05 June, 2017

Hadrian’s Wall; milecastles, turrets and artillery

Problems; what problems? 
To the casual observer, what Hadrian’s Wall was for and how it worked is fairly straight forward, but dig a little deeper and all is not as it seems. 
If you accept the Wall as an exercise in military architecture and that warfare is ultimately about numbers, then the figures just don’t add up quite as you might expect.
 The “problem” is as much perceptual as academic, since it has long been patently obvious to serious study that the Wall “garrison” as represented by the milecastles and even the forts would be hard pressed to defend the wall in a conventional sense.  It would be fair to say that the Wall has been seen as much a base for counter-attacking the enemy as a barrier to be defended; with good early warning from outlying forts it was envisaged that decisive actions would have been to the north of the Wall.[1]
Thus, for reasons which I will discuss further below the way we view the Wall as physical barrier should not be taken for granted; I use the word “view” advisedly, since so little of the Wall survives, the way we “picture” the Wall, both objectively in its visual culture and subjectively in our own mind represents a “understanding” which profoundly influences our other forms of thinking.
My interest started with the proposition that the spacing of the milecastles and turrets at c. 540 yards [2] might be related to use artillery [balistas/ scorpio], which lead me to build a CAD model a model of milecastle, turret and connecting Wall.  I am now happy about the potential use of torsion bows, but other more widely accepted ideas expressed and illustrated in our visual culture of the Wall, do not make sense in terms of military architecture or strategy.
It is not a new problem, because it is the central issue of Hadrian’s Wall; put simply how was the Wall supposed to work in terms of military architecture?
The Plan
Leaving aside the subsequent changes, the initial plan for the “Broad Wall” was for curtain wall with gates every mile protected by a small fort or “milecastle”, with a garrison of about 40 -50, separated by two intervening small turrets.  By any standard this arrangement is lightly manned, and the classic model presumed that if attacked these garrisons would be reinforced from forts on or behind the wall [1], counter attacking via the many gates to trap the attacker against the Wall. The outlying forts on the major routes from the north would give warning to deploy troops to appropriate forward positions, particularly the main fighting legionary units of the army based to the south at York and Chester.
Thus, the Wall has been viewed more as base with the initial decision to punctuate the Wall with Gates every mile allowing complete freedom of movement into the areas north of the Wall.   The Wall provided a physical barrier that cannot readily be pulled down or burnt which would prevent horses, carts and to a lesser extent hostile troops crossing into the south; it was an expensive, but permanent solution. The only practical way through was to capture one of the gates controlled by a milecastle.
It is simply a question of force ratios; in convention military thinking the attacker needs a certain superiority of numbers in order to come a defended position, dependent on the nature of the position and the forces. 
Typically, attacker might require an advantage of 3:1 over the defender.
In this period Roman adversaries would deploy forces measures in tens of thousands, as Mons Graupius in 87 where Agricola faced 30,000 Caledonians.  While Roman forces fighting from defended positions were highly effective, garrisons of 40 or even 80 in the milecastle would be hard pressed if surprised by a force of thousands.   The chances of holding any mile of Wall even with reinforcements is hopelessly optimistic, given that anyone with a ladder could climb the Wall and envelope a milecastle.
The critical technology – the gates and associated tower, is well known and proven in Roman Fort which typically had four entrances, very different from Hillforts or castles.   Fortifications are entered and usually attacked via their gate, but the Fort by offering four possible places to attack, also forces the attacker to divide their forces to cover all the gates or risk being counter attacked by the garrison.
However, forts are a very different context; a typical legionary fortress like Caerleon has 4 gates and 30 turrets in curtain Wall perimeter of slightly over a mile in length; defended by a garrison of 4000 – 5000 of the best infantry.  Caesar describes how a Legion commanded by Cicero’s younger brother successfully held their fort against the Eburones uprising in the winter of 54–53 BC [3], [while a force of 15 cohorts who abandoned their camp were ambushed and massacred].
This is very different to the Wall, where Britain’s entire 3 legionary garrisons is only roughly equivalent to 180 men per mile or one every 30’; the curtain Wall is only defendable if you can get sufficient forces there fast enough.  Ultimately, it is the 10’ wide gaps in the Wall that are the key links in the chain, and the time taken by the enemy to capture a milecastle becomes the only important parameter.
Turrets as Artillery Platforms
The Roman’s use of torsion artillery was a unique capability in this theatre, which, given the small scale of garrisons, represents only realistic available methods of interdiction available from the milecastles or turrets. Thus, their spacing at roughly 540 yards could be related to the range of certain types of weapon.
Julius Caesar tells us of an incident where to prevent his troops being out flanked and enveloped by a larger force, Caesar creates, at right angles to his front, physical barriers covered by his artillery dug in at either end.
 “….on either side of that hill he drew a cross trench of about four hundred paces, and at the extremities of that trench built forts, and placed there his military engines…” [Caes. Gal.2.8]
What is significant here is distance between his artillery units, suggesting that some of his pieces had a range of 400 yards, which would allow mutual covering fire.
There is view that some engines had a range of up to c. 400 - 500 yards, which is supported by archaeological evidence from the Jewish wars. [4]  Excavation at Gamla besieged in 67AD  found that stone Balista shot of 2 -6 kg / 10 -16 cm was fired from positions 330 yards [300m] from town, shot being found up to 65 yards [60m] inside the defences.

The spacing of the turrets and milecastles at 540 yards is realistically too great provide mutual covering fire, however, while crossing the Wall in the middle between the turrets might seem the obvious choice for the attacker, it is the place is exposed to maximum crossfire.
Thus, in terms of artillery, the turrets were placed just far enough apart to be effective; by contrast sections of Caesar’s rampart at Alisia in 52bc, like the fort at Caerleon, had Turrets spaced every 27 yards, which is probably reflects the effective range of pilum in this context.

While artillery platforms are known from the area, [5], it has to be said that I am not aware of significant finds of bolts, stones, or other ammunition associated with the turrets or milecastles.   The argument might seem better if the towers were closer together, but it’s in keeping with general minimalistic approach to the layout of Wall.   

A Milecastle Model #1
Theoretical Structural Archaeology was developed for the complex problems presented by the foundations of timber buildings, but is equally applicable to the less ambiguous remains of stone walled structures. It attempts to model structures directly from the evidence of foundations,  independently of the visual culture or textual conventions of the period.  The engineering of buildings and other structures is governed by general rules, as well as the context specific conventions such as those found in military architecture.  In addressing the spatial evidence of the archaeological record it produces different results from research that simply considers the conventions of textual summery found in the Wall’s literary construct. 
Theoretical modelling, while not simulating the process of building, forces the modeller to take design decisions and follow through the consequences.  While it is possible to build a structure from individual randomised blocks, there is a law of diminishing returns, and any exercise in modelling is usually directed towards testing a specific idea or resolving a specific problem.
In this case, over and above an interest in the use of artillery, the model wished to consider general elevations of structures on Hadrian’s Wall in relation to how they were roofed.
The model is based on the plan of the Broad Wall Milecastle 48 at Poltross Burn and Turret 48A [6], with the very important proviso that the site is treated as flat rather than steeply sloping to the North as is the case at mc48.
The connecting Hadrian’s Wall and the curtain Wall are modelled in general terms without detailed consideration of the battlements and parapet, which although visually important, are less relevant to the model, representing optional detail.
Models and the diagrams they generate may be used to demonstrate a range of ideas that are contradictory or false, for this reason  Steve Oles from SketchUp, the software I used for this exercise, is included to reinforce the point that this is a form of scaled diagram used to illustrate an argument, and not an artistic representation.
Roofing is an important design consideration in structures with wooden floors traditionally a feature of buildings, although this not always reflected in the visual culture of the Wall.
Very roughly the structure occupies about 650 m² of which about 40% is the curtain wall and towers all of which is hard surface that has to be drained.
The model wished to test, rather than a pitched roof design, the use of a simple Mono-pitched roof for the internal buildings which can be extended to articulate with the curtain wall, which in principle has a number of advantages.
  • It would drain water away from the narrow spaces between the building and the wall, where there is no evidence of a drain.
  • In an Ideal arrangement the runoff from the majority of the structure is directed towards the drains of central roadway and out through the gate.
  • Protects the steps which are built around an earth core with a retaining wall which is best kept dry for stability; 
  • Facilitates a NW Kitchen based around the oven, [baking is traditionally an indoor activity]; 
  • Covering the SW and SE corners provides a latrine and perhaps some form of loose box; 
  • Provides a complete drainage solution for the whole structure.
The key parameter is the roof pitch; 30.5° / 7:12 was chosen as the lowest practical pitch for a pitched roof with wooden shingles which could be replaced by tiles; 8:12 or 9:12 would also be considered practical.
 The remains of steps comprising four partially surviving 11” x 8” steps and a 15’ 4” retaining Wall which argues for 12’ rise on the steps; the logic of the model would be to match this with roof pitch which might suggest 8:12.  In general terms, the steeper the roof, the taller the curtain wall in this arrangement, and vice versa.  

The basic building design has 4 separate rooms, which are shown with a central hearth, and although this is not certain for all rooms, it does imply a single storey infantry barracks.  The width of the doors also indicates there is absolutely no provision for cavalry, apart for the potential of space for loose box against the southern wall. 
There is no apparent differentiation in the design to indicate the presence of an officer, and no obvious implication to be drawn from there being 8 rooms; the century, basic component we are familiar with, is normally comprised of 10 units of 8 men. The rooms are smaller than those in typical infantry barracks; however, the latter represent the permanent residence of a unit, which may not be the case for milecastles. The precise garrison, usually put around 40-50, was presumably deemed sufficient to man the milecastle and the turrets on either side; defending the intervening stretches of Wall being impractical.
The gatehouses are shown with windows to light the internal spaces and provide additional firing positions; these a nominally based on utility and the presumed position of stairs, and doors for which there are fewer options. The roof shape chosen is a N/S oriented pitched roof, as opposed to an E/W orientation or pyramid form, the other simple contenders; while this arrangement was considered in terms of practicality, it also creates an appropriate architectural effect at the front of the structure.   
The Turret
As with the Gate house the point was to demonstrate the need for a roof given that structure would have had wooden floors and ladders or steps.
In addition to consideration of drainage, headroom required for weapons and protection from missiles is important.
In this case, a pitched or pyramid roof are the simplest options, combining the two can create more complex forms; in any event, a roof is required with a standardised form, reflecting the generally uniform nature of the structure.  Apart from the addition of a roof and artillery, this is general how these structures have been pictured.
There role has been seen as compatible with the similarly scaled small lookout and signally  timber towers evident in archaeology and art. [23]
One advantage of wooden towers is that achieving a 50’ high platform is relatively simple with the appropriate timber, which is significantly taller than the elevations envisaged for the turrets on Hadrian’s Wall.
Thus, the use of wooden super structure in addition to a basic roof, floors and fittings cannot be precluded to enhance the height, and hence effectiveness, of the turret.  However, the most significant issue encountered modelling a turret was it relationship the intervening Hadrian’s Wall.
One possibility is that a walkway was accessed from a door in the Turret, Logically, the turret would have had three doors, potentially making it more difficult to defend.
The alternative is that there is no direct access, which raises questions about the form of Wall and its relationship not only to Turrets, but also to milecastles.

Not unusually, having built the model, I am unhappy with various aspects; which is precisely why models are built. 
The initial idea about that the spacing of gate towers and turrets might reflect the use of torsion bow artillery, works reasonably well, although, perhaps in keeping with the general character of the linear frontier, it would represent a minimal use of resources.The effectiveness of artillery and observation would be enhanced by height, taller the better, as is the case with most fortifications.
The issue of the relative heights of the Wall and milecastle curtain  is the principle issue highlighted, but not resolved by the model. An "Ideal" section, illustrated above, might resolve this issue and perfect the drainage.
To get this to work, the height of the wall must relate to angle of the roof and its starting height; an additional complication is the steps, which I have had to recess into the wall it give some extra height, or face issues with headroom at the top.
As this is a military structure top roof should not be higher than the wall, and it would be ideal from a drainage point of view if the wall was just higher than the roof, so that the rapid runoff from the hard surfaces of the walkway had somewhere to go. The estimated height of the wall using the 4 surviving steps puts the walkway at 12’,[6], the model is about 15’.
What is concerning about the steps is that they are not bonded with the curtain wall, while this simplifies construction, at some point it must integrate with the bonded stonework, and for the curtain to have greater height a more complex design has to be envisaged.

However, the height of the wall is problematic in other respects, in particular in how it related to Hadrian’s Wall itself. If Hadrian’s Wall and the curtain wall of milecastle are the same height, then it possible to enter via the Wall, which given how easy it would be to climb, is a very insecure arrangement that does not altogether make sense.  
For Milecastles to function as a fortification with a secure perimeter, the curtain wall has to be higher than the Wall itself, which as we have already suggested would benefit the model for roofing the buildings, if mono-pitch rather than standard pitched roof design was used.  
However you look at it, there has to be some reasonable differentiation, some method of isolating the two aspects of the frontier; since defending one mile of Wall is impractical, it makes sense to raise the milecastle / lower the Wall.

Here again we return to the essential question of how the wall was supposed to work, and whether it can be assumed to have had a walkway with battlements. If so, how was this accessed from the ground, or from milecastles and turrets?
This question is complicated by the differing phases and standards of construction evident in what remains of the Wall, where distinguishing between normal and exceptional is not always easy.

The normal arrangement seen in forts like Chesters and Rudchester is that the Wall intersects at a gate tower. At Houseteads [above] and Greatchesters, where the Fort is inserted, the junction with Wall is marked by Turrets, which in sense is ambiguous, since we have to assume that these towers are higher than the curtain wall of the fort.  In these latter forts, and Birdoswald where both cases apply, the intersection of the Wall with corner turrets at an angle makes any provision for access unlikely.   
So while the arrangement in forts makes it clear that they a physically separate from the Wall, it does not shed any light on the same relationship for milecastles as the presence of turrets obscures any inference about the relative height of the curtain wall.   
Both milecastles and turrets in their Broad Wall form were built with similar 12’ “wing walls”, a stub of wall that facilitated a junction with the linear sections of walling built by different gangs that would join them up at a later date.
To some extent this restricts the likely height of Hadrian's Wall, since this is necessary to prevent a butt joint, an offset between the successive courses of the facing [like a flight of stairs]. 
For the sake of argument; if including mortar we a looking at standard 8” course of 12” blocks [7], then a 12’ wing wall is equivalent to 24 x regular 6” offsets, which at 8” per course, suggests a height of 16’; shortening the offset / block length would increase the estimated height. It is clear that precise height of any of the elements is difficult to calculate from the existing evidence, and that the separation evident in the forts sheds no direct light on the relationship between milecastles, turrets, and the Wall.
Given the crucial role of milecastle in maintaining the integrity of the frontier, I would be inclined to the view that that they must be physically separated from the Wall in order to maintain its perimeter.  This can be achieved by a height differential, or, perhaps, by building the Wall with a top that slopes on the inside and has no walkway.   
Thus, building a model to understand roofing, one of the main approaches to Theoretical Structural Archaeology, has taken an unexpected turn, which in this case, results in a reconsideration of underlying assumptions or presumptions, which is the point of deductive reasoning, and one of the advantage of evidence based model building.
Whether, the visual culture, and the expectation it has generated, could ever conceptualise a more simplistic and less elaborate form of Wall is an interesting question, but what is more important is to be able to distinguish between what derives from artistic or textual convention, and what is rooted in archaeological evidence.

Sources and Further Reading

[1] Handbook to The Roman Wall. J. Collingwood Bruce. Published by Hindson & Andrew Reid Ltd, Newcastle upon Tyne (1966) p.26
[2] op. cit. p. 22
[3] Caius Julius Caesar "De Bello Gallico" and Other Commentaries English translation by W. A. MacDevitt, introduction by Thomas De Quincey (1915) http://www.gutenberg.org/etext/10657
[4] Aviam, M., 2007, The Archaeological illumination of Josephus; in Making History: Josephus And Historical Method, edited by Zuleika Rodgers,  pp. 354 -355,  361-2, 381.
[5] eg.  High Rochester; https://en.wikipedia.org/wiki/Bremenium
[6] Gibson, J.P. & Simpson, F.G. 1911. "The Milecastle on the Wall of Hadrian at the Poltross Burn"; Trans. CWAAS XI (New Series) Art XXIII pp390–461
[7] Handbook to The Roman Wall. J. Collingwood Bruce. Published by Hindson & Andrew Reid Ltd, Newcastle upon Tyne (1966) p.33


Odin's Raven said...

Surely, even if the fields of fire overlapped, the volume of fire would not have been sufficient to defend the curtain wall. If the Romans expected to fight to the north, perhaps the Wall was a fallback position which would have been difficult to outflank, and the Romans need just have manned the portion to which they had retreated, without normally manning the whole length of the Wall.

Geoff Carter said...

Thanks for your comment, you make a good point, the Wall with or without Artillery is only a deterrent which should prevents the Romans being outflanked; essentially, it prevents undetected movement.
It suggests that the Romans were confident of their control of the territory to the North and would have time to manoeuvre.
The artillery allows them to interdict or control the Wall without leaving the safety of the Turrets or Milecastles.
The important point is not getting over the Wall - but getting your horse over.
So while the Wall was lightly manned - I think it was still continuously manned for observation and deterrent reasons.