Play Nine Men's Morris Online

The aim is not mills. It is leaving them nowhere to go

Nine Men's Morris has two ways to lose, and beginners only ever notice one of them. You lose if you are reduced to two stones — and you also lose if you still have stones but no legal move at all. The second is how a surprising number of games actually end.

That changes what mills are for. A mill takes a stone, which is useful. But a stone in the wrong place is worth more: every point you occupy is a point your opponent cannot slide into, and stones that cannot move are stones that might as well not exist.

Build the double mill

There is one structure that decides games, and it is worth learning by name. Arrange five stones so that a single stone can move back and forth between two adjacent points, completing a different mill at each end.

Now every move you make closes a mill and takes a stone. Move out: mill. Move back: mill. Your opponent cannot block it, because both destinations are already yours and the stone is simply commuting between them. Two or three turns of that and the game is over.

Which also tells you what to watch for. If your opponent has two mills-in-waiting that share an adjacent pair of points, you must occupy one of those points before they set it up. Afterwards is too late.

You cannot take a stone in a mill

When you close a mill you take an enemy stone — but not one that is itself part of a mill, unless every enemy stone is. This is the rule that gives mills their staying power, and it has a direct consequence: once your opponent has a mill, every loose stone of theirs is a liability, because it is the only thing you are allowed to take.

Played the other way, it says something useful about your own position. If all your stones sit in mills, none of them can be taken at all. That is a rare and very strong place to be, and it is worth steering towards in the endgame.

Flying, and why three is not as bad as it looks

Down to exactly three stones, you may move any stone to any empty point rather than to an adjacent one. That is a real reprieve: a player with three flying stones can reach anywhere on the board, which usually means they can break any mill their opponent is trying to close.

So being reduced to three is not the end. It is a different game — one where you have nothing to defend and complete freedom of movement, against an opponent who now has to close a mill against a defender who can appear anywhere.

Where players go wrong

Chasing mills in the placing phase. Nine stones is not many. Spending them on an early mill while your opponent takes the junctions leaves you with material and no mobility.

Taking whichever stone is convenient. When you close a mill, take the stone that is about to complete something, or the one that gives their pieces room. A capture that improves their position is worse than useless.

Ignoring the blocking loss. Count your opponent's legal moves as carefully as their stones. Reducing them to one or two moves is often faster than reducing them to three stones.

Breaking a mill without a plan. Opening a mill is only good if closing it again is guaranteed — which is precisely what the double mill arranges.

What was verified on this page

The rules were checked against an independently written implementation across 1,778 position-and-colour pairs covering all three phases — placing, sliding and flying. Every legal move matched, including the awkward one: which enemy stones may be captured. Flying was confirmed to begin at exactly three stones and not before.

The search reuses one board and undoes each move as it backtracks, which is where these engines quietly corrupt themselves. Applying and then undoing a move was checked to restore the board and both players' reserves exactly, across 2,667 moves.

One thing was missing and has been added: the game had no way to draw. Twenty self-played games at equal strength all ran to a 400-move cap without ending — two engines of the same depth simply shuffle stones back and forth, and nothing in the rules as implemented stopped them. The game now recognises threefold repetition: if the same position occurs three times, it is a draw. Re-running those twenty games, every one now ends, the longest after 51 moves.

The difficulty settings were played head to head, and the honest result is that they do not all separate.

Medium beats Easy comfortably. But Hard and Medium finish level — the extra move of lookahead buys nothing measurable. That fits the game rather than indicting the engine: Morris is drawish enough that one more ply rarely changes the verdict.

A game older than almost everything

Morris boards have been found cut into the roofing slabs of an Egyptian temple at Kurna, and into the stonework of cathedrals across Europe where bored people sat through long services. It is among the oldest games still played in something like its original form.

It is also solved. Ralph Gasser proved in 1993, at ETH Zurich, that perfect play by both sides is a draw. Neither the first nor the second player can force a win from the empty board — which makes the missing draw rule above less of a footnote than it sounds: a drawn game needs some way to actually be declared drawn.

More games like this

If you like games where restricting your opponent beats capturing them, try these: