How Long does a Fight Take?

[Patrick Waterson]

This thread is prompted by a few remarks in a recent discussion where a question arose about how well we can determine the reliability and duration of the resisting power of various troop types, and hence how long an action between such types will last. 

For example, one can determine with fair precision how long one unit will last in melee against a superior opponent in a straight frontal fight (which I do in Optio).

You surprise me Justin.  That is one area where I'd expect us to struggle as a) very few fights featured situations where the only variable is troop quality and b) our sources aren't particularly good on timing (for various reasons).   But that is another topic.

This is of significance for rules systems that do not leave the duration of engagements and/or the durability of units within the provenance of dice.

I suggest we examine the following aspects:

1) Can we deduce how long a particular troop type will last against another given troop type, given no great disparity in training, experience, terrain or fatigue?  (There is an incidental buried sub-question here about quantifying the latter four attributes.)

2) Can we determine how long an infantry fight lasts compared with a cavalry fight, and is this meaningful?

3) How predictable is the outcome of such a fight?

For answers to these we turn, as ever, to the historical record, which I propose to take at face value except where sources contradict each other, in which case we rule which source is likely to be more reliable and follow that source.

Hannibal's battles form a good starting-point for such attempts at analysis, being described by Polybius in reasonable, and apparently accurate, detail - and by others with less detail and perhaps accuracy.


Trebia

Observations
Skirmishers have little effect on each other but the Carthaginian skirmishers are considered by Livy to have a significant effect on the Roman cavalry and by Polybius to have a noticeable effect on the flanks of the Roman infantry.

The cavalry battle is over before the infantry battle, and gives the Carthaginian skirmishers some (indefinite) time to operate against the Roman infantry flanks.

The collapse of the Roman flanks, which are fighting predominantly against Libyans and Spaniards, is heavily influenced by frontal elephant pressure and flanking shooting by Carthaginian skirmishers, in about that order.

The Roman centre, unimpeded by elephants or shooting, breaks through the Gauls and Libyans opposing them.

Conclusions
1) In a straight fight, Roman infantry break Gallic infantry and Libyans - this is the ultimate result of a long-drawn-out melee.

2) Where elephants and shooting are added, this result is reversed, at least against Libyans.  The melee is still a long one.

Moving on ...


Trasimene

Observations
Skirmishers do not appear to be a factor in this classic army-sized ambush.

Cavalry is noticeable mainly in the 'pursuit' (they ride down Roman infantry in the lake - Roman cavalry are nowhere to be seen, presumably vanquished at the outset).

Despite being ambushed, a force of 6,000 Romans (possibly the extraordinarii and leading ala or legion) break through a Libyan contingent and march off the field.  These seem to have been the only Roman troops who were able to take their opponents head-on as opposed to being flanked by them.

Conclusions
1) In a straight fight, Roman infantry still break Libyan infantry - never mind that the Libyans are veterans and the Romans are not.  (There is an evident weapon-systems-do-matter point buried in here.)  They seem to have done so before the main action was concluded, otherwise they would have been taken in the rear before being able to break through - although it may be noted that being taken in the rear actually helped the Roman centre to break through at Trebia - and would have been at least harassed while marching off the field.


Cannae

Observations
Again, skirmishers have little obvious effect on each other.

The cavalry fight on the Roman right is over quickly: Polybius times the conclusion of the action as taking place when the lines of heavy infantry met.  Things last longer on the left, the Numidians freezing Varro's men in place by use of unfamiliar tactics, but the appearance of the Carthaginian cavalry from the opposite flank is enough to clear the Roman cavalry off the field.

Just for once, although the Romans cause Hannibal's mixed line of Gauls and Spaniards to fold - apparently in fairly short order - the Roman centre does not break through, although the only troops it could have come up against would be the balance of the Gauls in Hannibal's army.

On the flanks, the Libyans execute a rapid enveloping manoeuvre and trap and mince the Roman wings.  This seems more rapid and effective than the Trebia combination of frontal pressure, shooting and elephants.  The commitment of the Carthaginian cavalry, apparently against the unattacked portion of the Roman rear, closes the encirclement nicely.

The Roman army, which judging by the timing of the Volturnus becomes seriously engaged about noon, is slaughtered by a point in the afternoon that allows the Carthaginians enough time to rescue their own camp, take the Roman camp and take prisoner about 2,000 fugitive Roman cavalry, although the latter was performed simultaneously (by the pursuing Numidians) rather than sequentially.

Conclusions
1) Hannibal's rearmed troops proved impervious to the usual Roman infantry breakthrough - as did the un-rearmed Gauls, once the Roman army was encircled.  Before it was encircled, neither Gaul nor Spaniard could hold the Romans for more than a short while.  Encirclement seems decisive.

2) The cavalry fights were over quite quickly compared to the infantry fights - at a guess, less than one hour compared with maybe 3 hours.


Zama

Observations
Skirmishers again had little effect on each other - the Roman skirmishers were anyway mostly busy with elephants.

Elephant pressure was largely negated by Scipio's preparations and tactics.

Cavalry fights in this battle seem to have been fighting withdrawals by the Carthaginian contingents, which were charged when off-balance and never really recovered.  The cavalry was off the field quite early in proceedings and the Roman cavalry quite late to return.

The first-line combat pitted Roman hastati against Celtic opponents of greater skill and experience.  The Romans won through superior weaponry and technique.

Next came Roman hastati against Carthaginian citizen troops.  The Carthaginians fought long and hard (and not only against the Romans, but also against the aggrieved survivors of their first line) before the survivors broke.

Finally came the clash of the veterans: two lines of troops equal in numbers, weaponry and just about everything else.  This fight went on for quite some time and Polybius hints that the Romans were getting the worst of it when they were saved by the return of their cavalry, which decided the action.

Conclusions
1) Romans still beat Gauls (or Ligurians) in a straight fight.

2) Romans still beat African-style heavy infantry in a straight fight (unless the Celts fighting against both sides made a crucial difference).

3) Romans cannot overcome Hannibal's rearmed 'imitation legion' types in a straight fight.

4) Cavalry fights are much more rapid than infantry fights (even allowing for a little help from the elephants).

5) Getting hit in the rear by cavalry is decisive.


The above actions suggest that infantry fights, at least in the Second Punic War, tended to be of long-ish duration, and that Roman infantry would beat Gallic or Libyan infantry in a straight fight (we can add Spanish infantry to that list, again in a straight fight).  Conversely, a straight fight between Roman infantry and Hannibal's post-217 BC infantry seems to have gone, if anything, in favour of the latter (buried rearmament point there).

Cavalry fights seem to have been over much more rapidly, although a certain disparity of force was present in the above battles, so whoever won the cavalry fight would be able to influence the ongoing infantry action.

Infantry collapses seem to have been very rare in this period - the only obvious one is the Spanish at Dertosa (Ibera) in 215 BC, and that, we are told, was because they did not want to go to Italy and so were determined to lose the battle!  Troops on both sides seem to have been highly motivated, with considerable staying-power.

That is a very rough and ready dive into the topic, but I believe demonstrates that we can determine with reasonable accuracy which troop type will win and to an extent how long the loser will last - the latter will need reference to other battles, notably Caesar vs Gauls and some Roman vs Roman actions in which duration is specified, but this post is already long enough!  We can extend this type of approach by looking at other armies and periods, though classical armies and battles are perhaps the best-described as a whole.

[Imperial Dave]

Excellent article Patrick and much food for thought.

To my untrained eye, classical battles (in the wider arena) with predominantly infantry were long drawn out affairs, ie othismos (whether you take this to be the shoving match beloved of many commentators or just "staying power" generated by deeper formations). Cavalry engagements, as you point out, tend to be over quickly or occur in several discrete phases throughout a battle.

Here is a question that I am musing over outloud and that is

Do battles between unfamiliar antagonists with or without technological differences generate quicker battlefield resolutions?

I dont have the answer, I am merely extending your line of reasoning....! 

[Justin Swanton]

Just one question, Patrick: is the speed of the cavalry fights in the examples given at least partly due to the disparity of numbers between the opposing sides? What does the record say about cavalry vs cavalry combats in which the protagonists had roughly equal numbers?

My own impression is that a horseman's morale was much more brittle than an infantryman's, due perhaps to his more exposed position on his horse, which meant that a slight change in fortune in a cavalry fight could lead to a rout more easily than in an infantry vs infantry contest. But what evidence is there for this?

[Erpingham]

That is a very rough and ready dive into the topic, but I believe demonstrates that we can determine with reasonable accuracy which troop type will win and to an extent how long the loser will last

I'd refer you back to my earlier comment and particularly Justin's comment that led to it.  Justin was comparing rate of defeat in combat versus rates of march to determine whether a flank attack would arrive before a centre gave way IIRC.  I think Patrick you have demonstrated we can't do this, even with your level of mastery of the sources.  Yes, we can say fights were long or short, or even longer or shorter than other fights but can we can't put a stopwatch on it.  I am also not convinced for much of the period we are interested in, people on battlefields made precise records of how long a fight took.  We are used to time in much more precision than they were.  This is without the psychological elastic perception of time is stressful situations.

So, while I actually think your idea of comparisons is a clever one and a good model (indeed there is a computer modelling technique which allows you to predict the winners in sporting matchups that never happened which uses a series of overlapping comparisons), I'm dubious of claims of timing precision.

[Justin Swanton]

I'd refer you back to my earlier comment and particularly Justin's comment that led to it.  Justin was comparing rate of defeat in combat versus rates of march to determine whether a flank attack would arrive before a centre gave way IIRC.  I think Patrick you have demonstrated we can't do this, even with your level of mastery of the sources.  Yes, we can say fights were long or short, or even longer or shorter than other fights but can we can't put a stopwatch on it.  I am also not convinced for much of the period we are interested in, people on battlefields made precise records of how long a fight took.  We are used to time in much more precision than they were.  This is without the psychological elastic perception of time is stressful situations.

I would qualify what I said in that knowing exactly how long a straight infantry or cavalry fight lasted is obviously impossible since we weren't there with a stopwatch. However I agree with Patrick that we can form a rough comparative idea. Cannae is the best example: Hannibal's entire plan rested on knowing how long his Gauls and Spanish would last in the centre whilst his left wing cavalry beat their Roman counterparts, marched around the Roman rear and beat the Roman left wing cavalry, then marched to the centre of the Roman rear to attack the back of the Roman infantry.

The question here is to what extent the Romans could keep up the pressure on the centre after the Libyans had wrapped around their flanks. If the wrap-around occurred after the defeat of the Roman cavalry on both wings then the implication is that the Roman legions were able to lay it on for some time without interference on their flanks. If however, the wrap-around occurred before the conclusion of the cavalry fights - i.e. Hannibal was able to tempt the Roman infantry to advance ahead of their mounted flank guards - then the Allied legions at least would have found themselves in trouble virtually from the debut of the infantry fight. That still left the Roman legions in the centre relatively unaffected until the Carthaginian cavalry began to pressurise their rear.

Final conclusion: for wargaming purposes an infantry fight between a Roman and a romanised infantry line should last long enough for all the cavalry action of Cannae - fighting and movement - to be concluded, without either infantry line being broken, even if one line is numerically superior to the other. One doesn't need a stopwatch for this, just rules that ensure the infantry engagement will last as many turns as are needed for superior cavalry to beat inferior counterparts then pile into the backs of the infantry.

The same sort of reasoning can be applied to the other examples.

[Imperial Dave]

Just one question, Patrick: is the speed of the cavalry fights in the examples given at least partly due to the disparity of numbers between the opposing sides? What does the record say about cavalry vs cavalry combats in which the protagonists had roughly equal numbers?

My own impression is that a horseman's morale was much more brittle than an infantryman's, due perhaps to his more exposed position on his horse, which meant that a slight change in fortune in a cavalry fight could lead to a rout more easily than in an infantry vs infantry contest. But what evidence is there for this?

Easier to make a break for it when things go against you in an engagement? Why hang around if you feel the battle is going badly and you are sat upon the biological equivalent of a motorbike

[Tim]

No expert at this but I can think of two or three Cavalry engagements where the mounted stayed the course.  Arbela where one flank of Alexander's army last long enough for him to defeat those facing him and come across to the other who seem to have been hard pressed.  Hastings where the Norman cavalry make repeated charges uphill (the ability to rest behind their own foot may be a factor). Kadesh where again the Egyptian mounted holds on against impossible odds, probably because it has foot to go and hide behind, slay all the enemy, and then grant them very favourable terms (I accept that our primary sources for this one may be a little unreliable...)

[Patrick Waterson]

Tim is right, and we can add Issus, where the Thessalians and Greeks on Alexander's left bottled up many times their number of Persian cavalry, although in this case they did not have to hold them for very long because Alexander's first charge took him to Darius, who remembered a pressing appointment elsewhere once his bodyguards started falling in numbers and thus pulled the rug from his own troops' continuing prospects.

At Arbela (Gaugamela to many) Parmenio's cavalry held out - just - against Mazaeus, but Parmenio did not view his prospects very brightly, sending repeated appeals for assistance right from the start of the fighting and ultimately being saved by the Persians streaming off the field following Darius' departure (again), although Alexander had turned back to help.  Conversely, on the Persian left, Alexander tempted Bessus' cavalry out of position and, by Arrian's account, held the 20,000-strong but deeply packed Persian cavalry contingent with two squadrons (maybe 600 men) and hit them in the rear/flank with the prodromoi, another 300 or so, who drove them off in rout.  This did not take long - just long enough for Parmenio to send his first plea for help.  (Alex ignored him, went in after Darius and won the battle.)

One case of apparently evenly-matched cavalry was at Heraclea (Pyrrhus vs Romans, 280 BC) - not a particularly well-described battle, but one in which Plutarch says the action swirled back and forth with "seven changes of fortune".  Now I have yet to see an infantry phalanx survive even one 'change of fortune' in a battle, so it looks as if Plutarch was referring to the cavalry action, and this suggests that it surged back and forth at least three times (initial Epirot charge, initial Roman rally and repulse of Epirotes, Epirot rally and second charge, etc.) until Pyrrhus cheated by adding the elephants so that 'fortune' was replaced by 'pachyderm fright' and the Roman cavalry were cut to pieces.

Assuming Plutarch and his source could count, this gives us a cavalry fight moving at at least three times the rate of the infantry fight: the cavalry action was 'decided' three times (or six times?) while the infantry were still hard at it, albeit without a clear winner, and the seventh clash - with elephants - sorted matters out very quickly while the PBI were still shoving away.

I'd refer you back to my earlier comment and particularly Justin's comment that led to it.  Justin was comparing rate of defeat in combat versus rates of march to determine whether a flank attack would arrive before a centre gave way IIRC.  I think Patrick you have demonstrated we can't do this, even with your level of mastery of the sources.  Yes, we can say fights were long or short, or even longer or shorter than other fights but can we can't put a stopwatch on it.  I am also not convinced for much of the period we are interested in, people on battlefields made precise records of how long a fight took.  We are used to time in much more precision than they were.  This is without the psychological elastic perception of time is stressful situations.

I think Justin's approach has validity, in that Hannibal's deployments and approach at Cannae worked really well - the timings all came together and Polybius' description has the events in sequence and related to one another, giving us a useful 'chronology' for this particular battle.  The value of Cannae is that Hannibal evidently made his battle plan with the resisting times of various troop types as key building-blocks in his scheme ("I love it when a plan comes together!" ) - it would not have worked otherwise.  The interaction of the Carthaginian manoeuvres and the almost 'timed' collapse of Hannibal's Gallo-Spanish centre plus the rapid defeat of the Roman right were integral to Hannibal's battle and it could not have worked if his cavalry took too long to defeat the Roman right or if his veterans had taken too long to gift-wrap the Roman wings.  The precision and neatness of Cannae still fascinated general staff officers 2100 years later.

Granted that we cannot apply a stopwatch to these events - in any event, with hours of variable length depending on the time of year it would not work too well - but we can see how they relate to each other, and that in itself is valuable for determining relative combat persistence.

Just one question, Patrick: is the speed of the cavalry fights in the examples given at least partly due to the disparity of numbers between the opposing sides? What does the record say about cavalry vs cavalry combats in which the protagonists had roughly equal numbers?

My own impression is that a horseman's morale was much more brittle than an infantryman's, due perhaps to his more exposed position on his horse, which meant that a slight change in fortune in a cavalry fight could lead to a rout more easily than in an infantry vs infantry contest. But what evidence is there for this?

Easier to make a break for it when things go against you in an engagement? Why hang around if you feel the battle is going badly and you are sat upon the biological equivalent of a motorbike

I hope the example of Heraclea illustrates the course of a cavalry fight when capabilities are pretty much exactly equal.  Cavalry if disadvantaged seems happy to give ground and reculer pour mieux sauter, which they will do for as long as their cohesion holds.  This is why I see the cavalry fight at Zama not as a rout and off-board pursuit but as a running fight in which the Carthaginian cavalry maintained their morale and cohesion but never managed to regain their formation, so they were giving ground for a long time (and well off the battlefield) before they finally came apart.  Conversely, where there is absolutely no question of giving ground they will stand and fight, as the Roman right did at Cannae, even dismounting to avoid any retreat.  However cavalry caught on the wrong foot, or facing the wrong way, can be very vulnerable to panic and rapid rout, witness the Persian left at Arbela/Gaugamela and Tigranes' cataphracts at Tigranocerta.  Each of these contingents collapsed and ran without further ado when unexpectedly hit in flank.

I think we can safely say that cavalry is quite resilient while its morale is good, that its battlefield system is much more flexible than infantry, especially when it comes to yielding and regaining ground, and that it can panic more rapidly when unpleasantly surprised, unless this is simply panic travelling at the same speed but four-legged panickers being faster at implementing it!

[Justin Swanton]

I would make one additional comment: the timing of infantry and cavalry fights was something sufficiently precise for Hannibal to be able to build his battleplan around it. He did not form up at Cannae crossing his fingers he would throw a few lucky 6's in the centre whilst his opponent threw a few unlucky 1's. Hannibal was anything but a gambler.

[aligern]

I oncedid a week's bookbinding course and was impressed at how they taught us to get measurements right.  The tutors did not keep measuring paper and card to get them to equal size. Wherever possible they made one measurement and then matched the paper to that.  That way it was impossible to end up with a situation where, for example the holes fir sewing the paper into the cover did not match with the holes in the cover.
I suggest that the way that the Ancients made objects and developments relied a lot upon mebthods where the initial measurement was made and then other processes followed and were matched to that measurement.
Hence to say Cannae. Rather than Hannibal being able to make a fine calculation as to the time events would take I envisage him learning from previous actions  and thus interspersing his Gauls and Spaniards to stiffen the Gauls and bowing their line forwards so that it would become denser as it was pushed back  and then having the African assault triggered by the Celto-Iberians passing a certain point. The retirement of the Carthaginian skirmishers will have been  triggered by the advancing Romans reaching a certain range and the attack of the Carthaginian cavalry by the arrival and deployment of the skirmishers which gave them superiority they needed over the Roman horse.
So the Cannae plan runs like clockwork and is based on H's accurate  perception of the way the Romans would react, but it is even more like clockwork  because each Roman move triggers the next Carthaginian  action. I would not call the Carthaginian moves responses, though, because they are  moving to Hannibal's plan, it is just that the Romans dictate the timing of  the Carthaginian actions.
If Cannae works as I suggest then it does not need Hannibal to have a stopwatch on events, because his men only move  as the Roman advance triggers them.
Roy

[Patrick Waterson]

Roy's point about interdependence of actions is a good one, in that Cannae depended upon a lot of parts slotting into place - with the cooperation of the Romans, which is why Hannibal waited for a day when Varro was in charge.  He needed a belligerent ignoramus in charge of the opposition for his plan to work.

Naturally, integral to the self-adjusting mechanisms involved is some idea of how long it will take those mechanisms to do the adjusting - there is little point in designing an encirclement that traps empty air because the 'anvil' could not hold the enemy for long enough to close the trap.  Hannibal would know almost to a breath how long it would take troops to execute a particular manoeuvre, and he had just had a whole winter to work on getting it exact, so the combination of interdependent activities and precise timings for execution of the parts under his control left him fully in charge of events on the day.

As Roy says, the whole thing is not run according to an inflexible schedule, but the respective manoeuvres are triggered as the appropriate battlefield conditions/criteria are reached.  Once these are reached and the manoeuvres triggered Hannibal knows exactly how long they will take to execute, having had the opportunity to practice for most of the winter.  We can also see how Hannibal's dispositions helped these trigger criteria to come about: the curved line of Gauls and Spaniards would temporarily stop the Romans where they touched it, so the ends of the Roman line would advance further than the centre and, more significant, further than their protecting cavalry, leaving a space for Hannibal's enveloping veterans to get behind them without risk or delay.  The curved line was not an affectation, it was a means of opening up a gap behind the Roman flanks.  It also meant that when the Gauls and Spaniards could resist no longer the Romans would follow up starting from the centre, straightening the line again just as the encirclement was going in and hence moving forwards the troops in the Roman rear that might have impeded the encirclement had they remained where they were.

Hannibal's plan was a classic combination of knowing what his troops could do and anticipating what the enemy troops would do.  He still needed to anticipate how long a particular phase of the battle was likely to take, otherwise the best-laid plans would be thrown out of kilter and he would have, at best, an incomplete victory.  As it happened, he got everything right (including the enemy leadership), and Cannae was his best-ever battle.

[Erpingham]

If Cannae works as I suggest then it does not need Hannibal to have a stopwatch on events, because his men only move  as the Roman advance triggers them.
Roy

So, more IGO-UGO than simultaneous moves then 

[Erpingham]

Hannibal's plan was a classic combination of knowing what his troops could do and anticipating what the enemy troops would do.  He still needed to anticipate how long a particular phase of the battle was likely to take, otherwise the best-laid plans would be thrown out of kilter and he would have, at best, an incomplete victory.  As it happened, he got everything right (including the enemy leadership), and Cannae was his best-ever battle.

I suspect it was more like a medieval mastermason than a modern civil engineer.  He knew his materials and their properties by experience, not by scientific experiment. 

Having disposed of the canard of high precision, what, if any, general conclusions can be drawn about the length of combat in the ancient and medieval periods?

[Patrick Waterson]

I suspect it was more like a medieval mastermason than a modern civil engineer.  He knew his materials and their properties by experience, not by scientific experiment. 

While I can see and appreciate the thinking, I am intrigued by the assumed distinction between experiment and experience: what would be the practical difference in applicable knowledge, if any?

Having disposed of the canard of high precision, what, if any, general conclusions can be drawn about the length of combat in the ancient and medieval periods?

I shall take that up if we ever dispose of the 'canard'.    Hannibal was able to execute manoeuvres with precision and to assess how long it would take for certain troop types to overcome other troop types.  Familiarity was, as correctly observed, an integral part of the ability to make such assessments, though Hannibal was not the only one to be able to do so: Alexander delighted in knifing through opposing formations believed to be tough enough to stop him.  His handling of Gaugamela was a remarkable balancing act, as with about 600 cavalry he kept 20,000 enemy cavalry occupied, and then routed them with another 300 of his own cavalry in the right place.  If that is not high precision, I do not know what is.

[Erpingham]

I suspect it was more like a medieval mastermason than a modern civil engineer.  He knew his materials and their properties by experience, not by scientific experiment. 

While I can see and appreciate the thinking, I am intrigued by the assumed distinction between experiment and experience: what would be the practical difference in applicable knowledge, if any?

Interesting question.  How you get to proficiency would be different - theoretical learning, academic as against hands on.  And you could become proficient quicker.  Also, I suspect that innovation would be easier/ more likely.  I leave it to you to apply that to a battlefield, though I would draw the parallel to the staff college model developed in the 19th century against what came before.