Sorry, but what you are seeing here is most likely statistical variation.
You need to run a _lot_ of battles to be able to reliably discern differences
of not much more than one percent. I reran them and got the opposite
results.
> The cause here is that the peasant in this case became the stack leader.
> He attacks, but cannot be attacked, so the stack's attack factor is
> effectively halved.
That makes no sense.
Carl Edman