The Lost Legion
2012. For a possible further 18 acre camp identified since publishing see addendum at the end of this article (end of Pt.3).
In our article – “The Eagle of the Ninth, What Really Happened” – we set out in 2007 to address the probable fate of the famous Roman Ninth Legion – the IX Hispana – which mysteriously disappears early in the 2nd C AD from the historical record. In so doing we challenged the prevalent view these days that the legion must have been defeated in battle somewhere other than in Scotland.
The assumption that the legion was lost out-with of Scotland (the legion was then stationed in Britannia) has since the 1970’s become fashionable, an anomaly attributable to the persuasive power of constant repetition by a vocal minority.
However there is no quality evidence – either artefactual or epigraphic – that supports these pieces of guesswork, a state of affairs that makes them simply smack of wishful thinking.
On the other hand in our Eagle of the Ninth article we looked closely at the epigraphic references which the ancients themselves made of these times; tantalising records giving glimpses of endemic warfare with the tribes of ancient Scotland and Rome’s setbacks with noteworthy troop casualties here in the troubled years of the early 2nd C AD.
Rather tellingly this is all occurs around the same period of time as the last accepted surviving artefactual and epigraphic record of the Ninth legion in the archaeological body of evidence.
Now this is a sounder basis to work with.
Since 2007 two things have happened:
First: Roman Scotland identified the site of the battle of Mons Graupius (83 AD) at the Clevage Hills next to Dunning in Strathearn in 2009.
This has been well received due to the methodical approach adopted to analyse competing sites. It demonstrated that we can still identify the site of otherwise “lost” actions in history with good probability if we look at all the available information methodically and - most importantly – evenly (link to Mons Graupius article).
Second: two movies are currently in production whose plots revolve around the Ninths loss in Scotland.
Good….and frankly long overdue!
Clearly we do not yet know the quality of these movies however anything that promotes among Scots a consciousness of Rome’s activities in Scotland can only be a good thing, and, will help put paid to the all too prevalent myth that the Romans stopped at Hadrian’s Wall, never venturing north thereafter.
The first film; “Centurion” by writer/director Neil Marshall is potentially set to follow the horror genre employed in “Dog Soldiers” and promises to be an enjoyable light hearted romp. Filming has taken place in and around the Glenfeshie Estate in Inverness-shire which in itself however is a bit of a cause for concern.
Image courtesy of Aaron Sneddon, Scottish Contract Press Photographer
The second film “The Eagle of the Ninth” is currently receiving headlines as it has signed up a brace of well known actors. It is directed by Kevin MacDonald and is set to bring Rosemary Sutcliffe’s well known 1954 novel to the big screen, a long awaited airing since the - now cult - BBC serialisation of 1977 was inexplicably never re-screened.
It is little wonder therefore that nearly everyone we meet in the field is avidly discussing these new films.
Indeed Roman Scotland friend Jim Brown from Heddon on the Wall has gone one step further and challenged us with finding the site of the legions loss using the same methodical approach we used in the Mons Graupius search.
This did not take much persuasion. Our Eagle of the Ninth article originally set out to challenge the prevalent theories noted above for the legions loss somewhere other than Scotland. However with Scots interest in this period thankfully growing now is the right time to take our original hypothesis in the Eagle article forward.
Well the films for cinematic effect will no doubt sow yet again the vivid Hollywood ideal of scenic and mountainous highland landscapes in the public consciousness of all things concerning Scotland – filming has already taken place as mentioned above in Glenfeshie in the Cairngorms.
It should be remembered however that the Romans appear never to have set foot within the Highland massif in strength.
And surely we must set our mark in writing now, before the films receive the almost inevitable hoots of derision from the usual academic suspects, a reaction that yet again will merely confuse the debate over the legions probable loss in Scotland.
So now is the best time to present our hypothesis. By using a similar approach to that used in our Mons Graupius article we can suggest a possible setting for the legions last march and its ominous terminus.
In our Mons Graupius article we looked at four key factors when analysing a hefty list of sites competing for the honour of being recognised as the scene of that battle. These were:
- The Primary Sources; i.e. what the ancients themselves wrote,
- The Archaeological Source; i.e. relevant remains surviving on the ground from antiquity, in this case marching camps,
- A Practical Analysis of the Sites; investigating any given locations strengths and weaknesses in a manner devoid of preconceived ideas or partisan bias,
- The name “Mons Graupius”; some very few place-names still exist which can be traced back to names recorded in antiquity.
Clearly in the context of the search for the Ninth we have no detailed commentary available to us like Tacitus’s “The Agricola”. Nor do we have a name for the place where the Ninth Legion was lost, so we are unable to consider modern names and analyse the applicability of their likely ancient antecedents.
However locating the site of Mons Graupius was no easy task. We are not therefore disheartened by academic’s remarks that “we know nothing of these times” as fortunately that is far from correct.
So what are the “factors” available to us in our investigation over the loss of the Ninth?
- The primary written sources from antiquity,
- The archaeological source in the shape of marching camp remains.
Further we will consider the tribes known to have been in conflict with Rome as well as casting a canny eye over the landscapes that these factors point towards.
The Primary Written Sources.
We do not have Tacitus’s remarkable work the “Agricola” to rely on, however it should be realised that the survival of such a detailed account into modern times is a fabulously rare occurrence, indeed a great boon for the study of the early Roman period in Scotland as a whole.
The vast majority of ancient texts relating to ancient times are much shorter and fragmentary. It is common practice for historians to relate many of these different sources from similar periods together to help “build the bigger picture”.
In this we are fortunate to have the short – but relevant - surviving remarks from Juvenal, Spartianus and Cornelius Fronto.
These ancient sound-bites taken as a whole begin to paint a telling picture of the period we are looking at, and they remain the unique point of reference anywhere for the search for the lost Ninth.
What are these ancient sound-bites?
Aelius Spartianus (probably late 4th C AD) is the author of the biography of the reign of the Emperor Hadrian (117- 138 AD) in the Scriptores Historiae Augustae. This early section of the SHA is considered reliable, often directly quoting the respected 3rd C AD historian Marius Maximus.
In chapter 5 of “The Life of Hadrian” Spartianus wrote;
“On taking possession of the Imperial power Hadrian at once resumed the policy of the early emperors, and devoted his attention to maintaining peace throughout the world. For the nations which Trajan had conquered began to revolt; the Moors, moreover, began to make attacks, and the Sarmatians to wage war, the Britons (sic) could not be kept under Roman sway,”
As there are no signs of disturbances by the indigenous tribes of Wales and England this can only be a reference to the lands of southern Scotland from which the Romans had only recently been forced to withdraw but which they probably still considered to be under Rome’s control.
Cornelius Fronto (100 – 166 AD) was an orator from Cirta and lived through the period we are looking at as a young man. Some of the letters he later sent as an old man to the Emperors Marcus Aurelius and Lucius Verus (dateable to the early 160’s AD) survive. That of interest to us here is a letter he sent to Marcus, consoling him over Roman losses in the east early in his reign (161 -162 AD);
“…. And again when your grandfather was Emperor, how many soldiers were killed by the Jews, how many by the Britons?”
Decimus Iunius Iuvenalis (Juvenal).
The poet Juvenal (late 1st – early 2nd C AD) - author of the Satires – makes a tantalising side reference to war in Britain around 105 AD. Here he refers - in Roman upper class salon talk - to the barbaric antics of the tribal chief Aviragus in the then current troubles in the south of Scotland.
This slightly precedes the events of 117 AD but - when taken with the archaeological body of evidence of destruction in southern Scotland and along the Stanegate in 105 AD - it assists in painting a tantalising picture of the troubled years immediately preceding Hadrian’s ascension to power in 117 AD.
Vindolanda Tablet 164
The nature of these warriors is recorded in the famous Vindolanda Tablets (link available in our Link section). These date in the main to the very late years of the 1st C AD and precede the period we are concerned with here by only a handful of years.
The excavators considered that many of these tablets are the surviving remains of shoddily burnt rubbish originating from the Commanding Officer house.
Tablet 164 is of particular interest and may have once formed part of an intelligence report. Here it describes the fighting nature of the belligerant tribesmen "Britunculli" - a derogatory term which translates along the lines of "Wretched Britons".
The clear reference to warfare locates these tribesmen as hailing from the troubled area to the north of the Stanegate line on which Vindolanda sat. The local tribes around and to the south of the Stanegate had been succesfully subjugated - and most probably disarmed - for over a generation. Therefore this will in all probability have been a description made of the combatative tribes of southern Scotland. It reads:
"..the Britons are unprotected by armour (?). There are very many cavalry. The cavalry do not use swords nor do the wretched Britons mount in order to throw javelins".
The Archaeological source; marching camps.
Scotland has the finest tally of marching camp sites known anywhere.
This is an archaeological legacy of ongoing warfare spanning over three hundred years. With improved technology the list will undoubtedly increase as further camps come to light.
These marching camps were the overnight bivouacs of Roman armies on campaign. Such a great number of these survive in Scotland that it is possible – once a few basic rules of “camp morphology” are understood - to attempt to track the progress of specific ancient armies on campaigns.
How can we do this?
By analysing both the size (which suggests the likely number of troops the camp may have contained) and morphology of the camps layout (which helps define certain general periods of likely use based on known structural sequences elsewhere in Scotland) we can begin to suggest relationships between those camps currently known.
This can be done with reasonable conviction even between camps spaced quite far apart. These are the fragmentary remains of the once-mighty chains of camps which marked the progress of many Roman armies on campaign in Scotland. Those camps in-between these known ones either remain to be discovered or are lost forever under modern development or the abrasive action of the plough over the intervening centuries.
This is an incredibly fruitful – if speculative - line of enquiry; one where we feel Scotland - not least with recent success identifying Mons Graupius – should and can be to the forefront.
(Link to Marching Camps article).
Those tribes with a track record of conflict with Rome.
Common misconception envisages the Romans combating only those tribes of Scotland’s northernmost reaches. This is fundamentally incorrect.
By setting aside such preconceptions and relying instead on what the primary literary sources from antiquity tell us (over a considerable span of time) we can glean a fair understanding of which tribes were frequently at odds with Rome’s hegemony in the British Isles.
This gives us another extremely potent line of investigation.
Further, sufficient -if incomplete - knowledge is available from ancient cartographers to give a reasonably good picture of the distribution of the tribes of ancient Scotland (link to Ptolemy’s map of Scotland).
How does this help us?
It enables us to focus on marching camps which sit in the problematic lands of those notably intractable tribes.
Using camp morphology we can further refine our search to those camps of a size and morphology which is of interest. Finding relationships between any such camps allow us to suggest a sequence and route as well as - most importantly - to plot the forces destination.
A practical analysis on the ground.
It is absolutely imperative that a quest like this is undertaken out in the countryside, not as an abstract academic exercise within the cloistered comfort of a Classics Department.
A keen eye able to read the ground will never go amiss, indeed such skill will be crucial in identifying that most difficult terrain where the Romans are factually known to have gone; not the scenic mountainous Highlands which Hollywood would have us (incorrectly) believe to be the setting for all interaction between invaders and the peoples of Scotland throughout the centuries.
What sort of conflict should we be looking for?
The lack of any further definitive records of set piece pitched battles between native and Roman after Mons Graupius (83 AD) strongly suggests that the tribes learnt the lesson from that fateful encounter.
Roman organisation, training and discipline usually gave them the advantage in traditional set-piece confrontations, a situation the tribes appear to have recognised after Mons Graupius.
Indeed it was recorded over a century later (Cassius Dio and Herodian on Severus’s 209 AD Scottish campaign) that the tribes successfully fought the Romans and inflicted serious casualties using hit and run tactics; a manner of taking on Rome with a greater likelihood of success.
An ambush therefore on a Roman column toiling through difficult terrain, or alternatively a nocturnal assault on their marching camp (the Caledonians it should be remembered almost succeeded against the same legion in 82AD under just such circumstances) is the best scenario to explain how the tribes may have engineered the circumstances where Rome’s battlefield superiority could be neutralised and tribal success best guaranteed.
So where did the events take place and how do our 4 factors noted above fit in?
Clearly the records made by Juvenal, Spartianus and Cornelius Fronto are sufficient to warrant the hypothesis we made in our Eagle of the Ninth article. This made two assumptions;
First; that the legion ventured north at less than full strength - allowing a plausible explanation for the existence for those very few individuals in later years who are claimed at some time in their lives to have been members of the Ninth.
We suggested therefore that perhaps as many of two of the legions cohorts may have been on vexillation service on the continent – a particularly common occurrence for the legion based in York, while it is reasonable to suggest that at least one cohort (perhaps even the double strength first cohort) may have been left as a skeleton garrison in the legions fortress at York.
There is of course no direct evidence from antiquity telling us this, or indeed contradicting it, this is merely a working hypothesis but one that we consider is compelling enough to merit further investigation.
The legion at full strength will have mustered somewhere between 5,200 and 5,500 men. The equivalent of 3 cohorts taken out of the equation reduces the force by around 1,500 men to some 3,700 to 4,000.
We also posited that the legion may have marched alone, without the support of brigaded auxiliary regiments; units who may have been left in their forts to better control a clearly disturbed frontier zone in 117 AD.
Secondly; as recent disturbances in southern Scotland appear to have penetrated into north east England along the line of Dere Street (on a similar line to the modern A68) we suggested that that this then is likely to have been one of the main areas of concern for Rome in Britain at this time.
Dere Street therefore is of continued interest in our search.
Identifying the likely culprits for the havoc dispensed in the early years of the 2nd C AD is no great historical whodunit, the finger usually being pointed squarely and convincingly at the Selgovae: the hill tribes of upland central southern Scotland.
This truculent tribe - or perhaps the name covered a loose confederacy of tribal septs - seem to have been an ongoing problem for Rome for most if not all of Rome’s dealings with ancient Scotland.
Certain forts built in southern Scotland during the first phase of the Antonine occupation around 140 AD – not least the reconstruction of Newstead on a grand scale with an unusual legionary / cavalry garrison- suggest that the territory on the edge of Selgovae territory was the problem that had either prompted Roman intervention, or conversely was considered so acutely problematic at the time to warrant special treatment all of its own.
Later in the 2nd C AD (following the withdrawal of troops from the mainline Antonine occupation) the Selgovae and their neighbours seem to have transmogrified into a tribal grouping known as the Maetae (link to the Tribes of Ancient Scotland article), a people who it appears caused Roman governors and the garrison of Hadrian’s Wall no end of problems in the late 2nd and early 3rd C AD.
By 208 AD enough was enough and the governor pleaded for help from the Emperor Septimius Severus. He in turn arrived with substantial reinforcements to bolster the beleaguered British garrison and campaigned first (in overwhelming strength) in Maetae lowland Scotland in 209 AD before then proceeding north against the Caledonians (with it appears a slightly smaller force). The Caledonians we are told had angered Severus by supporting the Maetae of southern Scotland.
Clearly these years had seen something of a field day with all the tribes of Scotland taking a metaphorical pop at the neighbouring super-state and even more worryingly for Rome by showing a degree of unity of purpose in the process.
However, of greatest relevance to ourselves here is that the largest camps in Britain - the massive 160 acre sites in southern Scotland - are confidently attributed to these Severan Maetae campaigns and hint at the puissant threat Severus had to contend with in 209 AD. Indeed Cassius Dio and Herodian record the incredible losses that Severus’s army suffered to tribal harassment and the commonplace modern assumptions that locates these events among only the more northerly Caledonians is recorded nowhere in the historical sources.
The Selgovae and their associated septs and satellite tribes, for long the sackers of Roman forts, and a tribal confederation of noted and long lived resistance to Rome seem therefore to be the most probable opponents that a small Roman task force (such as we are looking for) would seek to chasten in a punitive foray, located within acceptable striking distance of Romes’ frontier - then situated on the Stanegate line.
Why should we not think that they advanced further north, against the Caledonians?
Roman intervention appears to have been infrequent north of the Forth – Clyde line, the frequency – based on marching camp remains – apparently decreasing the further north you travel.
It is fairly safe to assume during the twenty or so years of the Antonine occupation that small forces could pass to and fro behind the Forth – Clyde running barrier in comparative safety as long as due diligence was paid in entrenching their overnight stops. And indeed this would appear to be confirmed by the small size of many camps commonly attributed to this time.
On the contrary and especially in the late period to which most camps in the far north belong -most camps here have the elongated proportions typical of late Roman marching camp morphology - Roman armies were staggeringly large, their camps rarely under 100 acres in size and often considerably larger. These are large enough to hold armies of well over twenty thousand men and in most cases considerably more.
To more fully appreciate this remarkable level of manpower and the enormous logistical muscle required for service against the Caledonians in the north of Scotland then a sense of scale is required through comparison.
In England (and even Wales where we are told Rome met stiffer though equally short lived resistance) the largest known marching camp is the Claudian invasion bridgehead work at Richborough – Rutupiae - postulated to be near 140 acres in size. This is unique and abnormally large for England and Wales where in the main camps of around 20 acres are considered large, those around 40 acres very big and in a very few cases only those approaching the 60 acre mark are considered exceptional.
The great bulk of camps in England in particular are tiny, usually around 1 to 4 acres in size and cluster in the main behind Hadrian’s Wall. They are in all probability labour camps housing troops employed on the construction or rebuilding works on the wall that were undertaken throughout the Roman occupation.
Richboroughs’ suggested size is extremely unusual for the south of Britain; and is simply a reflection of the size of the original invasion force and the area required to accommodate its store dumps prior to splitting into separate columns to pursue and -fairly rapidly- defeat the tribes of southern England in detail.
This size of camp would not therefore be repeated and subsequent years saw a staccato though regular expansion, radiating out from the far south east corner of England in steps between posts constructed from the start as permanent legionary fortresses.
The picture in Scotland - and particularly the north of Scotland - is much different and in the years which follow the period we are concerned with here, punitive military action was less the decision of governors, more the mandate of Imperial directive.
Incredibly, in an amazing number of occasions the ancient sources record direct intervention by the Emperor in person!
Indeed these were no sham show-boating performances like that of Claudius at Colchester. The large marching camps known in the far north of Scotland confirm that the armies that accompanied these Emperors were sized to suit Imperial prerogative and safety in a locality remote from the comfort and safety of the southern half of Britain.
While this speaks highly of Caledonian belligerency, as indeed do the even larger forces Severus brought to bear on the puissant Maetae of southern Scotland, it remains highly improbable that either the Governor or the Ninth Legions legate - even in a fit of foolhardiness - would attempt a penetration through hostile southern Scotland all the way to northern Scotland with such a small and unsupported force.
The events of 82 AD -when Agricola split his army into 3 smaller columns with near disastrous results- were never as far as we can tell repeated and as such it is unlikely that it was forgotten in a singular instance a mere 35 years later in 117 AD.
Further it is not entirely without profit to ponder over the fate of numerous similarly sized southern armies of later ages who, engaged in raids in southern Scotland, suffered spectacular defeat to the hands of Scots, particularly when exercising insufficient caution such as at Piperdeen, Haddon Rig, Ancrum Moor, Roslin and Nisbet Moor to name but a few.
And finally, to underline the point no Roman marching camps (currently known) survive in the far north of Scotland small enough to suitably accommodate the modest numbers we have suggested were involved in the most probable scenario surrounding the loss of the Ninth Legion.
So our working hypothesis develops; a force of around 4,000 men advance into Selgovae territory in southern Scotland’s hill country along Dere street to around the burnt out husk of the fort at Newstead.
Curle’s investigations into the Newstead fort site in the early 20th century unearthed an incredible tally of abandoned Roman equipment buried on the site, dateable in the main to the forts destruction around 105 AD.
These artefacts, now mostly held in the national Museum of Scotland in Edinburgh are an amazing and very tangible “real” link to these days in Scotland, the very same period that some academics would have us believe we know nothing of! Nonsense!
How did the legion get there, and did they go beyond this?
To answer this we must look at the fossilised footsteps of Roman armies in time - the archaeological remains of marching camps - and consider if we can piece together a possible picture of the Ninths movement based on our 4 factors.
Obviously not all marching camps remain to this day, or their remains are so slight or elusive that they still remain to be detected. However we need not shackle ourselves to the negative opinion of those who would claim that marching camps tell us nothing. True, they are notorious for the scarcity of dateable artefacts recovered in them, however as discussed in our Mons Graupius article several generally regarded “series” of camps can be deduced based on the camps morphology; i.e. its length by breadth proportions and some stylistic features - “clavicular” gates for instance (link to marching camps article).
Unlike Mons Graupius though we do not have the sure square or sub-rectangular plans of Flavian period works to rely on. This classic shape has a proven association with the earlier Flavian period (79 – 96 AD) campaigns in Scotland.
They are datable by artefacts found in some key sites such as Carey but in the main by the structural sequence of certain key sites in relation to certain other more readily datable features such as roads and permanent forts. However this Flavian camp morphology need not necessarily remain true for 117 AD, some 34 years after Mons Graupius.
Some small square or sub rectangular marching camps – more accurately described as work camps- around Hadrian’s Wall can be attributed in many cases to that frontier defences original construction which commenced in 122 AD, a mere 5 years after the troubled times we are concerned with.
Some of the great many other small camps here may belong to this time too, however it is possible that many of these more elongated examples are the bivouacs of troops employed on any one of the great many later reconstructions of the mighty barrier following violent attack from the north as well as more mundane deterioration in its fabric.
So if these square or sub rectangular camps do indeed belong to this period then it readily suggests that such forms may perhaps still have been the norm. However we feel it is fairly safe to consider that slightly more elongated camps could equally be attributable to this period as well given the evolution in their proportions that camps underwent throughout the 2nd C AD (as recorded by Pseudo Hyginus).
Writing his military manual much later in the 2nd C AD he advocated tertiate proportions (1:1.5) to camps and suggested each legion (around 5,200 to 5,500 men) would require a mere 15 acres to accommodate them in such temporary works.
Hyginus’s figures are hotly contested though, his work is often considered to be an ancient theoretical desktop exercise that paid scant regard to the real life practicalities of cramming such a large number of men, beasts and equipment into such small enclosures nor of reflecting the practical exigencies of armies on campaign.
What is known is that Hyginus work post dates the events of 117 AD.
Polybius’s older but more realistic calculation, which we reasoned was still in force during Agricola’s Flavian campaigns was less congested, giving a much more practical and satisfactory figure of around 25 acres per legion.
Finally it should be noted that Hanson has suggested a figure of 335 men per acre, approximating to 3 acres per thousand men. This figure was not based on any ancient source; rather it appears to have been suggested in an attempt to reconcile the differences between Polybius and Hyginus’s figures. However it still appears unrealistically small and as it not supported by any reference in antiquity we shall leave that figure alone.
The safest route for us therefore is to continue to follow Polybius’s model in 117 AD though to acknowledge that innovations in camp densities may have come into vogue later in the century, particularly in the post Antonine period.
Distilling Polybius’s 25 acres per legion arithmetically gives us:
25/5.5 = 4.5 acres per thousand men.
In the Eagle of the Ninth article we suggested the Ninth may have marched north with a force somewhere between 3,700 and 4,000 men.
Using the rule of thumb above this force would require its marching camps to be in the region of 16.5 acres to 18 acres give or take.
(In the interests in fairness and for the benefit of Hyginus’s die hard advocates the areas which can be extrapolated for this size of force using his suggested 15 acres per legion of 5,200 to 5,500 men would be in the region of 10 to 11 acres. However as mentioned above we feel this lower figure and density of camp was unlikely ever to have been achieved, even later in the century).
Further allowance should be made for a degree of discrepancy in the physical setting out of consecutive camps defences in any given “series”. Setting out after all in ancient times regardless of the surveyor’s handy rule of thumb ratios still remained firmly in the realm of “man in the loop” technology, entirely dependent on the soldiers’ ability to pace out the distance and direction accurately.
Such discrepancy is almost impossible to predict, but where lengths of camp defences take a distinct deviation in line - usually to miss difficult geographical features - then the actual area enclosed by the resultant camp defences will vary considerably from the area originally planned.
To compound matters, where high ground obscures the surveyors direct line of sight to all four corners of the camp then misalignment of these corners can be expected, a matter which increases the level of error in the camps constructed area compared to the surveyors original intended area.
©2009 Roman Scotland. All Rights Reserved
First Published November 2009