Download file in PDF format: TRAC 1994: Reading the Tea Leaves, Signalling as a Means of Prophesy on Roman Frontiers (pp. 90–98)
Reading the Tea Leaves, signalling as a means of prophesy on Roman frontiers
by D. J. Woolliscroft
Some years ago, the writer put forward a paper offering a possible explanation of the signalling
system of Hadrian’s Wall (Woolliscroft 1989: 5f). Before that time, it had always been assumed
that signalling on Roman frontiers would have operated in a linear manner, with signals from
minor installations, such as Wall turrets and milecastles, being relayed from site to site along
each link of the frontier chain until they reached a fort. The reality may have been rather more
efficient, however, at least in the Wall’s earliest phase when the main garrison forts still lay a
little to the south on the Stanegate road. For, with one possible exception, every single one of
the 82 Wall installations studied had a clear view to a Stanegate site, making direct
communications possible using Roman visual signalling techniques, whilst the Stanegate itself
could be linked together via an inter-fort chain. Moreover, this was not simply coincidence, for
quite a number of these links had only been made possible by allowing distinct irregularities in
the milecastle and turret spacing system, which suggests that signalling had been a definite
Since that paper was published in 1989, a few doubts have been shed on specific details,
notably the exact history of T’ 45a and MC 39 (Crow 1991: 62f), but the theory does seem
largely to have stood the test of time and similar direct signalling systems have now been
recorded during surveys on the Cumberland Coast defences, the Gask Ridge in Scotland and
the Wetterau Limes in Germany. But, in any discipline with scientific pretensions, there is one
sure way of putting a theory to the test and that is to use it to make predictions whose validity
can then be tested. For it is simply the case that if any academic theory is correct then other
things should follow that can be tested experimentally.
In the case of Roman frontier systems such experiments involve looking for gaps or other
oddities in a system and using them to predict the existence and location of missing sites or
other factors which can be looked for archaeologically and such prophesies have, so far, turned
out to be both possible and surprisingly accurate. In Britain, for example, a fairly rich harvest
of new military sites is gradually coming to light, including a second tower on Barcombe Hill
near Vindolanda, one and possibly two new fortlets on the Gask Ridge and a possible tower,
now under excavation, on the Antonine Wall, all of which would appear to support the
conclusions reached on Hadrian’ s Wall, but as most of this work is now either published or
about to be published in this country (Woolliscroft 1988: 23ff; 1994a; 1994b; 1994c;
Woolliscroft, Swain and Lockett 1992: 57ft), the remainder of this paper will concentrate on
the results from the German Limes (Woolliscroft & Hoffmann 1991: 531ft).
Here, a sample study length was surveyed, consisting of the 65km sector from WP 4/47 in
the northern Wetterau, to Grof3 Krotzenburg on the Main (Fig 10.1), which is the only Reading the Tea Leaves, signalling as a means of prophesy on Roman frontiers 91
SIGNAL TOWER !
ROMAN ROAD -EI7
Fig. J 0.1. Wetterau Limes 92 D. J. Woolliscroft
continuous stretch long enough to act as a significant (c. 14%) sample of the whole and largely
free of the modern forestry which elsewhere makes studying intervisibilities almost impossible.
It takes in 96 separate installations, including 14 fortlets and eight forts.
At first sight this study sector appears homogenous. The terrain is similar throughout. The
frontier installations are all sited a little behind the line, with the fOlts at reasonably regular 6-7
km intervals and the watch towers are sited so as to be intervisible with their neighbours rather
than to any regular spacing system. A closer inspection reveals a rather different picture,
however, in which the sector breaks quite neatly into two, around WP 4/94b, to produce
northern (Arnsburg – Ober Florstadt) and southern (A1tenstadt – Grof3 Krotzenburg) zones,
of 36 and 29 km respectively, each of which contains four of the eight forts, but which
otherwise show markedly different characteristics. Most noticeably:
I. Although the southern forts lie only metres behind the Limes, those of the northern zone
stand much further back. For example, Arnsburg and Echzell, are about J’I2 km behind the
frontier and Ober Florstadt is 21/2 km from the line. Furthermore, in the North the forts are
often built on prominent ground whilst, in the south, they tend to be built down in the river
2. The entire 29 km run of the southern zone is formed by just two almost perfectly straight
sections, hinging on the fort of Markobel. The more northerly of the two then continues
slightly into the northern zone, but thereafter the line makes frequent twists and turns right
around the northern Wetterau. Of course, the Limes is, here, turning anyway, to form the
Wetterau loop, so that changes of direction are inevitable. But this overall 180
extends over almost 20 km and is barely discernible on the ground, yet within it the Limes
zigzags constantly, often against the general trend of the loop.
3. fn the south, the Limes takes no account of the terrain through which it passes. The rolling
countryside offers considerable tactical opportunities, which could have been exploited
easily by aIJowing small deviations in the line, yet it remains remorselessly straight.
Consequently, it often faces steeply rising ground immediately outside the line and many of
the towers have little or no view forward. In the northern sector, on the other hand, the
priority seems to have been to gain the maximum possible tactical advantage from the
terrain, and the Limes here runs from high point to high point as a much stronger line.
Most thinking about signalling on the German Limes has hitherto been based on the fact
that its towers and fortlets are intervisibJe with their neighbours, and again, the assumption had
grown up that signalling was carried out laterally, tower to tower, along the frontier line. But
such a system would have been both inefficient and poor military practice Why?
To be effective, any military communications system must fulfil certain basic criteria.
Firstly, of course, it must work and, secondly, it must be as efficient as possible, which means
that it must transmit as lI1uch information as possible, as quickly and reliably as possible, whilst
using as few resources as possible It must also strive to be as invulnerable as possible to enemy
attack, which means that it must contai n redundancies and fail-safe mechanisms to allow the
system as a whole to survive the destruction of at least some of its component parts Reading Ihe Tea Leaves, signalling as a means ojprophe”y 011 Romanfronliers 93
ARNSBURG -OBER FLORSTADT
LOCAL SIGNAL LINK
SIGNAL LINK VIA RELAY
INTER FORT SIGNAL LINK
…. _ …. –
TRUE SITE POSITION
Fig. 10.2. “Vellerou Limes. Signalling. Arnslnl/g-Ober- Flors/adl 94 D. J l1i
We can apply these criteria to the two basic choices of Roman signalling system available:
the traditionally envisaged lateral system and the direct system found on Hadrian’s Wall and
elsewhere. Obviously both systems can be made to work, but the direct system is more efficient.
It is faster, since most signals do not need relaying and, even if we accept that Roman long
range signalling could transmit little more than emergency beacons, it also conveys more
information because, as the signals are direct, the recipients can see where they came from and
thus where the emergency was. It is also very much less vulnerable to attack.
Because the lateral system passes signals via every tower between their source and
destination, an enemy has only to neutralise one tower for the system to fail. But, on a direct
system, the destruction of anyone minor installation has absolutely no affect on the others.
Weaknesses might continue in inter-fort communications, which do usually have to be carried
via relays, but even here the direct system is superior because its inter fort communications rely
on specific installations, which can be protected. The lateral system can again be disabled by
the neutralisation of any site and so one would expect any sensible military organisation to
We can now turn to the actual position on the Limes, and again the study sector breaks
neatly into the same two zones, with very different signalling arrangements in each. The
northern zone (Fig 10.2) is almost identical to the Stanegate phase on Hadrian’ s Wall, for
almost every installation enjoys direct intervisibility with a fort and of the handful that don’ t,
all have a choice of at least two different one stage relays, so that none are dependant on any
one other site for their communications (1) The northern zone also possesses an efficient interfort system, for although none of the forts are actually intervisible, each can communicate with
its neighbour using just one relay and in each case the relay site is a fort let, a rather more
hardened installation than a mere tower.
The southern system, however, veers much more towards a lateral arrangement (Fig 10.3).
Here 35% of the installations need relays before their signals can reach a fort and some require
double relays via two other sites. Furthermore, plentiful opportunities to improve the position
were ignored For example, had Markbbel been just a few hundred meters further back from the
line it would have been able to see five currently blind sites and yet it was built right up against
the Limes. Worse still is the inter-fort system. Again none of the forts are intervisible, but here
only one of the inter-fort links was accomplished by a single relay and, of the three remaining
links, two required double relays, whilst that between Ri.ickingen and GroB Krotzenburg needed
no less than foue the neutralisation of any of which would have cut the strategic chain.
We are, therefore, confronted ,””ith a problem. Throughout its length, the study sector is
made up of the same basic building blocks and runs through similar terrain. Yet, it breaks
down into two distinct zones which, although contemporary, have different layouts and
operating systems, one of which appears clumsy and vulnerable. Such a situation requires
For example, if the southern system had operated over the whole sector, we could have
argued that, as this is a very early system, Roman frontier design was still in a primitive state
which had advanced by the time of Hadrian’s Wall and most of the other British systems. But it
does not. Nor can we argue that we are seeing a process of learning, with the builders starting
from GroJ3 Krotzenburg using one design, then realising that what they were building was
flawed and improving that design, for we can find the same two patterns repeated elsewhere on
the German frontier, not just on the Domitianic Rhine-Main and Odenwald lines and not just in
Germania Superior, but even on the much later, Antonine, Outer and Raetian Limes. Reading the Tea Leaves. signalling as a means o(prophesy on Romanfrontiers 95
ALTENSTADT – GROrl KROTZENBURG
LOCAL SIGNAL LINK
SIGNAL LINK VIA RELAY
INTER FORT SIGNAL LINK
– – – – TRUE SITE POSITION
0 2 4
Fig J 0. 3. IFelterau Urnes. Signalling: A Itenstdl-Groft Krotzenburg 96 D. J lVoolliscroft
Another approach might be to suggest that the two zones were built by different legions,
But again, the two patterns survive into the Antonine parts of the line and although it is true
that there were still legions in the area that had been there in Domitianic times, it seems
unlikely that a unit would have been allowed to perpetuate obviously inefficient methods for so
long. Presumably, then, whatever forces caused the emergence of two different frontier designs
on the Flavian, Wetterau Limes were still operating in the Antonine period, nearly a human life
One example of such a constant might be the strategic position and it is certainly true that
it has always been easier to approach the Wetterau from the north than from the east. To cite
this as a reason for the differences within the study sector, however, might be to misunderstand
the role of Limes systems, which are probably more bureaucratic than strategic in nature, and
there is no evidence, for example, from Hadrian’s Wall, that other Roman frontiers relaxed the
quality of their signalling systems in strategically less sensitive areas. Besides, there are still
some routes into the eastern Wetterau, notably the Main valley itself and the smaller Kinzig
valley, which now carries autobahn 66 and a modern railway right past the fort of Riickingen.
Neither of these produces local strengthening of the system and nor do there appear to have
been any specific local threats to the northern zone, for although there are some large hill forts
in the vicinity, they do not now appear to have been occupied in the Roman period (Herrmann,
1985 Mildenberger 1977/8: 157ft).
There is, however, another possibility. So far we have simply dismissed the southern
system as inferior, and by the criteria by which we have been judging it, so it was. Yet the
Romans built it, and continued to build sectors like it. If we assume that they were not fools and
would not deliberately perpetuate an inferior system, we must conclude that, by whatever
criteria they were employing, the system was not inferior, or at least, since both systems
continued to be used, that under certain circumstances it was not inferior. Theoretically, the
direct system is so superior that it is hard to think of circumstances in which a lateral system
would be built in preference to it, but need the fact that something was built necessarily imply
that it was preferred? A more likely scenario is that under certain conditions the lateral system
was all that was possible and once this point is reached an explanation suggests itself
immediately, especially in a German context.
The direct system does, in fact, have one potential weakness, in that its successful
operation requires a condition which the lateral system does not. As such a system functions by
passing visual signals back from, as well as along, the frontier, it can only operate in open
country. Yet southern Germany is, and always has been, a land of forests. The Romans may,
therefore, have been faced with a choice between clearing trees, a massive undertaking even if
only the lines of sight were cleared, and modifying the system. They would appear to have
opted for the latter, because the great advantage of a lateral system is that, given one proviso, it
can operate in forest.
Unlike the direct system, the lateral system operates in one dimension only, because all
signalling takes place along the line of the frontier. It has always been assumed that a corridor
around the Limes would have been kept clear of trees, to deny cover to an enemy and enhance
the views of the watch towers. This means that, so long as the frontier and its installations were
confined to a narrow and absolutely straight line, the presence of forest is irrelevant because the
towers could see down this linear clearing and pass their signals freely along it. This is exactly
the configuration of the southern part of the study sector and it is noteworthy that the one real
bend in the zone hinges on the fort of Markobel, which would explain why it was built so close Reading the Tea Leaves. signalling as a means oj prophesy on Roman frontiers 97
to the Limes, where it would have had a view along both legs of the line, and not in the position
further back mentioned earlier. The obvious conclusion then is that the southern part of the
study sector was forested in Roman times whilst the northern part was not.
This obviously has potential ramifications well beyond the subject of frontier studies,
because if we were to succeed in using signalling to determine the Limes’ environmental
context in this area, we could do the same for any part of the line. But, for the moment, such
potential must be approached with caution since not enough environmental archaeology has
been done in the area to either confirm or confound the theory, although a major survey of the
environment of the ancient Wetterau is now under way. But we have seen that the two basic
layout types recur throughout the line and limited environmental work elsewhere has produced
the expected results (Firbas 1930 75ft Knapp 1973 115ft Knerzer 1973: 71ft Rosch 1988:
114ft Streckhan 1958 61ft)
There now remains one more problem with which signalling may be able to help: the
history of the Limes’ forts. These are still controversial, but it does seem certain that the full
unit forts were not part of the original plan. They are almost all later, some of them
considerably later. As a result, it has sometimes been assumed that the Limes initially stood
alone, simply as a chain of watch towers, with no real backing by forces in its immediate
vicinity. This would, however, have been such appalling military practice as to be incredible.
It is assumed that Limes systems provided an observation screen and monitored and
controlled movements across the frontier, but there is obviously little point in observers unless
they can tell someone what they see. Likewise, a system lacking the man power of a fort chain
may well be able to monitor frontier movements. but there is a limit to what it can do to control
them. This means that from the outset the frontier needed forts, for which the minor
installations were merely eyes and ears. A system without forts would simply have provided
hostages to fortune, for to leave small groups of men, unsupported, in isolated positions, in
potentially hostile country is positively to invite their destruction.
In fact, there has long been a certain, limited, amount of evidence that the original plan
did, indeed, involve forts, albeit smaller in scale than the later full size auxiliary bases, because
excavations at a number of the principle Limes forts such as Saalburg and Altenstadt have
revealed rather smaller, turf and timber forts underlying them, which seem to date from the
earliest days of the system. But until recently few of these structures had come to light
However, during the signal survey it became very apparent that, in the northern sector, the
Limes often followed the exact horizon of the forts and, moreover, this appeared to have been
deliberate policy as the line had been weakened in a number of places and various tactical
oddities had been caused as a result. It seems most unlikely that such exactness could have been
achieved unless the forts had already been there in some form, or at least planned, when the
line itself was laid out, and the only conceivable reason for allowing such weakening would be
to ensure signalling links with the forts.
At the time tile Limes survey was being written up, air photographs became available that
confirmed the existence of earlier forts at Ober-Florstadt and Inheiden (2), whilst excavations
suggested that Echzell (Baatz 1965 139ff) may also have had such an ancestor (3) and so my
1991 published report (Woolliscroft & Hoffmann, 1991 543) took the risk of saying that
Arnsburg, the remaining fort in the northern zone, must also have had such a precursor, either
underlying the known fort or very near by. Fortunately, just a few months ago an air
photograph of the site was published (Hessisches MinisteriuJ11 Fur Wissenschaft und Kunst,
1993, 5+ Abb5), which appears to show a double ditch, obviously underlying stone work from 98 D. J. Woolliscroft
the known fort and although this has not yet been tested by excavation, it could be that the last
of these prophesied early sites has now been found.
1. The signalling maps have been slightly schematisised in the interests of clarity by artificially
displacing some of the lines of sight from their true termini. This merely prevents lines from
crossing or running on top of one another.
2. My thanks to Dr P.IlIe for allowing me access to and discussing with me air photographs
from the Landesamt fur Denkmalpflege Hessen archive.
Baatz, D. 1965. Limeskastell Echzell. Kurzbericht tiber die GrablU1g 1963 lU1d 1964, SJ 22: pJ39ff.
Crow,1.G. 1991. A Review of CUlTent Research on the Turrets and Curtain of Hadrian’s Wall, Britannia
Firbas, F 1930. Eine Flora aus dem Bnumenschlanun des Romerkastells Zugmantel, S.J. 7: p75ff
HerrmafUl, F.R. 1985. Der Glaub erg am Ostrand der Wellerau. Archiiologische Denkmiiler in Hessen 5 L
He1Tmann, F.R. 1986. Der Diinsberg bei Giej3en. Archiiologische Denkmiiler in Hessen 60, Wiesbaden.
Hessisches Ministerium Fiir Wissenschaft tmd KlU1St. 1993. Zeitspuren, Luftbildarchiiologie in !-lessen.
Knapp, R .. 1973. Die Vegetation der UmgeblU1g von Butzbach in der Gegenwart tmd zur Romerzeit, 5.J.
Knorzer, K.H 1973. Romerzeitliche Pflanzenreste alls einem Brunnen in Butzbach. Hessen, S.J. 33:
Mildenberger, G. 1977/8. Die germanische Besiedltmg des Diinsberges, PH. 17/18: pI 57ff
Rosch, M. 1988. Botanische FlU1de aus romischen BrllfUlen in MlIrrhardt Rems-Murr-Kreis, A.A.B. w.:
Streckhan, HU. 1958. Vegetationsgeschichtliche Untersuchtmg einer romerzeitlichen Torfbildtmg am
Schrenzer bei Butzbach in Hessen. SJ 17: p61 if.
Woolliscroft. D.1. 1988. The Outpost System of Hadrian’s Wall. Transactions of the Cumberland and
Westmorland Archaeological Suciety (second series) 88, p23ff
Woolliscroft. D.l. 1989. Signalling and the Design of Hadrian’s Wall. Archaeologia l1.eliana (fifth series)
Woolliscroft. D.l . 1994a. Signalling and the Design of the Cumberland Coast System, Transactions of
the Cumberland and Westmorland Archaeological Society (second series) 94.
Woolliscroft. D.1. 1994b. Signalling and the Design of the Gask Ridge System. Proceedings of the
SOCiety 0/ Antiquarians of Scotland.
Woo II iscroft, 0.1. 1994c. Garnhall. an Interim Report of the 1991-3 Seasons. Manchester
ilrchaeological Bulletin. forthcoming.
Woolliscroft. DJ. & HoffinafUl, B. 1991. Zum Signalsystem tmd Aufbau Des Wetterau-Limes,
Fundberichte Aus Baden-Wiirttemberg 16: p53 Iff
Woolliscroft. D.1. . Swain, S.A.M. & Lockett. N.] 1992. Barcombe. B. A. Second Roman ‘Signal’ Tower
on Bat’combe Hill, Archaeologia Aeliana (fifth series) 20: p57ff