The Repository Stone Record

The Letter That Explains Why This Castle Was Never Finished

August 22, 2026 · Runtime 9:48 · Watch on YouTube

In February 1296, the king's own master of works wrote to the Exchequer to explain exactly why his castle at Beaumaris was over budget. Nine years earlier, the same architect had finished a different castle, Harlech, for close to half the price. A payroll and a letter, read together, explain something no single castle can: what it actually costs, in pounds and in people, to be afraid of the right threat at the right time — and what happens to a building project the moment that fear moves somewhere else.

Chapters

  1. 0:00 Two castles, one architect
  2. 1:10 Harlech: the one that worked
  3. 1:37 The 1286 payroll
  4. 3:00 Beaumaris: built for an emergency
  5. 3:58 The 27 February 1296 letter
  6. 5:52 Three wars, one treasury
  7. 7:10 The stone keeps its own account
  8. 8:44 Why an account roll can be trusted

Transcript

Two castles, one architect 0:00

Two castles. Same architect, same king, six years apart. One of them was finished, roughly on budget, in six years. The other cost almost twice as much, took more than three decades, and still is not finished. Not damaged. Not ruined by a siege. Unfinished, the way a sentence stops when somebody runs out of ink. The accounts for both survive. So does something rarer: a letter, from the middle of the second project, in which the architect himself explains, in writing, exactly why the money was running out.

In February of twelve ninety-six, the king's own master of works wrote to the Exchequer to explain why his castle at Beaumaris was over budget. Nine years earlier, the same man had finished a different castle, Harlech, for close to half the price. A payroll and a letter, read together, explain something no single castle can: what it actually costs, in pounds and in people, to be afraid of the right threat at the right time, and what happens to a building project the moment that fear moves somewhere else.

Harlech: the one that worked 1:10

Start with the one that worked. In twelve eighty-three, King Edward the First is finishing his conquest of Wales, and part of that conquest is a new ring of castles down the coast: Conwy, Caernarfon, and on a crag above the sea at Harlech, a fourth. The man in charge of the stonework is a mason from Savoy named James of St George, and Harlech is where the record starts giving him a title: Master of the King's Works in Wales.

The 1286 payroll 1:37

The crown paid for this in weekly wages, and the clerks wrote down who was paid. In the summer of twelve eighty-six, the payroll at Harlech names two hundred and twenty-seven masons, five hundred and forty-six labourers, one hundred and fifteen quarriers cutting stone out of the hillside, thirty smiths, and twenty-two carpenters. Nine hundred and forty people, on one headland, all at once, all named in a document that has survived seven hundred and forty years.

At that rate the crown was spending close to two hundred and forty pounds a month. The castle was, by any working definition, complete by the end of twelve eighty-nine. Total cost: eight thousand, one hundred and ninety pounds. On schedule, on a budget the Exchequer had already agreed to, and finished. This is the version of a medieval building project that goes right, and nobody makes a documentary about it, because nothing about it needs explaining.

Now the other one. Beaumaris, on the Isle of Anglesey, had actually been planned first, surveyed as early as twelve eighty-four, the year before Harlech's payroll starts naming St George by title. It was shelved. There was no money for it. For a decade, Beaumaris is a castle that exists only on paper.

Beaumaris: built for an emergency 3:00

What put it back on the table was not planning. It was an emergency. In the autumn of twelve ninety-four, a Welsh noble named Madog ap Llywelyn led a national revolt against English rule. Anglesey rose with it, and the king's own sheriff there was killed. Edward crushed the rebellion over the winter, reoccupied the island in April of twelve ninety-five, and restarted the shelved plan for Beaumaris immediately, not as a long-term project, but as an emergency one.

Emergency shows up in the accounts as speed, and speed shows up as cost. In its first summer, twelve ninety-five, the wage bill at Beaumaris alone ran to about two hundred and seventy pounds a week. Harlech, at its own peak, spent two hundred and forty pounds a month. Put those two numbers on the same scale and Beaumaris is burning through cash at something close to five times Harlech's rate, from the very first season.

The 27 February 1296 letter 3:58

Seven months into that spending, on the twenty-seventh of February, twelve ninety-six, a letter goes to the Treasury and the Barons of the Exchequer, at Westminster. It is signed by James of St George and a colleague named Walter of Winchester. It is not a progress report. It is an explanation, in the first person, of a project that is outrunning its funding. It is the only document quite like it to survive from the entire Welsh campaign.

They itemise what the work actually requires. Four hundred masons, cutters and layers together. Two thousand lesser workmen. Two hundred quarrymen. Thirty smiths. Carpenters, for the joists and the floorboards. And, to move the stone and the sea coal, one hundred carts, sixty wagons, and thirty boats. Set that beside Harlech's entire payroll of nine hundred and forty people, and the letter is describing a workforce several times larger, at the same king's expense, at the same time as everything else the crown is paying for.

And then the sentence that is the actual point of the letter. Quoting: the men's pay has been and still is very much in arrears, and we are having the greatest difficulty in keeping them because they simply have nothing to live on. That is not a boast about scale. That is the king's own master of works telling the king's own Treasury that the workforce is one missed payment away from walking off the site.

Here is what that letter is not. It is not a confession of incompetence. It is written by the same man who delivered Harlech on budget nine years earlier, the most experienced castle-builder in the king's service, at the peak of his career. If James of St George could not make Beaumaris's numbers work, the numbers were the problem. Not him.

Three wars, one treasury 5:52

So look at what else the crown was paying for in twelve ninety-six. Since twelve ninety-four, Edward has been fighting a war in Gascony, in the south of France. Within months of this letter, still in twelve ninety-six, he opens a second front in Scotland. Beaumaris is not the only claim on the Exchequer. It is the third.

Harlech never had that problem. In twelve eighty-three, Wales was the war: the only one, the whole treasury's whole attention. By the time Beaumaris needed money, Wales was formally finished business, and the money that would have built it was on a boat to Bordeaux, or on the road to the Scottish border. Same architect, same crown, completely different position in the queue.

And the queue is where Beaumaris stayed. Funding grows increasingly erratic after thirteen hundred, thins through the next reign, and effectively stops around thirteen thirty. By then Edward the First is dead, in thirteen oh-seven, and so is James of St George, in thirteen oh-nine. The south gatehouse and the six great towers of the inner ward never reach their planned height. The outer gate toward the town is barely begun. The royal apartments are never built at all.

The stone keeps its own account 7:10

You do not need an account roll to see the interruption. It is in the stone. The lower courses of Beaumaris's walls are limestone mixed with a dark green schist, quarried close by. Above a certain height, the schist disappears completely. Its use stops after a documented pause in the building work, in twelve ninety-eight, two years after the letter, and it never resumes. A castle can keep an account roll in its own masonry, if you know which layer to read.

And yet nobody calls Beaumaris a failure. The World Heritage listing, shared with the rest of Edward's Welsh castles, names it among the finest examples of late medieval military architecture in Europe. Historians of castle design rate its plan as the most technically sophisticated concentric layout the king's masons ever built, more resolved than Harlech, more resolved than Caernarfon. An account roll can tell you a project ran out of money. It cannot tell you the design failed. Those are two different questions, and they need two different documents.

There is a harder version of the same point. Beaumaris was designed to withstand a siege it never had to face. No army ever tested those half-finished towers, because after twelve ninety-five, nobody rose against English rule on Anglesey again. An unfinished castle that is never attacked looks, on the account roll, exactly like an overrun. It looks, on the ground, like a threat that stopped needing an answer.

Why an account roll can be trusted 8:44

Which is really what these two documents are for. A payroll and a letter are not written to flatter a king, the way a chronicle is. They are written for the crown's own auditors, to be checked against other numbers by people with no reason to be impressed. That is exactly why Harlech's finished account and Beaumaris's stalled one can both be trusted, seven hundred and thirty years later, to say what actually happened: one a story of money arriving on schedule, the other a story of money going to fight somewhere else. Neither one is spin. That is the entire point of an account roll.

Next time, the stone keeps a second kind of record, and you can learn to read it standing in front of the wall. Two towers at Beaumaris stop dead after only a few courses of stone, and the type of rock changes at exactly the height where the money did. That is not decoration. That is a document, built into the wall itself, and it dates precisely to twelve ninety-eight, the year the funding first broke down.

Description and sources

In the summer of 1286, the Exchequer's payroll at Harlech Castle names 940 people by trade — masons, labourers, quarriers, smiths, carpenters — and the castle they built was essentially complete within three years, for a total cost of £8,190. Nine years later, the same architect, James of St George, is building Beaumaris Castle on Anglesey, and it is a different story: nearly double the cost, and still unfinished more than three decades on.

On 27 February 1296, James of St George and Walter of Winchester write directly to the Treasury and the Barons of the Exchequer, itemising the workforce Beaumaris needs and admitting that wages are so far in arrears the men "simply have nothing to live on." It reads like a confession of failure. It isn't — it's written by the one man in England who had just proven he could deliver a castle on budget. The difference between the two projects isn't skill. It's that in 1283 Wales was the crown's only war, and by 1296 Beaumaris was competing for money against Gascony and a new front in Scotland.

The interruption is even visible in the masonry: a band of dark green schist in Beaumaris's lower walls stops appearing after a documented funding pause in 1298, and never returns. A payroll and a letter aren't written to flatter a king — they're written for the crown's own auditors. That's exactly why both documents, one recording success and one recording an overrun, can still be trusted 730 years later.

PRINT-READY, FROM THIS CHANNEL

How to Read a Castle, Vol. 01 — the method, the measured drawings and the documents behind the first two episodes

https://therepository.gumroad.com/l/read-a-castle

PRIMARY SOURCES

ARCHIVE PLATES — public-domain documents, with author, date, archive and licence.

  • British Library — 2020 | Wikimedia Commons | Public domain
  • Photochrom Print Collection — between 1890 and 1900 | Wikimedia Commons | Public domain
  • British Library — 2020 | Wikimedia Commons | Public domain
  • Photochrom Print Collection — between 1890 and 1900 | Wikimedia Commons | Public domain
  • Buck, Samuel, 1696-1779 Buck, Nathaniel, engraver. — 1742 | Wikimedia Commons | Public domain
  • Photochrom Print Collection — between 1890 and 1900 | Wikimedia Commons | Public domain

Stone Record reads castles as documents — what the surviving paperwork and the standing masonry actually say, against what gets repeated about them. Primary sources for this episode are linked above.

The historical images in this video are public-domain material: two 1742 engravings by Samuel and Nathaniel Buck, and a Library of Congress photochrom print of Harlech Castle from the 1890s-1900s. They are genuine period depictions of the real buildings, not illustrations drawn for the channel — this episode uses no generated artwork.

#castles #medievalhistory #beaumariscastle #wales #archaeology