06 · The Shadow War the Empire Forgot

How the frontier actually worked

Picture a band of country two hundred miles deep and a thousand miles long, running from the Black Sea coast at Trebizond down through Cappadocia to the Cilician plain. This is not a line. There is no line. There is a graded zone of mountain and high plateau, dotted with forts you can hold for a week and forts you can hold for a year, threaded by perhaps a dozen passes the Arabs can use to bring a real army through, and crossed at any given moment by raiders, drovers, smugglers, monks, pilgrims, deserters, and tax collectors, all of them carrying news.

The Roman system had, in 838, four moving parts. They are worth knowing by their proper names, because your players will hear soldiers use them and will think them more impressive than they are. They were not, on the whole, impressive. They were ordinary and clever and ran on the labour of bored men in small stone rooms.

The kleisoura

A kleisoura is a fortified pass. The word means, literally, a closure. The eastern frontier has perhaps a dozen — the Cilician Gates above Tarsos is the most famous, but there are others above Adata, above Melitene, in the Anti-Taurus, on the Pontic route. Each kleisoura was held by a permanent garrison under an officer called the kleisourarches, who reported in 838 directly to the emperor, not through the local thematic command. The garrison's job was not to stop an army. An army of forty thousand could not be stopped by a wall in a pass; it could only be delayed, harassed, and watched.

The garrison's job, instead, was three things. To watch the road, and light the beacon when a column was seen. To slow the column — by felling timber across the road behind the army's vanguard, by collapsing the small bridges, by burning the forage in the mountain meadows the cavalry would need. And to send fast riders down the back routes to the nearest thematic command, with the column's strength, banner-count, and apparent direction.

A kleisoura garrison in 838 might have been two hundred men in peacetime and four hundred when the campaigning season opened. It was, by the standards of regular Roman regiments, irregular: men recruited locally, often of Armenian extraction, paid badly, fed worse, and astonishingly loyal because the alternative to loyalty was to be the man who let the army through.

The beacons

By 838 the Empire had, on the eastern frontier, a beacon chain of which the Romans were unreasonably proud. It ran from a peak above the Cilician Gates — a hill called by the soldiers Lampegos, the Lamp — north and west through nine intermediate stations, ending on the hill of Mount Auxentios above Chrysopolis, opposite Constantinople across the Bosphoros. Each station was a stone tower with a roofed brazier, two men in summer, four in winter, and a small horse-stable for the relief courier who carried the written dispatch behind the beacon's flame.

A beacon, well-laid, fired in clear weather, would carry the news from Lampegos to the Bosphoros in — depending on the chronicler you trust — an hour, an afternoon, or a day. The truth was probably four to six hours. This is, by any standard, an absurd transmission speed for the ninth century, and the Empire knew it, and built whole strategies around it.

The beacon could carry, by a code of how long it burned and how many secondary fires were lit beside it, four messages. Army sighted on the road. Army strength estimated. Direction of march. And — the message that mattered above all others — emperor required.

It is a near-certain bet, though no chronicler troubles to record it, that the beacons fired in the early summer of 838. The Empire knew, days before the army crossed the frontier, that the enemy was coming.

The signal — how the beacon chain actually spoke

The nine stations of the chain are named here in the order a flame travels them, south to north, because that is the order they fired in the summer of 838. Each station name is the modern peak the chain almost certainly ran across, with the Byzantine soldier's name where one is recorded.

1. Lampegos (Çakıt Tepe) — above the Cilician Gates, the southern eye. 2. Argaios (Mount Erciyes, lower northern shoulder) — covering the Cappadocian plain. 3. Samos (Hasan Dağı) — the bridge station, sighting Argaios south and Aigilon north. 4. Aigilon (a peak above Aksaray on the salt-lake road) — the long western leg across Lykaonia begins here. 5. Mamas (a hill above the modern village of Sivrihisar, east of Pessinous) — the closest station to Amorium itself; the city sees this fire directly. 6. Kyrizos (the Köroğlu range above Beypazarı) — relays west toward the Sangarios. 7. Mokilos (a ridge above the Sangarios crossing near modern Geyve) — the bridge to the Bithynian forests. 8. Saint Auxentios's lesser peak (Kayışdağı, on the Asian side opposite the City) — the penultimate. 9. Mount Auxentios proper, above Chrysopolis — the last fire, watched from the windows of the Great Palace across the Bosphoros.

The chain spoke in four states. A single steady flame, no secondary fires: army sighted, strength not yet read. A single flame with one small fire kindled to the east of the brazier: a column under ten thousand. With two eastern fires: ten to thirty thousand. With three: above thirty thousand, and the courier behind the flame is already in the saddle. A flame extinguished and re-lit within the same watch, three times in succession: the emperor is required in person. There is no signal for 'all clear.' The chain goes dark when the danger passes, and the next watch lights nothing at dawn.

Two men hold each tower in the summer months, four in winter. The men sleep in shifts. The brazier is laid the night before with seasoned pine and a clay pot of pitch buried in the kindling; the pitch is what gives the flame its colour and reach, and is the most expensive single item on the kleisourarches' annual ledger. In clear high-summer weather the chain runs nine stations in four to six hours. In the equinoctial storms it may not run at all for a week, and the relief courier — there is always a relief courier mounted and waiting, carrying the written dispatch the flame cannot — becomes the only message. The courier rides station to station, changes horses, sleeps four hours in twenty-four, and reaches the Bosphoros in eight days if nothing on the road is broken.

Three failure modes the players will eventually meet. (a) A station fires the wrong code — usually by mistake, occasionally by sabotage; either way the relief column rides out and the kleisourarches downstream has eighteen hours to work out whether to believe him. (b) A station does not fire when it should — fog, drink, treachery, or the men dead in their tower with their throats cut. The chain breaks at that station, and the next northern station has to be reached on horseback, costing a day. (c) A station fires correctly and is then over-fired — a second, false flame lit in the hours after, contradicting the first. This is the cleverest sabotage and the hardest to detect. It costs the empire its certainty, which is the thing the chain exists to give.

GM table use. Treat the beacon chain as a clock the GM may advance independently of the party. When a fire is lit at any station, every station north of it lights within the hour on a clear day, within four hours under cloud, and not at all in heavy rain or thick fog. The party may be at any one of the nine. Their decision — to confirm, to override, to extinguish, to ride for the next station themselves — propagates up the chain at the speed of fire, and that is faster than anything else in their world. Use this. It is the single mechanic in this chapter that will make your players feel the geography of the Empire.

The refuge forts

Behind the kleisourai, scattered across the plateau, were the kataphygia — the refuges. These were small fortified hilltops, sometimes no more than a stone enclosure two hundred yards across with a single dry-stone wall and a covered cistern. There were perhaps four hundred of them in the Anatolian provinces in 838. They were not garrisoned. They were maintained by the villagers of the surrounding countryside, who paid no tax for the privilege, and whose obligation in return was to keep the cistern clean, the wall passable, and the gate hinged.

When the beacons fired and the village priest read the news in the square, the village moved. Within a day — sometimes within hours — the cattle were driven up the hill, the grain was carried, the children were lifted, the old were carried last. The doors of the village houses were left open, because doors are valuable timber and the Arabs would take them anyway. The village burned, or did not burn, depending on whether the raid passed through that valley. The villagers stayed inside the kataphygion for as long as the army was on the plateau — two weeks, six weeks — eating the grain they had carried up and drinking the cistern. Then they came down, rebuilt the houses, and were back at work within the month.

This sounds barbaric to a modern player. It was not barbaric to the men who designed it. It was the cheapest defence-in-depth ever conceived, and it worked, and the Empire knew, in 838, exactly how to do it. The villagers around Amorium had drilled the procedure within living memory. So had the villagers around Ankyra. So had the villagers between Tyana and the Cilician Gates, who would, in fact, drill it again that summer, with great efficiency, while the field army marched east to die at Anzen.

The aplekta — the assembly camps

Finally, behind the refuge forts, were the great army camps — the aplekta — at which the thematic levies and the imperial tagmata gathered when the campaign season opened. The principal aplekta in Anatolia in 838 were at Dorylaion, on the road from Constantinople to the south-east; at Kaborkion, a half-day's ride west of Amorium; and at Koloneia in the north. An aplekton had to feed and water twenty to forty thousand men and their animals for two weeks while the army assembled. The maintenance of the cisterns, the granaries, the smithies, and — particularly — the bread ovens of an aplekton was a substantial and continuing imperial expense, justified by the fact that an army that has finished assembling in nine days can be in the field two weeks before an army that begins assembling on the day it is needed.

An aplekton was the visible counterweight to all the small invisible things — the beacon, the refuge, the kleisoura. The army that gathered at Kaborkion in early July 838 was the army that was supposed to use them. It chose not to.

Shadowing warfare — the doctrine

There is a Byzantine military treatise called the Peri Paradromes, On Shadowing — known in modern editions by its Latin title, De Velitatione Bellica — written under Nikephoros Phokas in the tenth century. It is later than 838 by a hundred and thirty years. But it is a written-down version of a body of practice that was already old when the men of 838 inherited it, and it is the clearest single statement of what the Romans were supposed to do, and what they did not do, in the summer the Caliph came.

The doctrine, in short: do not give battle. Let the column come. Hang on its flanks. Cut its scouts. Take its drovers. Light night fires on the ridges to deny it sleep. Foul the wells behind it as it advances and the wells in front of it as it retreats. Strike the baggage train at dawn on the third day, when the men are most tired and the horses most footsore. Above all, do not stand in line of battle, because the Arabs are very good in line of battle, and we, the treatise says with what may be honesty or may be flattery, are not as good as we would like to think.

The campaign for which the doctrine is written is a campaign of refusal. It does not save Amorium. The kataphygia do not save Amorium, because Amorium is too large to be a refuge fort and its walls are not the walls of a small refuge. The doctrine saves the countryside, the livestock, the women and children of the countryside, the harvest, the next year's planting. It does not save the great fortified city the Arab army has decided to break. It buys time for that city — a week, ten days — in which the relief column from the field army can arrive.

The field army did not arrive. That is the entire story of the summer of 838.

Anzen, 22 July 838 — the doctrine refused

Theophilus took the field at the head of the imperial tagmata and the Anatolian theme on the seventh of July, marching east-south-east from Kaborkion toward Dazimon — modern Dazmana — where his scouts reported the southern column of the Arab army under Afshin advancing through the upper Halys valley. The southern column was perhaps twenty thousand strong, two-thirds of it Turkic horse-archers of Afshin's personal command, the remainder Persian and Arab levies. The northern column, under the Caliph himself, was elsewhere — south of Ankyra, marching toward Amorium directly.

The Roman strategy as described to the basileus — the title his own court used; English calls him emperor — by his senior commanders, the strategos of the Anatolikon theme, Aetios, among them, though he was outvoted — was the doctrinal one. Shadow the southern column. Force Afshin to detach his outriders. Refuse battle. Wait for the columns to converge or to separate further; in either case the Roman force, fighting on its own ground, could choose where and when to strike.

The strategy chosen instead was the opposite. The emperor advanced to Anzen, a plain east of Dazimon, drew up in formal order of battle, and on the morning of the twenty-second of July gave the order to attack. The Khurramite contingent under Theophobos held the left. The Roman regulars held the centre. The tourmarchēs of Cappadocia held the right.

The first Roman attack went well. The Persian and Arab levies broke. The pursuit went on for an hour. Then Afshin's horse-archers — held back, on the wings, for exactly this — wheeled in from both sides and shot the centre to pieces. The emperor's bodyguard was cut off; the emperor himself was nearly captured, was saved by the Khurramites, was rescued from the field with a small escort and rode without stopping for two nights. The Roman army did not exist by the next morning.

Amorium was, from the day of Anzen, a city that knew the relief was not coming. It would still be six weeks before the Caliph's vanguard sighted its walls.

The shadow doctrine — rules for the table

What follows is a small, system-agnostic ruleset for running a shadow campaign at the table — bolted onto whatever resolution system you already use. It assumes the party is a band of frontier auxiliaries, four to six strong, attached loosely to a tourmarchēs or to a kleisourarches, with the freedom to ride where the orders do not reach. It does not assume they will be obeyed.

The Column Clock. The Caliph's army (or any large enemy column) is tracked by three values the GM keeps on a card behind the screen. Pace — the days until it reaches its next named objective; reduce to advance, increase to delay. Strength — a single number from 1 to 10 representing effective fighting power; this is the number Aetios and his successors care about. Morale — a single number from 1 to 10 representing the column's willingness to push hard on a given day; below 4 the column halts to rest, below 2 it begins to fragment.

Shadow Actions. On any day the party is in contact with the column — within a day's ride, with eyes on the dust — they may attempt one Shadow Action. Each is a single contested roll against the column's screening force, modified by terrain, weather, and the party's frontier skill. The five canonical actions, with their effect on a success:

Burn the Forage — fire the standing barley and the meadow hay along the column's next day's march. Effect: Pace +1 (the column must halt half a day to send foragers further afield), Strength −0 but morale erosion begins next encounter. Failed: the party is pursued by light horse for d6 hours; one party member tests for capture.

Foul the Wells — drop a dead sheep into the cistern, smash the wellhead, or — for the truly committed — empty a bag of dried hellebore into the spring. Effect: Strength −1 for the next three days as men and horses sicken; Morale −1 immediately. Failed: the party leaves sign; the column's scouts find it and the next Shadow Action this week is at a penalty.

Cut the Scouts — ambush an outrider patrol of four to eight men at dusk. Effect: Morale −1, and the column is blind for the next d6 hours (the GM rolls the column's movement that day without telling the party where it went). Failed: one of the patrol escapes; the column's main body now knows the party is in the field and detaches a screening force.

Night Fires on the Ridges — light a dozen small fires on the hills above the column's camp at the second watch, so the sentries cannot sleep and the cavalry horses will not settle. Effect: Morale −1, and the column starts the following day at half pace until noon. Failed: nothing — this is the safest action and the one a tired party falls back on. Repeat use within a week halves the effect.

Strike the Baggage — a dawn raid on the third day of any march, when the men are most tired and the rearguard most slack. The hardest action, the highest reward. Effect: Strength −1, Pace +1, Morale −2; the column may turn aside to chase the party for one full day, which is also the point. Failed: party takes losses; resolve as a sharp combat against a force twice the party's number.

The Refusal Rule. The doctrine is built on refusing battle. A commander who refuses battle when his subordinates expect battle pays in legitimacy: track a counter called Standing for any named NPC commander. Each ordered refusal costs 1 Standing. Each refusal that turns out, in hindsight, to have saved a force restores 2. When Standing reaches zero the commander's subordinates begin to disobey written orders; when it reaches negative three the commander is recalled, replaced, or — in Theophilus's case — feels in his own chest the need to give the wrong battle, and gives it. This is the mechanical reason Anzen happened. The emperor's Standing, on the morning of 22 July, was at minus four.

Relief and the Field Army. When the beacon chain fires its 'emperor required' code, the field army at the nearest aplekton begins to march within d6 days (GM's roll, modified by season and supply). Its travel time to any point in central Anatolia is one day per fifty miles by good road, half that across country, and it cannot move at all in the first week of a thaw or the last week of an autumn storm. The party may, by Shadow Actions against the enemy column, buy the field army the days it needs. Or they may, by accident or by sabotage, ensure the field army arrives too late. Both outcomes are designable. Neither is a failure of the system.

GM note. None of these rules will save Amorium in 838 — that outcome is fixed. They exist so the players can feel, in concrete numbers on a card, what it cost the empire to forget its own doctrine for a single summer.

Why Aetios was right, and was overruled anyway

Theophilus had reasons. He was twenty-nine. He had spent his reign losing — to Mamun in 830, to Mamun again in 832, to al-Mu'tasim's lieutenants in 833 — and the only victory he could point to was Sozopetra, which had ceased, by the summer of 838, to feel like a victory. He needed Anzen. He needed it more than the campaign needed the doctrine.

The doctrine is not, the historians who study these things will tell you, a doctrine for emperors. It is a doctrine for strategoi, who can lose a province quietly and recover it slowly. An emperor who loses cannot recover quietly. An emperor who refuses battle three times in a single summer becomes, in his own court and in the chronicles of his successors, an emperor who refused. Theophilus understood this with terrible clarity, and chose the worse outcome over the lesser shame.

Aetios, who survived Anzen and rode west to Amorium with what was left of his theme — perhaps four hundred men — would be in the city when it fell. He is one of the forty-two officers taken to Samarra. He is one of the forty-two killed there, on the bank of the Tigris, in March 845. He had argued, at the emperor's council, for the shadow war. He had been overruled. He did not, by any account that has come down to us, mention this again.

For the table — the shadow campaign your party can run

This is the part the players can have. The frontier system above is the canvas. What follows is a six-to-eight-session campaign arc that a party of frontier-themed characters can run in the months before the sack — in the spring and early summer of 838 — and that lets them feel, in their bones, the system the Empire had built and the doctrine it was about to forget.

Session 1 — The Beacon Lit Out of Turn. A station above the Adata pass fires its beacon in late April, six weeks before the campaigning season. The relief column rides out. There is no army. The two men at the station are dead, with their throats cut by a small blade. Someone wanted the relief column out of position. The party is sent to find out who, and why, and from where.

Sessions 2–3 — The Cistern at Loulon. A kataphygion above Loulon has been quietly fouled — a dead dog dropped in the cistern, the cover wedged so the villagers will not notice until the day they need it. The party investigates, drains the cistern, finds two further bodies beneath the dog, and works backwards through the village's quarrels to a man who has been paid in Abbasid dirhams to make sure this particular refuge cannot be held when the army comes.

Session 4 — The Long Ride. The party carries a written dispatch from the kleisourarches of the Cilician Gates north to Dorylaion — eleven days of hard riding through country where the beacons are already firing the wrong codes. They meet an imperial courier who knows a man on the southern road who can be turned. They meet a Khurramite scout who is not yet sure which side he is on. They meet a village priest who asks them, plainly, whether his people should go up to the refuge now or wait a week. They have to answer.

Session 5 — Anzen, Eve of. The emperor's council. The party is in the camp, attached to Aetios as auxiliaries. They hear, through tent walls, the argument. They are sent at midnight on the twenty-first of July with a sealed message to a tourmarchēs in the cavalry reserve, telling him to hold his men back from the dawn advance. They can choose to deliver it. They can choose not to. They can choose to alter it. Whatever they do, the battle still goes the way it goes.

Sessions 6–8 — The Walk to Amorium. The party rides west with Aetios and the remnant. They reach Amorium six weeks before the Caliph does. They have, in their saddlebags, the memory of a system that worked, and the knowledge that the men who ran it are dead. The campaign continues into Chapter 8.

GM note: do not let the party save Anzen. Do not let them, even by stratagem, change the outcome of the battle. The campaign is good because the players cannot win it. The thing they take to Amorium is not a victory. It is the truth about how the Empire used to defend itself, carried by the only people in the city who still believe in it.

What survives today — plates for this chapter

The book will, in this chapter, carry a sequence of nine then-now plates. They are listed here for the GM who wants to print them out and put them on the table.

1. The Cilician Gates / Gülek Boğazı. Modern photograph from the southern approach, looking up the pass. The road still goes this way. The army came this way.

2. The hill of Lampegos — modern Çakıt Tepe, twelve miles north of the Gates. A modern photograph of the summit, where a small ruined stone enclosure with a fire-blackened interior is, by the best identification, the southernmost station of the beacon chain.

3. Hisarköy mound (Afyonkarahisar) — the citadel of Amorium today. From the photograph by Klaus-Peter Simon. Already in the gallery.

4. Derinkuyu underground city. The first of the great Cappadocian refuges. Eight levels deep, ventilation shafts cut to the surface, millstone doors. Not a kataphygion in the strict sense — it is a refuge city, larger — but it was in use through the ninth-century raid period, and a player who climbs down its first ladder will understand, instantly, what a refuge meant.

5. Açık Saray, Cappadocia. A rock-cut village of the eighth to tenth century, with stables, oil press, dovecotes, and a small church. The closest surviving analogue to a ninth-century rural farmstead in the empire. The honest answer to the question 'what did a Roman farmhouse of 838 look like.'

6. The walls of Ankara. The inner Byzantine citadel walls still stand, on Roman foundations, on the hill above the modern city. They were rebuilt in part by Michael III a generation after the events of this book, but they preserve, in their lower courses, the masonry that watched the Caliph's northern column pass beneath in July 838.

7. Binbirkilise / Karadağ, Lykaonia. The Thousand and One Churches — a complex of perhaps fifty surviving ecclesiastical buildings, dated from the fifth through the ninth centuries, on the slopes of an extinct volcano in central Anatolia. A working religious landscape in 838, abandoned by the eleventh century, photographed beautifully in the dry summer light.

8. The pass of Dorylaion — modern Eskişehir. The plain where the imperial army gathered. The aplekton itself is invisible, but the geography is the same: a flat watered ground at the foot of a low ridge, looking south-east toward the campaign road.

9. The field at Anzen / Dazimon — modern Dazmana, in the Tokat province. A wide level plain bounded north by the Yeşilırmak. The plate is a single landscape photograph, taken in late July at the time of year the battle was fought, with no caption beyond the date.

Each plate, in the printed book, faces a half-page artist's reconstruction of the same site in 838 — the beacon lit, the refuge populated, the army drawn up. The reconstructions are commissioned. The photographs are credited. Both are printed in sepia, on the heavy stock the rulebook uses for plates.

A closing word, for the GM

The reason this chapter exists is that the Empire's defence in depth is one of the most interesting and least-romanticised military systems of the early medieval world. It is interesting because it is humane. It assumes the survival of the peasant, the cistern, the goat, the priest. It refuses the heroic battle in favour of the unromantic week of bad sleep and worse food and small fires on a ridge.

The Empire forgot this in the summer of 838 for reasons that were not stupid. It is worth giving your players, through whichever sessions of the shadow campaign above you choose to run, the chance to remember it on the Empire's behalf. They cannot save Amorium. They can carry, into the city as it ends, a quieter kind of knowledge — that there was once a way to do this, and that the way was good, and that the people who ran it were almost no one anyone has ever heard of, and that this is what the Empire was, when it was working.