The Lost Legion (Part 2)
So what camps are known that may suggest the route taken by the Ninth?
There is a dire paucity of marching camps of this size heading north out of the legions base at York (Eboracum).
The trail therefore is effectively cold up to Corbridge (Corstopitum). This noteworthy Roman station is located at the junction of the lateral Stanegate road and the north-bound Dere Street after it bridged the River Tyne. However at this time the post was simply an auxiliary fort.
Rather interestingly it had been built by members of the Ninth legion to replace the earlier Flavian fort on the site which appears to have been devastated during Aviragus’s successful invasion of Roman Britannia in or around 105 AD and is a nice link with the unit we are concerned with.
The Corbridge hoard, a chest containing scrap iron, is dateable to this period, and it is on the basis of this find that it is possible to reconstruct Roman Lorica Segmentata armour of the late 1st C AD. The hoard, and its loss is yet another nice and very tangible link with the years we are looking at.
Lying only a short distance to the west of Corbridge lay the remains of Agricola’s supply station at Red House, built to supply his advance into Scotland in 79 AD and the years of campaigning that followed. It appears however to have been abandoned fairly soon after this, and certainly by the time the first Flavian fort at nearby Corbridge was built around 90 AD.
At 25 acres Red Houses’ ramparts could conceivably have been pressed into service as shelter, however the lack of marching camps to the south of here would appear to suggest peaceful conditions in Brigantia, a situation that may have merited nothing more defensively formidable than a perimeter of timber stakes.
This distinctly minimal level of security would merely prevent unauthorised coming and going and discourage petty theft. All of which is indicative of the peaceful conditions prevailing in the north of England.
North of Corbridge matters rapidly change. Brigantian tribal lands or hegemony extending north of the Stanegate (later Hadrian’s Wall) has been much speculated and certainly greatly overstated.
What is certain is that by the time our column of troops reached High Rochester (Bremenium) they were deep in Votadini territory.
(Note; the current Brigantium archaeological reconstruction centre here is misnamed.
The information contained in Ptolemy’s map lists Bremenium among the lands of the Votadini.)
Before this however the force passed the Roman station at Habitancum. The fort here – Risingham- is probably a later construction while the marching camp (West Woodburn) at 27 acres (and most probably squarish in plan) is likely to be either contemporary with our period or belonging to the earlier Flavian one.
Notwithstanding this, at around 27 acres the camp is too large for us so our force marches on until, near the currently undated fort at Blakehope on the River Rede we come across our first serious candidate site for our forces overnight stop; Dargues.
The marching camp at Dargues has tertiate (1:1.5) proportions. This may raise some doubts about its applicability to the period we are looking at but as it is no more elongated than that we will tentatively accommodate its identification in our list of sites the legion may have used.
Its size; 14.8 acres is of interest, if marginally too small. There are however no notably difficult features on the site that may explain an error in setting out. Further it is close in size to another group of camps of around 13 acres in size that we shall meet time and again so it remains possible that the camp may be part of that group – particularly as that series contains more camps of classic tertiate proportions than we should perhaps anticipate in 117 AD. Tentatively however, we will note this as the first halt of the Ninth after leaving Corbridge – some 20 miles or so (a good days march) to the south.
Across the River Rede lies the field of Otterburn, scene of no less violent and calamitous conflict some 1271 years later when the Scots under Douglas crushed Percy’s larger army in a vicious moonlit battle in 1388.
The local tribe, the Votadini, are not in general terms thought to have been at loggerheads with Rome; however this territory, skirting the looming high ground is worryingly close to the likely lower reaches of Selgovae territory and here the Romans - near bandit country- looked rightly to their defence when encamped at night.
From Dargues the Roman road heads high; ever north westwards into the wild area around Otterburn training grounds. No camps here match the size we are looking for.
Not even at the abandoned or sacked Flavian fort at Bremenium (it likely shared Corbridge and Newsteads’ fate in the fighting of 105 AD) do we find one, and although it is ringed by marching camps of many styles and likely dates none match our criteria.
So onward our force marches till at the extremity of the Northumbrian National Park we come across the remarkable little Roman station at Chew Green, and, some 5 miles to the north west across the modern Scottish border the matching grouping of marching camps at Pennymuir.
The log jam of camps at these sites offers very interesting investigation for us, the sites being located either side of difficult high ground (above 400m OD) in the Cheviots.
As such these camps – not unlike modern motorway service stations - will have seen continual reuse so we should be careful about suggesting that any camp belonged to the movements of our footsore column coming up from Dargues (10 miles to the south over demanding ground) and no-one else.
That said, Chew Green has two marching camps of note. That of particular interest is an almost square camp measuring in at just below 19 acres. This is effectively within our broad target range of 16.5 to 18 acres “give or take”. This camp in all probability was built during the earlier Flavian period due to its square proportions and its clearly early position in the multi period sites “structural sequence”.
We can safely add Chew Green to the road map of our suggested march of the Ninth.
Before we move on it is worth casting an eye over the other camp here, the 13.75 acre marching camp. This size of camp is below our target range. The camp is also slightly more elongated than the 1:1.5 tertiate proportions mentioned above. It is reasonably safe therefore to attribute this camp to a later period and acknowledge that it will have likely been reused on many occasions.
The short hike across the Cheviots here, coming down next to the mighty Woden Law escarpment is difficult but under the circumstances perhaps insufficiently challenging for a Roman column – now undoubtedly deep in enemy territory - intent on a fast penetration to stop over for the night. Therefore we could be forgiven for overlooking Pennymuir, the mirror image location to Chew Green.
Pennymuir has some impressive upstanding remains, yet on first appearance seems to lack a camp of the size we are looking for.
However an interesting structural relationship can be seen in camps A, B and D.
We suggest that a tentatively reconstructed camp D is earliest- and Agricolan- as it would underlie Dere Street which would have been built soon after his invasion of 79 AD.
Next on site is camp A, and due to its proportions it is fairly likely that it has an Antonine foundation date. In keeping with this timeline it respects Dere Street but overlies a portion of the underlying Agricolan.
This overlap portion is then reinforced in a later reuse of the site giving us camp B. Is this the end of the story for us at Pennymuir?
Perhaps not.
As already mentioned the Ninth may have pressed on without stopping after descending from the Woden Law escarpment, especially if they were marching fast on a lightning campaign. However on the basis of our understanding of the likely structural sequence at Pennymuir outlined above there remains an interesting possibility.
In 117 AD all that would have been on the site was the remains of the large Agricolan camp, with Dere Street rather inconveniently bisecting it. For the sake of discussion however if the Ninth were to have made use of that portion of the camp to the east of the line followed by the later camp A (perhaps even sharing a rampart) then rather interestingly the area this would appear to enclose (the erosion to the south east by the Kale Water is later) would be close to our target area.
So at best all we can say is that the Ninth may have camped here.
Onwards our force proceeds and regardless of whether they stopped at Pennymuir of not the next station along the line is of great interest.
Cappuck near Jedburgh is yet another site with a fine tally of marching camps and has one close to our target area. The small fort here has attested use (deduced from finds) not only of the Flavian and Antonine periods but crucially also from the late Flavian/ Trajanic period – exactly the period we are looking at.
Was this therefore a halt of no greater urgency than the simple expedience of an overnight encampment?
Or had the unrest which drew the Ninth north seen an outpouring of Selgovae violence at Cappuck?
We can not know with certainty but it is a compelling thought as well as reminding us - as indeed our ancient colleagues in the Ninth would have been presciently aware - that they were now operating deep in dangerous country, land where the numerous and highly mobile mounted warriors of the Selgovae - Vinolanda's "Britunculli" - campaigned with confidence.
The marching camp of interest here – “Cappuck 1” on Ulston Moor - is just that tiny bit large at 19 acres but similar in area to the early camp at Chew Green which we have already suggested the Ninth may have used.
The camps proportions are tertiate- and therefore later than the early Chew Green camps original construction.
So at least we may posit that if the Ninth did stop here that they may have been the builders of the camp as opposed to simply making fortuitous re-use of a pre existing camp such as Chew Green. For the purposes of our investigation therefore we give Cappuck 1 a tick and set off again north along Dere Street. At St Boswells a camp of over 12 acres sits by the Tweed, another example of the slightly smaller group of around 13 acres already encountered.
Soon Dere Street passes the major Roman complex at Newstead (Trimontium) near Melrose, lying in the lee of the prominent Eildon Hills. The fort here, sacked some 12 years earlier would remain a focus of Roman attention throughout the busy 2nd C AD, particularly around 140 AD when Pius had it rebuilt, giving it substantial defences. However there are no marching camps of the size we are looking for here, so we must press on.
And at this point we ask ourselves a very relevant question, where would the legion be pressing on to?
The monster 160 plus acre camps of Severus’s later Maetae campaign - the likes of which can be found absolutely nowhere else in the British Isles – certainly proceed north from here towards the Firth of Forth.
Another road however has long been sought hereabouts; the one that passes “from Newstead” westwards along the Tweed and eventually Lyne valleys.
The eastern portions of this road have long been sought around Newstead with no success. That the road did exist is proven by its known western elements and the multi period fort and camp sites along its route. This proves the region constituted an ongoing area of concern and campaigning for Rome.
These valleys also hold remains of considerable Iron Age tribal activity and it would be disingenuous to suggest that with such close Roman supervision being the order of the day over several periods that these settlements and hillforts were not inhabited in the Roman period.
The area therefore was in all likelihood one with a major Selgovae tribal focus and the find spot in this general area of the Hawkshaw head confirms less than cordial relations with Rome here at this very time.
But where exactly did the westbound road go from Newstead? That question has long vexed academics.
Could the marching camps of our suggested route taken by the Ninth legion offer a solution?
We think so.
Newstead controlled Dere Streets crossing of the Tweed and it is possible that the crossing here was by bridge. Following Dere Street north from this crossing (the same north side of the river it should be noted upon which the known portions of the westbound road runs) we soon encounter the marching camp at Kedslie near Earlston.
This camp has no upstanding remains and is known only from aerial photographs. It is however of significant interest.
It sits a mere four miles north of Newstead so perhaps explains why no halt is recorded there. At 20 acres it is again marginally larger than our target size and again, like Cappuck has proportions close to tertiate.
However this camp offers the point at which a plausible junction may have existed between the northbound Dere Street and the mysterious westbound road, and indeed the lie of the land here as well as further west would appear to corroborate the possibility.
Before we follow this line however we must ask ourselves if any camps exist further north along Dere Street which would suggest that the force which marched and camped in these camps did not turn west here into the difficult ground of Selgovae territory?
To the north, close to the Firth of Forth at the Antonine period complex at Inveresk are the remains of a camp of 20 acres. This makes partial re-use of the large Flavian/ Agricolan enclosure there. Could this have been a terminal destination for the troops who used Kedslie?
Possibly, however given Inveresks’ pivotal role in the later Antonine occupation it seems most likely that this enclosure dates to that period. Further, and most tellingly, Inveresk lies squarely within friendly - or at best- less hostile Votadini territory than that occupied by the truculent Selgovae.
If the Votadini were the intended recipients of Roman punitive action then the settlement on Traprain Law near Haddington in the Lothians is unlikely to have enjoyed the unbroken occupation that archaeology suggests it did. This is quite unlike the treatment meted out in Selgovae territory; the incumbents of the oppidium on Eildon Hill north for example appear to have been ejected to make way for a Roman signal station.
No. The Inveresk camp probably records activity later in the Antonine period, and whether its builders came via Kedslie cannot be guessed with certainty though reusing earlier structures would be just as viable a proposition for troops in the middle of the 2nd C as it would for those in 117 AD.
So west we turn towards the known stretches of Roman road further along the Tweed valley, following easy high ground (usually much favoured by Romans when setting out their roads) via Langshaw, Colmslie and Buckholm to the old crossing of the Gala Water at Torwoodlee above Galashiels.
By now we can safely assume that the Ninth, now deep in Selgovae heartlands, were busily meting out devastation upon the unlucky locals, an approach clearly repeated by invaders from the south in a later era given the number of ruined tower houses hereabouts.
Once across the Gala Water - following the natural corridor in the lie of the land - it is a fairly straight line via Caddon Mill at Clovenfords (the name ford is an apt pointer when looking for old roads), down again to the northern bank of the Tweed opposite Peel, possibly following a broadly similar line to the modern road here.
Whether the Roman road followed the Tweed close to its north bank (as does the modern highway) or further uphill is uncertain. Difficult terrain such as the constricted ground where high ground comes down close to the rivers northern bank at places like Thornielee and Pyat Hill would tend to suggest the former.
At Innerleithen we pass the marching camp situated here. Only part of this camp is currently known though the positions and lengths of these ramparts and gate suggest that this camp is probably larger than our target area.
It belongs then to one of the many other Roman forays into deepest Selgovae territory. The camps position does however confirm that we are indeed on the line of the Roman road, especially as within a few miles further we come to a site of the greatest interest in our quest; that at Eshiels near Peebles.
Eshiels, lying on the north of the Tweed has two camps. The earlier is an almost square camp complete with clavicular gates and measures over 27 acres in size. We can reasonably date this camp therefore to the Flavian period, most probably to Agricola’s campaigning in central southern Scotland against the Selgovae in 80 AD.
Inside this camp is another. It reuses the earlier camps western side and at 16.3 acres – capable of accommodating around 3,600 men – it sits well in our target range. The period we are looking at remains sufficiently close to the Flavian period for camp defences of that period to remain upstanding, and as we have already pressed the Flavian enclosures at Red House and Chew Green into service by our suggested route taken by the Ninth legion then reuse of parts of the defences of the earlier camp at Eshiels is just as possible.
This camp measures some 280 by 236 metres, giving a proportion of 1:1.18. This is more elongated than square yet less than tertiate proportions so fits into the constraints of camp morphology at the period we are concerned with comfortably.
And tellingly, at that the trail goes cold.
Continue to Pt.3
Back to Pt.1
©2009 Roman Scotland. All Rights Reserved
First Published November 2009
