Breakpoints in land battles

A comment by kesler12 asked me to provide some support for my claim that in actual battle units usually break after losing a minority of their strength (the context was that simulated combat where men do not fear for their lives tends to have much higher casualty rates and much more resolute defences).

I couldn't find the original summarising article or report that I would have preferred to supply and I sure won't even try to assemble the dozens if not hundreds of separate tiny bits that form the general picture, but I did find an old study with a statistic analysis that seems relevant:

Robert L. Helmbold, RAND Corporation, 1971

He tried in typical (then still new) operational research style to develop a formula for calculating when a force would 'break' and retreat. His hypotheses had to be tested, and he used a database of 1080 historical battles, and the author produced these diagrams:

As you can see, the winning defenders suffered up to 20% casualties in about 90% of the battles and 20-25% casualties in another about 5% of the battles. Meanwhile, the attacks - which faltered obviously - suffered less than 40% casualties in more than 90% of the battles.

Next, the same from 612 battles won by the attackers. Again, about 90% of the losing defenders suffered "only" up to about 40% casualties.Keep in mind that it's a rule of thumb that usually more casualties occur during pursuit than before the defence broke!

These were historical battles, many of which were rather unlike modern combat engagements, but human nature largely stays the same, I suppose. Either way, the results are so very striking in support of the assumption that defending forces yield after suffering less than 50% casualties that this should be accepted as true IMO.

What I remember from the old summarising article that I keep not finding (I suspect it's in Infantry Magazine 1990's, Armor magazine 1980's or 1990's or in Military Review somewhere, or maybe it's some RAND report.) is:
Many defences break after 10-20% casualties over a time span of hours or days. The most resolute "stubborn" defending forces suffered up to 40% casualties before retiring.
Other sources support more assumptions:
Only encircled forces or forces that couldn't flee (such as Japanese island defenders) went to 80% casualties and beyond without routing. Attacks stall after similar (lower) casualty rates as defences break, and they stall the earlier the more "veteran" and less "youthful" the units are. "Veteran" units tend to be more stubborn (less fragile) defenders (and less vigorous attackers), and so are forces with a higher share of married men (who tend to be less prone to give up).



  1. So we can safely say that the casualty threshold for a defending force is 20% in most cases, and rarely more than 25%. And that the casualty theshold for an attacking force is less than 40% in most cases. Exceeding that threshold will result in their disintegration as an effective force, and loss of the battle in question.

    'Many defences break after 10-20% casualties over a time span of hours or days. The most resolute "stubborn" defending forces suffered up to 40% casualties before retiring.'
    I came to a somewhat different conclusion. In battles where the defenders lost, they typically experienced 20-25% casualties during the breakthrough phase. If the attackers continued on with an exploitation phase, the defenders would then suffer additional losses which brought the total up to 40% (which is when the battle ended). Of course, defenders with low cohesion and low morale could break after experiencing only 5% casualties...

    'Other sources support more assumptions: Only encircled forces or forces that couldn't flee (such as Japanese island defenders) went to 80% casualties and beyond without routing.'
    I would hesitate to accept such a conclusion. Defenders who had already suffered heavy losses and had broken would almost certainly surrender, once they realised there was no possibility of escape. The exception would be soldiers in the grips of fanatacism, caused through their ideology (WW2 japanese), patriotism (WW2 russians), or political circumstances (late WW2 germans).

  2. Sven,

    the diagrams you provided are avarage losses of a unit (brigade/division?). This does not account for very uneven distribution of losses within this unit.

    20% losses of a division, the norm after 6 weeks ostfront in 1941, meant that the infantry regiments and the pionier units had a much higher loss reate, the artillery a much lower.

    Did the infantry units break?


    1. Don't make up stats, please.
      There were more than about three million German men on the Eastern Front 1941, and losses (KIA/MIA) in June/July were a "mere" 79k.

      The diagrams I showed prove as empirical evidence that it is uncommon for forces to keep resisting up to 70 or 80% casualties. They break much sooner. Even taking into account that losses are not distributed evenly.
      This is important because in training where fear is no factor the opposing force is much more stubborn usually. This leads to a systemic bias towards attrition and a neglect of (preparation for) the important pursuit phase as lessons from unrealistic training.
      THIS is what I'm writing about here; I don't care about the exact figures or distributions at all, since they vary anyway.

    2. "Don't make up stats, please.
      There were more than about three million German men on the Eastern Front 1941, and losses (KIA/MIA) in June/July were a "mere" 79k."

      Please, WIA are the largest share of "Blutige Verluste" and there were no 3 million German soldiers in major actions on the east front 1941, likely not more than 2 million.

      At the end of 1941 more than 900.000 Germans were KIA/WIA/MIA; to assume that these high losses occured later than June is more than debatable.

      The point you miss is that in a division only a few units decide whether the division breaks and these units usually suffer very high losses. To take the average of a very large unit (division) with a lot of "combat support" gives IMHO an misleading picture, to start persuit too early is BTW not a good idea.


    3. 3.2 million German troops involved in Barbarossa is the typical figures given in sources, with variance from 2.5 to 3.7 depending on which reserves are counted, army only or Luftwaffe as well and so on.

      I didn't make up the rather specific 79k KIA/MIA from the first two months, I merely forgot to add a link:

      The ratio of WIA to KIA was actually quite low:

      Feel free to back up your 20% figure.
      And I didn't miss anything about the unit level. Fine data is available for battle level only.

      Collapse doesn't necessarily originate in the most-pressed units. Back in 1940 Sedan A French division panicked, and the panic started among the artillery, not the suffering infantry.

      To do something "too" anything is no good idea. "Too" always means "wrong".
      Yet when one reads WW2 veterans' writings on how to do battle one sees that at least some insisted on building and keeping reserves for the exploit/pursuit phase. Someone who's focused on attrition in MILES- or AGDUS-supported simulated combat would not do such a thing. That's also a consequence of small training areas, of course.

  3. This is a rich topic for exploration; perhaps more important is when are forces "neutralized" (using the artillery definition) enabling friendly forces the option to bypass these units.

    A closely related question is how does casualty rate affect cohesion. A unit that suffers many casualties quickly seems more likely to break than a unit that takes the same casualties stretch over longer periods.

    The current fetish is to focus on destruction (men or material), but the destruction of morale is paramount.


    1. Breaking the opposing force (or classic "werfen"/pushing from a position) has a very different meaning in low density warfare on land than in the the frontline-dominated World Wars. Nowadays it's either about capturing territory (Bosnia/Ukraine/Syria slow positional conflicts) or about forcing the opposing force to move (and thus expose itself) against its will, as a setup for its destruction. Back in WW2 breaking a defence was often not so much about the defenders as about the exploitation against much greater forces (piercing and pushing out two divisions to encircle an army).

      Maybe I'll write more about pursuit and exploration in the context of low density campaigns in the future.

  4. Is there a relation between casualties handed out and casualties received? The diagram does not pair them.
    It is possible, that there is a judgement on how much damage is being inflicted and how much is received. At some point, the destruction received would outweigh the destruction projected.
    If this is the case, then loss of expensive items could significantly affect perception, while strikes without the ability to retaliate would be devastating. The data is from symmetric conficts, while current conflicts are more asymmetric.
    A very cheap militia on foot could make a huge case out of the monetary value of one IED destroyed enemy tank or one helicopter taken down. The believeable distribution of the story would help to withstand a feeling of otherwise helplessness, when confronting the enemy in one sided shooting engagements.
    If the ration between damage dealt and damage received is extremely one sided, then the willingness to abandon such a situation might happen earlier, than in a situation, where something could be done against the enemy. The very low casualties of Western forces in the Greater Middle East could have made it impossible for their opponents to sustain much of a defense anywhere.

    1. In support of this idea using the same conflict and its data are figures for combat stress losses amongst First Canadian Army in NW Europe. The troops were able to attack extremely well and could weather the defence extremely well, but if left in the line to simply be in contact with the enemy then their loss rates for morale-psychological reasons increase dramatically under German mortar and artillery H&I fire. Improvements to counter mortar were particularly important at reducing these stress casualties; patrolling seems to have helped, but only the people who went on patrols. The question of agency you present is a very important one that is commonly overlooked: "But what can we do about it" is also what drives things that SO here has written about like the provision of DMRs and GPMGs down to sections and 2x section (multiple/half-platoon) patrols rather than using proper fire and manouevre or accepting some harassing fire as part of doing business in COIN.

  5. In my opinion it depends heavily on the culture of the military and it´s society. The Imperial Japanese Army in China and also against the russians in open terrain did not break and retreat with much higher losses even when were was a possibility to retreat. Even than not, when it would have been wise to retreat. So i think you underestimate the importance of culture in this context and in warfare generally.

    Yours truly


  6. There are indeed many questions around breakpoints in battles, from where and when to how and why. What significant trends can we observe in certain contexts, let us say the Eastern Front in WWII and how to they compare to more recent conventional events?

    One of the reasons of the catastrophic losses of the Red Army in Operation Typhoon were according to Lev Lopukhovsky among others the rapid destruction of obvious army and front headquarters by air strikes and the lack of radios. Once command and communiciations were shattered the isolated units fought their own battles with little coordination. At that point of the war doctrine and leadership ability was additionally often poor.

    Powerful electronic attacks are just one of the elements which can facilitate breaking a defence or attack by reducing the ability to communicate and fight. Widely dispersed and highly mobile formations relying on long-range firepower to large degree are arguably more vulnerable to the loss of long range communications. The right approach to training with a high regard to 'Auftragstaktik', activity of mind, independent actions, frugal operations with large 'onboard' supplies and flexibility of command, basically lots of the classic stuff would gain in importance.

    On the other hand I'm pretty sure there were battles which were won because units fought on to victory helped by the fact that they didn't know of the local defeat of friendly units not too far away. Napoleon seemed to have been quite adept at winning such encounters...


    P.S: You might be generally interested in the 'Space communications: Past, Present, and Future', especially the Lunar Laser Communications Demostrator

  7. Anon above, with his point about C3I and particularly C&C nodes getting knocked out, speaks to part of what I'm saying now: is it about total loss rates or perceived loss rates? Do I have to inflict 60% losses on the infantry regiments of a division to make the artillerymen and logisticians panic, a'la the France 1940 example, or do I only have to kill enough men that when Pte. Billy looks to the left, he sees what's left of Pte. Tommy and when he looks to the right, he doesn't see L/Cpl Willy because of intervals, conforming to terrain, and the general notion known as the "empty battlefield." The figures you've posted are clear, and perhaps sufficient to achieve this effect. If sufficient confusion is imposed, one may be able to cascade local failures of resolve into a rout.

    I look forward to your writings on pursuit in the low density battlefield.

    1. The panic of Bulson wasn't about casualties. The artillerymen knew there were heavy fights since they themselves were firing rapidly. A rumour about a German tank breakthrough did spread and the troops routed.

      High casualty rates are rather breaking the leadership's resolve to stay on the battlefield. They initiate a retreat.