Philip Sabin’s Analysis of Ancient Warfare in Lost Battles

I really like the analyse of Ancient Warfare in Lost Battles by Philip Sabin. So I’ve written up what captured my attention.

The book is available from Amazon USA, UK, and Canada:

Sabin, P. (2009). Lost Battles: Reconstructing the Great Clashes of the Ancient World. Bloomsbury Academic.

Interpreting Ancient Battles

We’re severely restricted in source material for ancient battles – that is why Sabin latched onto “Lost Battles” for the title of his book.

Unfortunately, there are severe limits to our understanding of these long distance [i.e. ancient] engagements. Not only have transformations in the face of war left us with no personal experience of the dynamics of massed close quarter fighting, but the ancient evidence that has survived is lamentably thin – typically, just a few pages of often ambiguous or unreliable text on any individual battle … [which means] modern scholarship teeters unsteadily above a narrow and unsatisfactory evidentiary base. (p. 1)

Because of that Sabin attempts to adopt a general’s eye view or grand tactical approach to analysing and simulating battles.


Sabin organises all armies into about 20 units. This is a fairly arbitrary decision but dodges the issue that we have a lot of detail about small scale units for the Romans but for other armies we’ve got almost no information. Although the armies have a similar number of units the units vary in size so the armies have quite a range in terms of number of men.


Units are categorised by type and class. Although Sabin’s “type” combines whether the unit is mounted (infantry/cavalry) and weight of unit (heavy/light). Deconstructing that most units are categorised in three ways:

  • Infantry or cavalry
  • Heavy or light
  • Veteran, average or levy (Sabin calls this “Class”)

In this model all light troops are dispersed skirmishers. Heavy troops are dedicated close combat fighters and those of a more intermediate nature.

Elephants and chariots are handled a bit differently. Elephants are African or Indian and can be accompanied by light infantry escorts or unaccompanied. Chariots are average class. Scythed chariots are different again and pretty rubbish.

Sabin largely ignores equipment differences however does give a few troop types special characteristics:

  1. Cataphracts – these fully armoured cavalry with lances are ponderous but protected
  2. Hoplites
  3. Close-order archers
  4. Phalangites – pike armed
  5. Roman legionaries with distinctive multiline deployment and overall ability to prevail and dominate
  6. Chariots
  7. Scythed Chariot
  8. Elephants

The first and most important measure of fighting power at the unit level, far ahead of any issues of equipment or fighting style, concerns the morale, discipline, cohesion, skill and reputation of the troops themselves. Lendon has highlighted very well how Greek authors such as Polybius tended to ascribe military success to better tactics, techniques or equipment, whereas Roman writers like Caesar and Livy were more concerned with the superior bravery and virtus of the victorious soldiers. The evidence from both Greek and Roman battles lends much more support to the Roman interpretation – how else can one explain the dominance of Spartan hoplites, or Eumene’s veteran Silver Shields, or of Caesar’s legionaries over much larger numbers of similarly equipped opponents. A key ingredient in such triumphs against the numerical odds seems to have been the intimidation of less-resolute adversaries. The Persian masses at Cunaxa fled from the 10,000 even before they had reached arrow range, and, by contrast, Xenophon has another very telling anecdote of how a small Spartan force with borrowed shields was massacred, in part because their more numerous opponents were misled by the shield blazons and did not realize they were fighting Spartans. (p. 19-20)

Following the Roman writer’s lead Sabin assumes that, man-for-man, veteran troops are about three times more effective than average soldiers, and average soldiers are about three times more effective than levies. At the unit level Sabin separates combat effectiveness and the number of men to approximate these effectiveness ratios. His combined model means two levy units have a similar combat value of a single veteran unit, but have eight times as many men in the ranks. The specific values are:

  • Number of men: Sabin uses a ratio 1:2:4 men in a veteran:average:levy unit
  • Combat values: Sabin gives values of 4, 3 and 2 for veteran:average:levy units

Lost Battles gives each unit a combat value depending on its type:

  • 4 – Veteran unit, average Roman legionary unit, Indian elephant unit
  • 3 – Average unit, African elephant unit
  • 2 – Levy unit
  • 1 – Scythed chariot unit

The direct contribution of horsemen in pitched battle equated very roughly to that of twice as many infantry soldiers of a similar calibre. This can be reflected in our unit system by making each cavalry unit represent half as many troopers as an infantry unit of the same quality. (p. 22)

Each battle/scenario has a troop multiplier from 1-8 (actually 1, 1.5, 2, 2.5, 3, 4, 5, 6 or 8). A multiple of 1 is for a small battle. Large battles have bigger units – just multiply the number of men in the unit by the multiplier.

In a small battle (battle multiplier = 1) the troop ratios are:

  • levy unit has 1,000 infantry or 500 cavalry
  • average unit: 500 infantry, 250 cavalry, 100 two-horse chariots or 50 four-horse chariots
  • veteran unit: 250 infantry, 125 cavalry.
  • Scythed chariot unit: 25 four-horse vehicles
  • Elephant unit: 5 beasts with 250 light infantry escorts or 10 unaccompanied beasts

Battle field

Usually infantry within an army were divided into left, right and centre. Cavalry would be on the far right or far left. Armies would also have a front line but option for a reserve, or space to recoil. Lost Battles simulates this by representing every battle field with an array of 5×4 zones, i.e. the frontage is five sectors and the depth is four swathes. The width and depth of a zone are roughly equal. The larger army fills all five sectors but the smaller army may not fill one or both flank sectors.

The relationship between army size and sector size is non-linear. Roughly speaking quadrupling the army size only doubles the sector size. Or, put another way, the frontage of the army is directly proportional to the depth, so increasing the frontage increases the depth. Doubling these dimensions – frontage and depth – quadruples the number of men necessary to fill the space created.

The frontage of armies did not vary with the number of troops in a linear fashion. Larger armies seem to have formed up instead in significantly greater depth to avoid the command problems and terrain constrictions that would have hindered a thinner but longer line. (p. 31)

Sector widths for each battle range from 300 m to 1,200 m, with the larger distances obviously being more common in scenarios with higher troop multiples, but with the scale in different scenarios with the same multiple varying depending on individual circumstances. (p. 32)

Camp locations are often highly speculative, and to avoid undue complication it seems best to ignore camps not on the actual battlefield and to treat all armies the same in terms of initial deployment, regardless of the length or direction of their approach march. (p. 33)


In some cases, however, especially in larger battles with deeper zones, one or both camps were so close to the fighting line as to warrant being placed on the actual battlefield. (p. 34)

Most battle fields were flat open plains. But significant features sometimes occurred.


On the fixed size table the sector width impacts the game turn length. A Lost Battles turn is 3 min for every 100 m represented by a zone. That gives a game turn of between 10 to 40 min.


In Lost Battles infantry move 1 zone per turn. Everything else is relative.

  • Some terrain is “difficult”, e.g. hills, woods, marshes. This slows mounted troops and restricts troop density.
  • Shallow rivers and streams don’t impeded movement
  • Full speed for infantry is twice as fast as their normal, cautious movement allowance
  • Cavalry and chariots go twice as fast as infantry unless they enter difficult terrain or a fortified camp
  • Full speed for cavalry is also twice as fast as their normal movement allowance, but it not possible in difficult terrain
  • Infantry changing face takes their entire normal movement allowance; for cavalry changing face takes half their normal movement allowance
  • Camps invite looting hence delay troops passing through


Sabin models the outcome of combat and is not interested in casualty calculation.

it seems hopeless to base our combat system on any form of casualty rate calculation grounded in the number and lethality of the weapons being brought to bear.

There was a paper/scissor/stone thing going on providing swings and roundabouts in combat outcomes (p. 23, p. 55):

  • Heavy cavalry could catch and ride down light infantry
  • Chariots have less of an advantage against light infantry
  • Steady heavy infantry could repulse heavy cavalry
  • Light cavalry could shoot at heavy infantry without danger to themselves but also without inflicting significant damage
  • Light infantry could out shoot light cavalry due to greater numbers and mounts of the cavalry being unprotected
  • Archers can out shoot light troops
  • Elephant enjoyed a huge advantage over heavy cavalry and lesser one over heavy infantry
  • Indian elephants intimidated African elephants
  • Chariots have the edge on heavy cavalry

Units get an advantage in combat when the unit is:

  • non-levy and attacking levy
  • infantry and attacking downhill
  • outside a river and attacking enemy in, or emerging from, a river
  • fresh hoplites or phalangites and attacking heavy infantry (unless the attackers are outflanked or in broken terrain)
  • heavy infantry and attacking hoplites, phalangites or archers who are not facing the attack
  • archers attacking light infantry or light cavalry
  • cavalry or chariots and attacking the flanks or rear of heavy infantry

Units are disadvantaged in combat when the unit:

  • has adjacent enemies to their front and rear or their left and right
  • is heavy infantry (not in it’s own camp) or non-infantry, and attacking into or from wood, marsh, hill or fortified camp. Light infantry are not affected by this terrain.
  • is infantry and frontally attacking fresh legionaries. This extends Roman infantry contests
  • moved a long way or changed facing
  • is frontally attacking cataphracts
  • is cavalry or chariots frontally attacking elephants
  • is chariots, archers, spent light infantry or light cavalry, or in or emerging from a river, and in wet weather. Representing wet bowstrings, bogged chariot wheels, and/or flooding.

Sabin disadvantages hoplites, phalangites and close-order archers when they are attacked from flank or rear. But not other troops.

Attacking the enemy in flank or rear is often felt to justify massive combat bonuses, but this may in fact be exaggerated. At Ibera, the Romans easily dealt with encirclements on the flanks after putting their frontal opponents to flight, and at Telamon and Ruspina, troops fought for a long time despite having enemies to both the front and rear … However, certain subtypes of heavy troops do seem to have bought their extra strength in frontal combat at the expense of added vulnerability to attacks from other directions … Hence, attacks by heavy infantry (of whatever kind) on hoplites, phalangites or archers not facing the attack should receive a +1 dice roll modifier (p. 54-55)

Hoplites and phalangites benefit from the ability to mass more units in a given area. [Presumably reflecting a deeper formation.]


It was not just direct confrontations with opposing troops that caused forces to break and run. Panic could prove infectious (especially in the later stages of an engagement), and it was rarely necessary to defeat every enemy contingent individually in order to prevail. Such escalating panics were the clearest instance of the non-linear ‘butterfly effect’ to which Kagan alludes. (p. 56)

An army gets an advantage in Army Morale when:

  • There are at least twice as many friendly as enemy units currently on the field

An army is disadvantaged in Army Morale for:

  • Friendly units lost in combat (but not through morale failure)
  • Friendly generals lost
  • Being outnumbered i.e. there are at least twice as many enemy as friendly units currently on the field
  • Enemy occupies relatively more of the battlefield, particularly the key zones for the army (usually the centre) and not particularly any river banks
  • Friendly camp having been occupied at any point

A unit is advantaged for Unit Morale when:

  • Veteran
  • Average legionary
  • Indian elephant
  • Heavy infantry (except for hoplites versus hoplites)
  • A long distance from the cause of the morale test e.g. lost friendly unit is on the other flank
  • A good general nearby

A unit is disadvantaged for Unit Morale when

  • Levy
  • Scythed chariot
  • Guard of a poor general
  • Spent
  • Close to the cause of the morale test e.g. adjacent to the lost friendly unit
  • Surrounded, i.e. with enemy on opposite sides of the unit (front/rear or left/right)

To preventing fighting to the last man an army automatically breaks if only 5/6th of the units in the army is broken.

Disclaimer: Some of the links contained within this page have my referral ID (e.g., Amazon), which provides me with a small commission for each sale. Thank you for your support

2 thoughts on “Philip Sabin’s Analysis of Ancient Warfare in Lost Battles”

  1. This is so very useful to my present area of research into developing a realistic rules for ancient combat
    Many thanks


Leave a Reply