How continuous was combat?

[Dangun]

One should be wary of automatically correlating low casualties with short engagement times: if an army runs at first contact, it can suffer heavily from pursuit even in a brief battle.

That is not what I am arguing.

In-so-far as we have any facts - the known fact, from the sources for hoplite vs hoplite (Kerantz data) - is low casualties. Frustratingly, as you say, we have very few facts bearing on engagement times.

I am not arguing that short engagements cannot produce high casualties.
(Although... and I am sure you will find this distracting, but the Kerantz data suggests we don't have any high casualty battles in the hoplite vs hoplite sample, so what are we talking about? Your point about the rout is not a relevant criticism of the Kerantz data, although it certainly explains why other battles were bloodier.)

I am arguing that low casualties can only be produced by short engagements. I think I am essentially agreeing with you?

And I go back to the thought experiment: if 100 hoplites faced another 100 hoplites, how long would it take them to produce 19 (10%) casualties?

Let's have some guesses.
10 minutes? 20 minutes? An hour?
An hour sounds hardly credible because it implies 1 casualty every 3 minutes. 200 hoplites pushing, stabbing, hacking, punching for 3 minutes and only 1 person dies. I find it hard to believe.

So I agree with you Patrick, I think battles were short.

[Erpingham]

So I agree with you Patrick, I think battles were short.

Actually, I don't think Patrick said this.  I think he holds a similar position to others that some were short and some were long based on our sources.  While a short battle can be all "Wham, Bham, thank you mam", a longer battle must be something else.  What is that something else?  Is it spent largely in contact but fighting methodically/non-vigorously?  Is it spent only partly in contact but fighting harder when you are?  Some hybrid between?

[Patrick Waterson]

I am arguing that low casualties can only be produced by short engagements. I think I am essentially agreeing with you?

Nearly, but not quite.   Low casualties can also be produced by spending a long time trying to make holes in well-protected opponents, which is what I am partially arguing in respect of hoplites - only partially, as there were cases when hoplites, caught from angles other than frontal, went down like ninepins, and when they went into what we might call othismos casualties would drop to practically nothing until the push was resolved.

Krentz' data is useful, but like all data a simplistic and unidirectional approach wastes most of what it can tell us.  I know that looking at the circumstances of each battle in turn can be really boring, but it is the only way to establish how the casualties came about, and to whom.

Take First Mantinea (418 BC).  Krentz comes up with the winners losing 3.3% of their force (300) and the losers 13.8% (1,100).  This allows the utterly misleading conclusion that on about a 1,000-man front (slightly more for the Spartans) each winning hoplite killed one opponent and each losing hoplite 0.3 of an opponent and the compilation of a lethality table based on this misrepresentation.  Statistics is something military historians need to avoid along with lies and damned lies, or better, to use very carefully and in the context of as full an understanding of the armies studied as possible.

The reality of the First Battle of Mantinea is that practically all the Spartan losses occurred in two contingents: the Sciritae and the Brasideans, each 600-700 strong, isolated on the Spartan left and facing a superior Argive right; the Argives' own left and centre collapsed or fled before they could inflict any meaningful casualties.  So what we really have is that 1,000 picked Argives plus maybe 2,000 less exalted hoplites (say 3,000 in all) downed 300 opponents in a few minutes and sent the remainder running back to their camp.  Given that only 1/8 of 3,000 would have been in contact, i.e. 375 or so, each average front-rank Argive finished off just under one Spartan in a few minutes - and would probably have done more if the remainder had not fled.

This of course was not the end of the battle: as the Argives returned from their successful encounter, the Spartans, who had seen off the rest of the Argive allies, slammed them in flank and helped to round off the day's lopsided casualty figures.

First Mantinea was a day of rapid encounters with flanking giving swift results.  Delium (424 BC) six years previously was more of a mixed battle.  Krentz, leaving aside Diodorus' dubious description, concentrates on Thucydides' account, which gives each side around 7,000 hoplites and gives a winner loss of 7% (500) and a loser loss of 14% (1,000).  What he does not note is that the winner's loss was concentrated mainly in a single contingent:

"The Boeotian left, as far as the center, was worsted by the Athenians. The Thespians in that part of the field suffered most severely. The troops alongside them having given way, they were surrounded in a narrow space and cut down fighting hand to hand; some of the Athenians also fell into confusion in surrounding the enemy and mistook and so killed each other." - Thucydides IV.96.3

The Boeotian left and centre were being thrust back and/or nibbled away, but meanwhile the 25-deep Theban right was advancing against their 8-deep Athenian opponents.

"... the right, where the Thebans were, got the better of the Athenians and shoved them further and further back, though gradually at first." - idem, 4

This progress by shoving does not appear to have been productive of many casualties - yet.  Then things went seriously wrong for the Athenians, and they really had only themselves to blame.

"It so happened also that Pagondas [the Theban commander], seeing the distress of his left, had sent two squadrons of horse, where they could not be seen, round the hill, and their sudden appearance struck a panic into the victorious wing of the Athenians, who thought that it was another army coming against them. [6] At length in both parts of the field, disturbed by this panic, and with their line broken by the advancing Thebans, the whole Athenian army took to flight." - idem, 5-6

Pursuit time!  Now the Thebans could settle down to some really serious killing, but ...

"Some [Athenians] made for Delium and the sea, some for Oropus, others for Mount Parnes, or wherever they had hopes of safety, [8] pursued and cut down by the Boeotians, and in particular by the cavalry, composed partly of Boeotians and partly of Locrians, who had come up just as the rout began. Night however coming on to interrupt the pursuit, the mass of the fugitives escaped more easily than they would otherwise have done." - idem 7-8

Delium was another of these battles when night stopped play, or rather ameliorated the loser's casualty level.  The winner's score thus topped out at around a thousand, but could have been significantly higher with more daylight, given that for once a Greek army had a substantial pursuing force of cavalry.  The battle had begin 'late in the day' (IV.93.1) so was not an all-day affair, and our best clue to duration is that the Theban right had managed to shove back the Athenian left 'slowly at first' and was still going while the Athenian right and centre were emerging victorious over outflanked opponents of broadly similar depth, i.e. there had been no large-scale rapid breaking and running as at First Mantinea. 

And I go back to the thought experiment: if 100 hoplites faced another 100 hoplites, how long would it take them to produce 19 (10%) casualties?

As I hope to have explained, or at least expounded, there is more to it than that.  Ordered hoplites facing ordered hoplites will have hardly any losses because both sides are well protected and fresh, so while there will be a number of wounds inflicted and spears broken, any more than a handful of deaths would be unusual at this stage.  Our 100 hoplites will probably be deployed 10 x 10, so after a few minutes of ten men thrusting with hard but not frantic strokes, 10 files of 10 men will be pushing against the other 10 files of 10 men and even the frequency of wounding will drop, probably to near zero.  Finally, one side will - or at least should - prevail, and the other will be tumbled back, losing cohesion and formation, and the successful pushers will resume use of weapons against foes who are now off-balance and/or deprived of mutual support.  This is when the casualty count will begin to rise, and one-sidedly, too.  The loser will probably crack soon, losing most of his front rankers while the rest drop their shields and run away (the difference in load between them and shielded pursuers allowing them to get away without much of a pursuit).

Let's have some guesses.
10 minutes? 20 minutes? An hour?
An hour sounds hardly credible because it implies 1 casualty every 3 minutes. 200 hoplites pushing, stabbing, hacking, punching for 3 minutes and only 1 person dies. I find it hard to believe.

It is possible between opponents of comparable strength and skill, but as noted above if no clear results emerged from the initial weapon play (which we term doratismos as a convenient label) both hoplite formations would tend to go for a shoving match (which we likewise term othismos) to decide the issue.  This is one reason hoplite battles tended to be shorter than their Roman counterparts; Romans seemed happy to go on all day with their weapon play (might we call this perhaps xiphismos?), changing lines when exhaustion or lack of success rendered the fighting line ineffective.  Hoplites, by contrast, used the weight and cohesion of their formation as a means of producing a result, and although we have no good timekeeping to tell us exactly how long they would keep this up for, it did allow them to stay in contact for rather longer than the doratismos stage without suffering many casualties; those would come when one side was overborne.

Anyway, that is my current understanding.  It seems to fit the patterns we get from Greek battles generally and Thucydides' accounts in particular.  I hope it illustrates that there is - or certainly should be - much more to battle and army technique analysis than statistical analysis of and derivations from the losses on each side.

[Dangun]

Statistics is something military historians need to avoid along with lies and damned lies, or better, to use very carefully and in the context of as full an understanding of the armies studied as possible.

Methodologically, I disagree. Its rare, but we are incredibly lucky when we can generate statistics. Sure, interpretation is required, but far less than for literary analysis for example.

Krentz' data is useful, but like all data a simplistic and unidirectional approach wastes most of what it can tell us.  I know that looking at the circumstances of each battle in turn can be really boring, but it is the only way to establish how the casualties came about, and to whom.

Its not that its boring. But unless you repeat the process for all of the battles and show there is something structurally wrong with Kerantz's data - all you have is anecdote. An exception is not convincing evidence. Sure reality admits exceptions, but a methodical and broad collection of data - like Kerantz - is far more persuasive.

Now I have not (and will not) repeat Kerantz's compilation to verify or refute, but that's what's required to convincingly show the conclusion is flawed.

Not a bad topic for a Slingshot article Patrick?

Finally, one side will - or at least should - prevail, and the other will be tumbled back, losing cohesion and formation, and the successful pushers will resume use of weapons against foes who are now off-balance and/or deprived of mutual support.  This is when the casualty count will begin to rise, and one-sidedly, too.

But this is exactly what Kerantz's source data says did not happen. The casualties were not one-sided. (Unless you've switched to non-Hoplite vs Hoplite.)

If you accept Kerantz data I think the implications are pretty clear.

I think its more interesting as to why some types of battles produce low casualties and no rout massacre e.g. hoplite vs hoplite (strongly suggesting short engagements) and some types of battles produce high casualties and rout massacres? If rout massacres are part of the prevailing culture, I am string motivated to fight longer.

[Mark G]

I think you guys should introduce hanson, if you want to focus on hoplite battles.

He is pretty convincing that the entire ethos if hoplites warfare was to get it over with as quickly as possible.  That the objective was not to kill the enemy army. But rather to have them acknowledge your hegemony, which facilitated a short , sharp, low casualty battle.

And add in that the spartans notoriously did not pursue, which defacto encourages the other side to put up a show and then scarper in safety.

All of which seems to be generally true, until the Peloponnesian war changed the dynamic entirely and saw the end of hoplite ascendency.

[Patrick Waterson]

Mark - yes and no.  It was night, not brevity and restraint, which put a stop the the Theban pursuit at Delium.  It was the inability of infantry with shields to catch infantry without shields that kept casualty rates down in an infantry pursuit.  I have seen no evidence of any tendency towards an "Oh, these are fellow Hellenes; we've won, let's go easy on them" attitude: quite the reverse.  Athens inflicted andropodismos - the slaughter of the entire male population - on the defeated Melians in 416 BC.  They almost did so to the Mytilenians (on Lesbos) in 427 BC (they reconsidered the following day and sent a reprieve).  The Spartans and Thebans slaughtered the entire garrison of Plataea when it surrendered in 427 BC.  Hence I think Hanson is quite wrong when he thinks the acknowledgement of hegemony was the principal aim: regrettably, blatant genocide was just as often on the agenda - and the 'democratic' cities were worst of all.  This is admittedly stating matters a bit starkly because not every victory was followed by some form of civil massacre; the mentions are in aid of making the basic point that Greeks were not particularly restrained or merciful with respect to other Greeks, so Hanson seems to be barking up the wrong tree with his short, sharp hegemony-focussed battle.

Hoplites incidentally remained in the ascendant well after the Peloponnesian War; in fact until the rise of Macedon in 356-336 BC.  This is why Darius III had so many hoplites in his armies: they were still the best available, at least outside Macedon.


Nicholas - I think you overrate statistics.  Statistics, properly handled, are useful for establishing apparent and sometimes real correlations: used for anything else they are a short road to overriding both observation and common sense.

Incidentally, I am referring to Peter Krentz, Casualties in Hoplite Battles.  Is this the 'Kerantz' you refer to, just to make sure we are talking about the same material?

Statistics is something military historians need to avoid along with lies and damned lies, or better, to use very carefully and in the context of as full an understanding of the armies studied as possible.

Methodologically, I disagree. Its rare, but we are incredibly lucky when we can generate statistics. Sure, interpretation is required, but far less than for literary analysis for example.

This reminds me of something attributed to the Mullah Nasrudin.

"O people!  Do you want knowledge without difficulties, truth without falsehood, attainment without effort, progress without sacrifice?*"

Its not that its boring. But unless you repeat the process for all of the battles and show there is something structurally wrong with Kerantz's data - all you have is anecdote. An exception is not convincing evidence. Sure reality admits exceptions, but a methodical and broad collection of data - like Kerantz - is far more persuasive.

I think this misses the point: Krantz' data is hopelessly generalised, and fails to deal with how casualties were actually incurred.  It is not that 'his data' is 'structurally wrong' but rather that it fails to tell us anything useful or even usable about hoplite warfare.  He puts together the 'what' and completely fails to address the 'how' and 'why'. 

Finally, one side will - or at least should - prevail, and the other will be tumbled back, losing cohesion and formation, and the successful pushers will resume use of weapons against foes who are now off-balance and/or deprived of mutual support.  This is when the casualty count will begin to rise, and one-sidedly, too.

But this is exactly what Kerantz's source data says did not happen. The casualties were not one-sided.

They were pretty one-sided once an army started crumbling, and we were discussing 100 hoplites vs 100 hoplites at this particular juncture, not the vagaries of success and disaster when army meets army.

I think its more interesting as to why some types of battles produce low casualties and no rout massacre e.g. hoplite vs hoplite (strongly suggesting short engagements) ...

Or rather strongly suggesting an inability to mount an effective pursuit, as pursuit usually generates the marked inequality of casualties between winner and loser through producing the majority of casualties for the latter.  If we are to conclude anything from Krantz' 'data' it would seem to be that pursuit was rare and, when it occurred, incomplete.

I think we all agree that hoplite battles were generally shorter than Roman battles, and not just because armies in Greece were usually smaller.  Hoplites had a way of shortening battles, and it seems that the major contributors to this relative brevity were:

1) Habitual outflanking of the opponent's left (often on account of the right hand drift Thucydides describes in his account of First Mantinea).
2) Marked differences in troop quality at some points in the line (usually on or near the right).
3) The use of coordinated pressure to decide the action once it became apparent that weapon play by the front ranks could otherwise go on all day on account of no clear superiority having emerged.

Concerning exactly how long the face-to-face combat portion lasted, the major time contributor would be how long it took to establish a superiority in othismos, and also how long othismos continued to be contested by the losing side once a superiority had been established.  I know of no battles where 'the shove' resulted in neither side succeeding, but it would be interesting if there was one.  In any event, fit men evenly matched might keep this up for a considerable length of time, leading to an Epaminondas calling: "Give me one more step for victory!" or words to that effect.

*When everyone said 'Yes!' Nasrudin replied: "I only wanted to know. You may rely on me to tell you all about it if ever I discover any such thing."

[Erpingham]

Just an observation, but since we have done hoplite battles to death (I think we have at least three other threads on them) perhaps time to move on.  This will also stop my teeth grating as patrick reruns his othismos theory   Anywhere else with a good record of battles over various lengths between heavy infantry (preferably not Republican Rome)?

[Mark G]

Pat, both of your examples if massacres are from the Peloponnesian period or later.

[RichT]

Yes agreed Anthony, hoplites are really a red herring in this context, the context of the thread being originally to discuss discontinuous combat ie (presumably) the dynamic standoff model of Roman (and their opponents) heavy infantry combat.

If all this is going to be is yet another thread where one person grinds on about the hoplite scrum theory ('othismos'), a 20th C invention for which there is no ancient evidence whatsoever (so there), then been there, done that, boring.

Peter Krentz's name seems to be causing a lot of difficulties. It is Krentz, not Kerantz, not Krantz.

My one observation about casualty statistics (and yes statistics are useful and widely used by historians - like any tool they must be used properly and can be abused, but getting an overall idea of average losses in battles is valuable) - is that the important figure is the losses suffered by the victor, rather than the loser (since it's not possible to separate out pursuit losses from combat losses for the loser). The finding from this is that losses are low which suggests (doesn't prove, isn't on its own conclusive evidence for, but suggests) short engagement times - I think everybody is in agreement on this for hoplite battles, so perhaps they can be laid to rest.

So for Roman battles, or others:

Anywhere else with a good record of battles over various lengths between heavy infantry (preferably not Republican Rome)

Well no, the problem is AFAIK there is no record at all of the duration of combats, and only a few rather vague hints about the duration of battles - for which we need to avoid assuming that the duration of the battle and the duration of the heavy infantry combat are the same (which was the point of my Waterloo analogy of a few pages back).

Since the multi hour melee (genuine, actual fighting, in toe to toe combat, multi (more than two?) hour, melee, and leaving aside 'line relief', since non-Romans presumably didn't do it) is being proposed as a viable theory by Patrick, I would extend Nicholas' 100 hoplite thought experiment - imagine two 8 rank deep lines fighting each other, and causing, over the course of the combat, 50% casualties each (including killed and significantly wounded) - much higher casualties than are ever attested, but let's take a worst case, and assume there were lots of unrecorded wounded. Now if the battle lasted 1 hour 20 minutes (to keep the sums simple), then each file leader (whoever is file leader at the time) is scoring an effective hit (killing or wounding) once every 10 minutes. If the combat goes on longer, then hits are naturally less frequent - in a two hour melee, there would be one hit every 15 minutes, three hours, once every 22 minutes, and so on. Now while this sort of fighting would doubtless be very energy efficient, it bears no relation to any of the fighting I have ever seen or imagined (in combat sports, re-enactments, riots, TV and film reconstructions) and (more to the point) it doesn't sound at all like the descriptions of combat that exist, which all sound more violent, continuous, vigorous and dangerous than one hit every ten minutes or more would suggest. This is the fundamental problem (aside from any other considerations) I have with multi hour melees.

[Patrick Waterson]

Yes, Mark, but so are Hanson's conclusions.

Anthony, please feel free to move the thread on as I have been rather hogging it so far.  The Late Roman Empire gives us a few potentially illuminating battles courtesy of Ammianus Marcellinus; Dark Ages warfare provides plentiful battles but very little description; what about the later Mediaeval era?

Before we move on, a last word (or exposition) from Vegetius, as promised.  This is from his Book III, about giving battle.  We can guess from the following how continuous combat was considered to be.

The wedge (cuneus) is a disposition of a body of infantry widening gradually towards the base and terminating in a point towards the front. It pierces the enemy's line by a multitude of darts directed to one particular place. The soldiers call it the swine's head (caput porcinum). To oppose this disposition, they make use of another called the pincers, resembling the letter V, composed of a body of men in close order. It receives the wedge, enclosing it on both sides, and thereby prevents it from penetrating the line.

Once a triangular shape has slotted into an opposing triangular hole, how easy or hard is it going to be to achieve some sort of separation for recuperative purposes?

The serra is another disposition formed of resolute soldiers drawn up in a straight line advanced into the front against the enemy, to repair any disorder. The globus is a body of men separated from the line, to hover on every side and attack the enemy wherever they find opportunity. And against this is to be detached a stronger and more numerous globus.


We note the implication that the serra is used for running repairs on the battleline.

Above all, a general must never attempt to alter his dispositions or break his order of battle during the time of action, for such an alteration would immediately Occasion disorder and confusion which the enemy would not fail to improve to their advantage.

Generals had a very hands-on role in Roman armies: decisions whether to press on or break off 'tactically' or otherwise modify the fighting tempo were the province of the general, not his subordinates, unless those subordinates were detached, e.g. with a globus watching for a chance to strike.

[Patrick Waterson]

My one observation about casualty statistics ... is that the important figure is the losses suffered by the victor, rather than the loser (since it's not possible to separate out pursuit losses from combat losses for the loser).

A good observation.  I really mean it.

The finding from this is that losses are low which suggests (doesn't prove, isn't on its own conclusive evidence for, but suggests) short engagement times - I think everybody is in agreement on this for hoplite battles, so perhaps they can be laid to rest.

They do by and large give the impression of being brief compared to other cultures' combats, so yes.

Well no, the problem is AFAIK there is no record at all of the duration of combats, and only a few rather vague hints about the duration of battles - for which we need to avoid assuming that the duration of the battle and the duration of the heavy infantry combat are the same (which was the point of my Waterloo analogy of a few pages back).

But in the classical period the duration of the heavy infantry combat was the majority of the duration of the battle.  Romans and their neighbours customarily opened with a period of skirmishing (Greeks and Macedonians apparently did not), but this just warmed everyone up for the main event.  Once the lines closed, they stayed closed, at least according to our sources, and this is a huge difference from Napoleonic warfare, which is why Waterloo goes out of the window as an analogy.  Hence the duration of the battle and the duration of the heavy infantry combat were pretty much the same, give or take the time consumed in pre-battle speeches and the initial skirmishing phase - plus the pursuit, if any.

Since the multi hour melee (genuine, actual fighting, in toe to toe combat, multi (more than two?) hour, melee, and leaving aside 'line relief', since non-Romans presumably didn't do it) is being proposed as a viable theory by Patrick, I would extend Nicholas' 100 hoplite thought experiment - imagine two 8 rank deep lines fighting each other, and causing, over the course of the combat, 50% casualties each (including killed and significantly wounded) - much higher casualties than are ever attested, but let's take a worst case, and assume there were lots of unrecorded wounded. Now if the battle lasted 1 hour 20 minutes (to keep the sums simple), then each file leader (whoever is file leader at the time) is scoring an effective hit (killing or wounding) once every 10 minutes. If the combat goes on longer, then hits are naturally less frequent - in a two hour melee, there would be one hit every 15 minutes, three hours, once every 22 minutes, and so on. Now while this sort of fighting would doubtless be very energy efficient, it bears no relation to any of the fighting I have ever seen or imagined (in combat sports, re-enactments, riots, TV and film reconstructions) and (more to the point) it doesn't sound at all like the descriptions of combat that exist, which all sound more violent, continuous, vigorous and dangerous than one hit every ten minutes or more would suggest. This is the fundamental problem (aside from any other considerations) I have with multi hour melees.

Largely because my respected interlocutor turns a blind eye to the phalangitic press of files which takes over following the initial period of we-have-been-thrusting-away-for-some-minutes-and-getting-basically-nowhere, and for which there is plenty of evidence in our sources, even if Xenophon and Polybius refer specifically to file pressure or 'weight' only once each.

But I think we have agreed to move on from here.

[RichT]

Patrick:

A good observation.  I really mean it.

Why, thank you

I don't think the Vegetius passages you quote provide any illumination at all, to be honest.

I disagree with everything else too.

Once the lines closed, they stayed closed, at least according to our sources

Begs the question - this is precisely the point at issue.

Hence the duration of the battle and the duration of the heavy infantry combat were pretty much the same, give or take the time consumed in pre-battle speeches and the initial skirmishing phase - plus the pursuit, if any.

But I put it to you that:

1) the deployment, the speeches, the advance, the skirmishing, the different elements of fighting, the collapse and the pursuit, if included in the duration of the battle (and we can't be certain which are or aren't included), could well make up a very, very large proportion (I would say, the vast majority) of the duration of the battle.

2) the assumption that all the heavy infantry closed and engaged at once, all along the line, as one humungous block, like on the battle diagrams we know and love, is just that, an assumption, without any really firm grounding, given the extremely brief descriptions of battles we have. There could well have been - and for Romans, with their small subunits, there must have been - all sorts of details going on below the level of the typical ancient account (which often amounts to little more than "side A with 40,000 men closed with side B with 35,000 men, they fought for some time, then side B broke and fled"). These are big armies, often with complicated internal structures, and a lot of details of what they did to each other must be lost in the brevity of the accounts that we have.

Largely because my respected interlocutor turns a blind eye to the phalangitic press of files which takes over following the initial period of we-have-been-thrusting-away-for-some-minutes-and-getting-basically-nowhere, and for which there is plenty of evidence in our sources, even if Xenophon and Polybius refer specifically to file pressure or 'weight' only once each.

Well, no. But moving on - my reference to '8 deep' may have led you astray. My little example was intended to be about Romans or their enemies, not phalangitic files. So imagine them if you will armed (and fighting) with swords and big oblong shields throughout. My point stands - is the strike rate described at all plausible?

I agree that other examples from other periods would help enormously.

[Erpingham]

I've had a bit of a look through some accounts for medieval examples.  We are a bit limited by the fact that many battles were not just infantry fights in this period.  Here are a few examples which were (primarily - cavalry were involved at Rosebeke)

Battle of the Standard 1138 2 hours long
Battle of Neville's Cross 1346 3.5 hours long
Battle of Rosebeke 1382 1.5-2 hours long

Neville's cross is interesting, as I have suggested before, because it was a particularly long and exhausting battle in the eyes of the chroniclers.  In the words of the Anonimalle Chronicle, the two sides "fought strongly and relentlessly for a long time and two or three times rested by agreement and then fought again".  Several accounts mention the exhaustion and the need for breaks.

I've also had a glance at one or two saga accounts of battles (again usually all infantry affairs).  We have to be careful with these, as they represent history from heroic poetry perhaps padded with contemporary (13th century) military practice.  But if we look, say, at the account of Clontarf in Njal's saga in the light of the "Irish Pharsalus" account, we can see similarities, as attacks wax and wane.  A famous incident in the account is the "cursed banner" - two attacks break through to Earl Sigurd's banner and the standard bearer is killed both times.  Sigurd's men drive off the attacks but no-one will take the banner because it is cursed, so Sigurd takes it.  A third attack breaks through again, Sigurd is killed and this time his men rout.

[Patrick Waterson]

I disagree with everything else too.

Some people are just hard to please.

Once the lines closed, they stayed closed, at least according to our sources

Begs the question - this is precisely the point at issue.

So - would sir like to provide a few source quotes showing them periodically opening up again?

But I put it to you that:

1) the deployment, the speeches, the advance, the skirmishing, the different elements of fighting, the collapse and the pursuit, if included in the duration of the battle (and we can't be certain which are or aren't included), could well make up a very, very large proportion (I would say, the vast majority) of the duration of the battle.

Which requires armies to be very patient for the pre-battle phase and very diffident about the actual fighting: much talk, little action and swift departure - a bit like a committee meeting.  Frankly, if actions were anything like this I would expect at least some hints in our sources.

2) the assumption that all the heavy infantry closed and engaged at once, all along the line, as one humungous block, like on the battle diagrams we know and love, is just that, an assumption, without any really firm grounding, given the extremely brief descriptions of battles we have. There could well have been - and for Romans, with their small subunits, there must have been - all sorts of details going on below the level of the typical ancient account (which often amounts to little more than "side A with 40,000 men closed with side B with 35,000 men, they fought for some time, then side B broke and fled"). These are big armies, often with complicated internal structures, and a lot of details of what they did to each other must be lost in the brevity of the accounts that we have.

If this is a serious proposal that such armies historically closed partially and piecemeal, I am going to have to ask for some source evidence.

But moving on - my reference to '8 deep' may have led you astray. My little example was intended to be about Romans or their enemies, not phalangitic files. So imagine them if you will armed (and fighting) with swords and big oblong shields throughout. My point stands - is the strike rate described at all plausible?

Caught me there.   However, when (or rather if, as I believe Anthony is not keen) we come to look at Republican Roman battles, we shall tend to find ourselves looking at somewhat different casualty totals and hence one would assume different loss rates.

[Patrick Waterson]

I've had a bit of a look through some accounts for medieval examples.  We are a bit limited by the fact that many battles were not just infantry fights in this period.  Here are a few examples which were (primarily - cavalry were involved at Rosebeke)

Battle of the Standard 1138 2 hours long
Battle of Neville's Cross 1346 3.5 hours long
Battle of Rosebeke 1382 1.5-2 hours long

Neville's cross is interesting, as I have suggested before, because it was a particularly long and exhausting battle in the eyes of the chroniclers.  In the words of the Anonimalle Chronicle, the two sides "fought strongly and relentlessly for a long time and two or three times rested by agreement and then fought again".  Several accounts mention the exhaustion and the need for breaks.

'Resting by agreement' at Neville's Cross is intriguing: given a 3.5 hour battle and 'two or three' breaks suggests that each session of fighting 'strongly and relentlessly' would have lasted - or perhaps averaged, as both sides would have become less fresh as the day wore on - about an hour.

This might be a handy rule of thumb for mediaeval infantry, which lacked Roman-style endurance training (or even Greek gymnastics) but had the underlying constant of stamina-demanding everyday rural activities - farming, for the most part.  We can hold the 'average hour' up against other mediaeval battle accounts and see how it looks.

That said, I am not sure how far the desire of both sides to stand on the defensive may have influenced their amenability to stop and chat about having an intermission, but it does at least give us some basic timing to try out for this particular culture and period.