and intermittently occupied for 5000 years' (Sharples 1991a, 1). Bringing the fieldwork dataset to publication in the space of four years was also, unquestionably, a major achievement, made all the more remarkable considering this was conducted in the aftermath of Verney Wheeler's death and in the face of many other commitments, such as the completion of the Institute of Archaeology, as well as the first years of the Second World War. Amidst all this, of course, one aspect of the Maiden Castle excavation deserves rejection: the absence of anything that could be defined as unambiguous, in situ evidence for a battle has effectively demolished the argument that the hillfort was defended and fought over at the time of the Roman invasion (Stewart and Russell 2017, 158-62). The hillfort had, it would appear, been largely abandoned by the early first century AD, the use of the eastern entrance for burial and specialized iron working reflecting the continued spiritual importance of the neglected site, but not large scale occupation. Absence of intensive settlement at Maiden Castle may, in turn, indicate why the hillfort was not reused by the Roman army in the immediate aftermath of the invasion, the [Colour figure can be viewed at wileyonlinelibrary.com] small garrison, which some have postulated for the site (e.g. Todd 1984), receiving no support in the most recent geophysical survey (Stewart and Russell 2017, 109-13).
If there is no sign of extensive occupation at Maiden Castle, nor indeed of the remodelling of its defences, in the early to mid first century AD, the siege being no more than an over-optimistic interpretation of the excavation data, what evidence is there for settlement, either from within or close to the site at the time of the Roman invasion? A full magnetometry survey of the hillfort interior, undertaken during the course of the 1985-86 excavation, provided considerable evidence for settlement prior to 100 BC, including multiple storage pits, hearths, roundhouse gullies and roads (Balaam et al. 1991). At the time, a major success of the survey was the identification of an apparently D-shaped enclosure, measuring 40 x 45 m, at the centre of the hillfort, located in a dry valley which separates the western and eastern ends of the hill (Balaam et al. 1991, 41). A break in the north-east line of the ditch circuit for this 'new' enclosure, was thought to indicate an original entrance, whilst the absence of magnetic anomalies inside the ditch suggested an internal bank. Although unexcavated, the overall form of the enclosure together with its location, implied a constructional date in the later Bronze Age (Balaam et al. 1991, 41;Sharples 1991b, 96). Resurvey of the hillfort interior (in 2013: Fig. 5) has helped clarify the shape and dimensions of the site, suggesting it was originally more of a flattened oval (c.40 x 75 m), whilst also demonstrating that, along all sides, it was overlain by Iron Age roundhouses (Stewart and Russell 2017, 111-12).
Two further ditched features, largely indistinguishable from the wealth of background noise seen in the 1985 survey, were far more clearly defined in the magnetometry survey of 2013 (Stewart and Russell 2017, 111-12). These two enclosures (Fig. 6) appear to comprise a polygonal (six-sided?) ditched feature, measuring 30 x 25 m, with possible entrance breaks along the northeast and central south-west edges, abutting (or joining) a flattened oval enclosure, measuring 30 x 20 m, with possible entrance breaks on both the north-west and the east (and more pointed) end. The precise nature of the relationship between the two ditch systems is unclear, although an obvious absence of overlap may indicate that both were broadly contemporary. Each appeared to contain a single ring ditch and several pit-like anomalies clustered together in small groups.
The two polygonal enclosures appear broadly similar in form to examples of Later Iron Age, Durotrigian farmsteads previously excavated at Gussage All Saints (Wainwright 1979), Rotherley (Pitt Rivers 1888), Tollard Royal (Wainwright 1968), Tolpuddle Ball (Hearne and Birbeck 1999), Winterborne Kingston (Russell et al. 2017) and Woodcutts (Pitt Rivers 1887), which were all occupied from the first century BC until at least the mid-first century AD. The most recently examined of these, at Winterborne Kingston to the east of Dorchester, was defined by a two-phase polygonal (kite-shaped) enclosure (Fig. 7), covering an area of 50 x 30 m internally with a single, 2 m wide, south-east facing entrance. Despite the absence of clearly defined buildings, a large number of pit features were detected within the area of the ditch. A total of 11 pits were fully or partially investigated, their fill comprising charcoal, baked clay and other burnt material. Activity within the enclosure appears to have ended abruptly, the earthworks being deliberately levelled, and the ditches backfilled with large amounts of cultural material, including a series of complete prehistoric pottery vessels, in the mid to late first century AD. This may suggest sudden abandonment, the land perhaps being cleared in order to make way for more intensive forms of agricultural practice, in the early Roman period (Russell et al. 2017, 110).
If, as seems likely, the recently defined polygonal enclosures at Maiden Castle represent two, roughly contemporary, farmsteads of Late Iron Age, Durotrigian affiliation, this would further indicate that settlement within the hillfort at the time of the Roman invasion was limited in the extreme. This was no densely occupied tribal town then, whose inhabitants rallied en masse to protect their economic and spiritual centre from a Roman legion, but, rather a sparsely inhabited space, largely gone over to agriculture, set within the long-abandoned ramparts of a politically defunct hillfort. The people who worked and lived on the hill in the early years of the first century AD cannot ever really have numbered more than a few dozen and, although it is certainly possible that some inhabitants were represented in Wheeler's east gate cemetery, we cannot know for sure the precise circumstances of their death. Whether any were killed in a futile attempt to protect their farm, were executed by a newly installed Roman government or, indeed, died fighting Rome (or some other enemy) elsewhere only to be brought back for interment within an ancestral plot, we have no way of telling. Whatever the case, the results from the most recent geophysical survey at Maiden Castle (Stewart and Russell 2017, 106-13) represent one more, and arguably final, nail in the coffin of Wheeler's great siege theory.