Holding a reserve

Why does a commander keep the freshest troops out of the fight until the battle is nearly lost?

There is something close to a scandal in the practice. Men are dying along the line while a commander's best formation stands in a wood a few miles back, doing nothing. If the fight is desperate, why is that strength not in it? Yet almost every doctrine ever written insists on it, and the commanders remembered as reckless are usually the ones who had nothing left rather than those who kept something back.

Run the counterfactual first, because it isolates the mechanism. Suppose a commander knew the whole battle in advance — every enemy move, every failure of his own. Would he hold a reserve? He would not. He would allocate every soldier at the outset to the place where that soldier does the most good, because he could compute it. The reserve does not exist because battles are hard. It exists because they are unknown.

So what does the uncommitted formation actually hold? Not combat power exactly — the same battalion in the line would be combat power. What it holds is the right to decide later, with information that does not exist yet, and the value of that right moves with how much the situation can still change.

Now notice what commitment does to it. Once the reserve goes in, it is in contact, its axis is chosen, its strength is being consumed, and it cannot be turned around at any speed that matters. Committing does not spend the force so much as spend the choice. An option becomes a bet on the reading of the battle you happened to hold at that moment.

Which is why the uses of a reserve are so revealing. Doctrine lists three: exploit an unexpected success, restore a situation where the line has been broken, and meet a threat that was not foreseen at all. All three are responses to surprise. The reserve is the part of an army shaped against the fog rather than against the enemy.

But hold the other side of the ledger up, because this is where the honest difficulty lives. Strength withheld is strength not applied. Every hour the reserve waits, the troops in contact fight with less than the army possesses, and some of them die of that shortfall. If the battle resolves while the reserve is still standing in its wood, the commander has paid the entire cost of holding it and bought nothing.

And the option expires, at no announced hour. Reserves need time and ground to move, so a formation committed too late arrives after the decision it was meant to influence. The two errors point in opposite directions: commit early and you have bet on an incomplete reading; commit late and you may hold a valid option past the moment it could be exercised.

So the judgement is whether the return from applying this force now exceeds the value of still being able to apply it later, given that later may not arrive. Both quantities are unobservable. That is precisely why this is the classic test of command — the fog of war operating not on the map but on a single irreversible decision, and no staff process resolves it.

One case, offered as an illustration of the difficulty rather than a verdict. In May 1940 Churchill asked the French commander Gamelin where the strategic reserve was and was told there was none; everything had gone forward into Belgium. A commander with nothing uncommitted has stopped commanding and started watching. Which explains a rule that otherwise looks fussy: the moment a reserve is committed, a new one is formed, even a scratch one scraped from a quiet sector. The point is not its quality. It is never to be in a position where the answer to the next surprise is nothing.

Calling the reserve an "option" is a modern framing, not the language commanders used, and it should not be pushed too hard. It captures the essential asymmetry — value under uncertainty, destroyed by irreversible commitment — but real reserves also do prosaic things the framing misses, like relieving exhausted units and providing depth against a breakthrough that has not happened yet.

Two further refinements. Reserves are layered by echelon: a company commander holds a platoon, a corps commander holds a division, and the higher the level the larger the reserve and the later its commitment, because the higher level's surprises take longer to develop. And the reserve is the natural complement to mission command rather than a competitor to it — delegated initiative lets subordinates absorb the deviations local to them, while the reserve is what the commander retains for those that exceed anyone's local capacity to absorb.

One caveat worth stating plainly: the rules of thumb about how much to hold back differ by army, era and mission, and I have deliberately quoted no fraction, because there is no universal one. There are also situations — a rigid timetable, or an attack where mass at the outset is what creates the breakthrough at all — where spreading strength across time forfeits the decision rather than preserving it.

• How uncertainty itself is generated and sustained in battle, in the collection's piece on the fog of war.
• Why delegating decisions downward changes what a commander needs to retain, in the piece on mission command.

Every term the collection defines is gathered in the glossary.