Saturday, 7 January 2012


It is a known fact in warfare that units, slowly, lose capability. This might be through loss of cohesion, loss of command capability, dispersal across terrain or casualties. However it happens, we have to find some way of modelling it.

We immediately have a wargaming problem with cohesion. Our little men are stuck to bases, and it is hard to show that they are wavering, disrupted, reforming or whatever. Our symbols (for that is what our toy soldiers are) cannot take all the attitudes of the originals they represent.

In a sense, this is inevitable. Even if we consider that our miniature battles are, in some sense, a model or simulation of the real thing, a model is never exactly the real thing. If it were, then it would be a duplicate of the real battle, and I’m not sure, however much we may appreciate re-enactment that a re-run of say, the Battle of Kursk would go down too well.

This, then, is one of the choices we have to make when writing wargame rules or playing with toy soldiers. Presuming that attrition is sufficiently important for us to model, how, within the limitations of our simulation and our symbols, do we manage it?

Over the years, a number of methods have been tried. The “Old School” wargames, which are coming back into vogue a bit these days, worked by treating each model as an individual. These individuals were shot and removed, or not, on an individual basis. Thus, at a glance, the general could see the status of a unit and its casualty level. Also included were officers, who could also become casualties and, if I recall the Charles Grant rules correctly, this had an effect on the morale of the unit.

This sort of ‘bang, you’re dead’ model is still quite widespread, I think. For example, while I’m not a 20th century wargamer, the games I’ve seen, particularly relating to tank warfare, seem to fit fairly well into this category. Tanks, in effect, duel, and the loser gets a puff of cotton wool to show the fact.

A slightly more sophisticated method of modelling attrition started, I think, with some of the early WRG rules. Here, a toy soldier was 20 ‘real’ men, and the casualties were worked out in ‘real’ men. Thus, if I had five musketeer toy soldiers shooting at you I could inflict 19 real men casualties on your five. You still had five figures to shoot back, but one of them was actually only one real man.

The upshot of this, of course, was a lot of book-keeping and, as I suspect I’ve mentioned before, an unrealistically high casualty rate. While some of the casualties could be rationalised as wounded, cowards running away and so on, comparison with real world battles still suggested that the rate was far too high. In early modern battles, at least, the bulk of the casualties occurred when one side or the other fled and was chased.

More recently, the DB* series removed the basis of attrition from the model almost entirely. A unit was a single base of toy soldiers and it either fought or fled. The only compromise with attrition at all was that units could be pushed back in combat, and suffered a penalty in the next round if that were the case. Units could be rallied from the rout in some circumstances, and could return to the fray as before.

Now, writing wargame rules is hard work, and it is not my intention to criticise anyone who has undertaken so to do. What I’ve described is a variety of ways of modelling attrition, each valid in its own way, but none of which capture the full extent of the concept. Indeed, given the nature of models, it is probably impossible to do so.

This is not to say that models cannot be improved, or that innovation is impossible. We can, by developing better and different models, try to obtain a better grip on the underlying reality, in this case, of a body of trained men under extreme stress.

What, then, might be a way forward? We need, it seems to me, some sort of double model. Units get both worn down by combat, and, under some circumstances, suddenly collapse. In fact, a number of the methods of modelling attrition had both models. By removing figures we show the direct effect of attrition, while the impact that has on morale (if unit strength is counted in morale, and it usually is) as well as other factors like being unsupported and outflanked can cause the unit to rout.

We could argue that having both methods included in the behaviour of units is accurate, and the best we can do, but those of you who have used such rules will be aware that constant calculating and recalculating of morale tests slows the game considerably. A single system, like DB* but perhaps a little more sophisticated, would be useful.

As you probably know, board wargames can use a ‘stepped’ approach, so a unit counter is flipped to show a degraded performance once it has been attrited a little. It is this sort of approach which we have adopted in the Polemos rules I have had a hand in. The problem is that it can reintroduce book keeping into the rules, which is a nuisance and can detract from the visual impact of the game.

My own solution, which I am slowly and painfully introducing, is to have casualty markers. For each level of attrition, (called ‘shaken-ness’), a marker is placed behind the unit until it is either rallied away or flees in rout. It is neat, simple and visually acceptable, even though the level of casualties depicted is probably still too high. Nevertheless, it works and can be rationalised by claiming that it shows the confusion that the rear areas of a unit can become when it does take casualties.

As I said, no model is perfect, but that is one way of modelling attrition without excessive amounts of paperwork.


  1. One of the difficulties is that attrition and morale seem to work in quite complicated ways. In the periods with which I'm most familiar (Napoleonics, WW2) then many units seem to have been able to take quite a large number of casuaties in certain circumstances (artillery bombardments, firefights) but were less able to take a much smaller number of casualties if delivered all at one moment, or if followed by being charged. The various rule sets in the Polemos system do seem to work reasonably well at modelling this however, compared to straightforward attritional models.


  2. I guess no model will be entirely accurate all the time (if it were it would be no model), so I suppose the aim is to try and extract or abstract the key elements in each period.

    Which raises an interesting question: did the key elements of morale or attrition vary over time? You might imagine that they didn't, but (say) Maya warriors might react differently to Napoleonic British. Further, I suppose Aztecs and Conquistadors should have entirely different systems.