Dunkirk: they saved the soldiers and left the army

Ten days after the German offensive opened on 10 May 1940, the shape of the campaign was settled. The armoured divisions came through the Ardennes, ran west and reached the Channel on 20 May. The Anglo-French main force in Belgium and northern France was cut off from behind.

The overlap of dates is the thing to hold onto. On that same 20 May, in the tunnels under Dover Castle, Admiral Ramsay convened the first meeting about how an evacuation would be run¹².

No evacuation had been decided. A plan to break south and join the French was still alive. Which means the procedure for losing was being built in parallel with a plan for winning that had not yet been abandoned.

The eleven cases on this site so far are all either the design of a winner or the structural failure of a loser. This is the only one where the losing side is the protagonist and its design worked.

Several hundred thousand British and French troops were inside the pocket. The defensive perimeter ran from Gravelines through Bergues to Nieuport, using the canals and waterways. It was not a line for winning. It was a line for buying time.

There were two exits: the East Mole at Dunkirk harbour, and the beaches running east from it.

The beaches are shallow; destroyers and large ships cannot come alongside. Men had to be ferried out by small craft, which caps the number per hour. 98,671 left that way.

The East Mole was a breakwater built to protect the harbour water, not a pier. But it had depth alongside. Planking was laid, men were queued in single file, and destroyers came alongside to load them directly. 239,555 left that way — more than two thirds of the total¹².

On the beaches, what was not coming home stayed. In places lorries were driven into the sea at low tide and lashed together into an improvised jetty. The equipment they had decided to abandon became the structure that carried them off.

Read Dunkirk as a miraculous escape and the centre of it disappears. They did not escape. They chose what to lose.

Home came 338,226 men. Left behind were 2,472 guns, about 64,000 vehicles, 445 tanks and 416,000 tons of stores. Britain lost almost everything it would have needed to defend itself on the ground. In the summer of 1940 there was no equipment with which to meet an invader in a straight fight.

The trade works anyway, because one side of it can be remade and the other cannot. Guns and vehicles and tanks come out of factories. Trained soldiers — above all the NCOs and officers — do not. What Britain carried home were the people who would train the army it had not built yet.

Here is the twelfth type: a design of what to lose — when you are losing, abandon what can be remade and carry home what cannot.

Do not confuse it with supply at Vicksburg or convergence at Waterloo. Those are designs for winning. What is designed here is the procedure for losing, and the objective is not victory but being in a condition to play again.

On 24 May Rundstedt, commanding Army Group A, ordered the armour to halt, and Hitler endorsed it. The advance resumed on the 26th, and many units took longer still to move⁴. Those 2–3 days became the time in which the perimeter was built and the ships were gathered.

The explanation that Hitler deliberately let the British go is persistent and is not supported by the scholarship. The reasons given for the halt are more prosaic¹⁴.

The armoured divisions had run hundreds of kilometres in ten days and their serviceability had fallen. The ground in Flanders is cut by canals and waterways and is poor tank country. The infantry was still far behind. The flanks were stretched. "Conserve the tanks for the next operation" was, at that moment, a reasonable judgement.

Which is the point. These 2–3 days were not an Allied design. They were time that the other side's reasonable judgement happened to hand over. A design of what to lose cannot make time. It can only build the shape that is ready when time arrives.

There are two losers in this case.

One is Rundstedt. The halt order let the British get away, but as a judgement at the time it holds together. Tank serviceability, ground, stretched flanks, infantry still coming up — all of those constraints were real. What he got wrong here was not the judgement but the estimate of how long it could be afforded. At Normandy he held the equally correct rule of not committing the reserve until the main landing was identified. The same man, with the same signature, loses again four years later.

The other is Lord Halifax. In the War Cabinet of 26–28 May he argued for exploring terms through Italian mediation³. The BEF was in a pocket, nobody knew how many would come out, and the equipment was certain to be lost. Seeking terms was not, in that moment, an absurd position.

A correct reading losing to a structure has appeared eleven times on this site. This one is different. Halifax's reading did not lose to a structure; it lost to another reading. And which was right can never be tested on the side of the option that was closed. That is a weakness of this type, not a strength.

You can choose how to lose because somebody who cannot is paying for it. In this case the bill fell mainly on the French army.

At Lille, about 40,000 men of the French First Army held 7 German divisions for 4 days before food and ammunition ran out. Those four days go directly into the completion of the perimeter and the start-up of Dynamo. The Germans accorded them the honours of war at the surrender¹⁴.

At Dunkirk itself, the French rearguard of about 40,000 men surrendered on 4 June. The last ships left because they stayed.

Of the 338,226 brought home, 139,997 were French. Most were returned to France and became prisoners weeks later at the surrender. The same ship, a different place to get off, a different outcome.

Anyone reading this as a well-managed withdrawal should not skip this section. Only the side that can post a rearguard gets to design its defeat.

Immediately after the evacuation Britain did not have the equipment for a serious ground battle at home. Had Germany been able to land in the summer of 1940, this case would read very differently. That it could not was a matter of air and sea, not a credit to Dunkirk.

But the equipment came back — from the factories and later from American supply. And the officers and NCOs who had been carried home became the people who trained the enormous army built after 1940. The difference between those two recovery times is the whole of this type.

The other result is in the alliance. The sense that the British had left first stayed with the French for a long time. The order in which people withdraw becomes the price of the relationship afterwards. A design of what to lose includes how it looks from your own side.

The popular memory is a miracle in which small civilian boats lifted an army off a beach.

The numbers say otherwise. 98,671 left from the beaches and 239,555 from the East Mole — more than two thirds from the harbour. The lifting was done mainly by destroyers and other large ships. The little ships ferried men from the sand out to those ships, which was indispensable and was not the main act¹².

Nor was it a miracle. Preparation began on 20 May, routes were surveyed, a breakwater was converted, and loading rates per hour were calculated. Most of what is called the miracle is procedure and clerical work.

So far this is the site's familiar substituted correction — the same shape as the rubber tanks at Normandy, where a photogenic cause takes the memory slot of the effective one. What is new here is not the kind but the question that survives the correction.

What was that story for? What Britain needed in the summer of 1940 was not equipment and not manpower but a reason that would hold up the decision not to stop. The telling of the miracle supplied exactly that. Churchill knew it, which is why the same speech carries "wars are not won by evacuations". A story can be inaccurate and still be doing real work.

So the correction does not end at the correction. After "that is not what happened", write "then what was that shape for". Stop at the debunking and the facts get more accurate while why they were forgotten stays invisible.

A: if the East Mole had been unusable. Bomb it down and only the beaches remain. The rate per hour falls by a large factor, and the number brought home is plausibly around a hundred thousand. Two thirds of this operation hung on one breakwater.

B: without the halt order. The time to build the perimeter disappears. This type cannot make time. It only prepares the shape; the time itself is handed over by the enemy or by circumstance.

C: if the War Cabinet had decided to explore terms. The meaning of the 338,226 men changes entirely. A design of what to lose is only that if it comes with the intention to play again; otherwise it is simply survival.

These are unverifiable thought experiments, not claims about what happened. Removing one element is how you see what it weighed.

Read Dunkirk as a miracle and the centre goes. They did not escape; they chose what to lose. And they could choose because the procedure had been written before the defeat was settled, because somebody rewrote the objective, and because other people stayed to the end.

That makes twelve. Austerlitz — space; Cannae — momentum; Midway — time; Trafalgar — command; Stalingrad — fixation; Agincourt — frontage; Waterloo — junction; Vicksburg — supply; Sekigahara — defection; Normandy — deception; Tet — the verdict. Dunkirk shows a design of what to lose: when losing, abandon what can be remade and carry home what cannot.

What separates this type from the other eleven is that the occasion for using it is certain to come. Few people will ever design space or deception. But a withdrawal comes to every organisation and every person. What is asked for then is not courage but whether the order of abandonment was decided in advance.