The Repository Stone Record

Arrow Slits Were Harder to Shoot Through Than Claimed

October 7, 2026 · Runtime 9:00 · Watch on YouTube

In 1863 Viollet-le-Duc shot a bow through a medieval arrow slit and wrote that its sides worked like a gunsight. Around 1980, two researchers ran the test again at White Castle in Wales, with real bows. The arrow snaked into the stone unless the archer stood back — and the loops meant to protect him proved extremely vulnerable to shots coming in.

Chapters

  1. 0:00 Two tests, a century apart
  2. 0:50 The belief: tiny target, wide arc
  3. 1:41 Viollet-le-Duc's 35 degrees
  4. 2:40 White Castle and its offset crosses
  5. 3:38 The tests of c. 1980
  6. 3:57 An arrow that snakes
  7. 4:47 View, not fire
  8. 5:20 The target that wasn't safe
  9. 5:36 Viollet-le-Duc against himself
  10. 6:24 A castle that never met a siege
  11. 7:52 How to read an arrow loop

Transcript

Two tests, a century apart 0:00

In eighteen sixty-three, the architect who restored Carcassonne published the result of a test he had run himself. He took a bow and shot through a medieval arrow slit. The sides of the slit, he wrote, did not get in the way. They worked like a gunsight, and made the shot surer than aiming in the open air.

More than a century later, two researchers took real bows to the arrow slits of White Castle, in Wales, and ran the test again. They did not get his answer.

The arrow loop is the narrow slit that every castle wall seems to have. The story told about it has two halves: a tiny target for the enemy, and a wide field of fire for the archer inside. Both halves were tested at the same castle. Neither held up the way it is repeated.

The belief: tiny target, wide arc 0:50

Here is the belief, the way guidebooks put it. From outside, the slit is only a few centimetres wide, so an attacker can barely hit the man behind it. Inside, the wall opens into a wide recess, so the archer can stand, draw, and sweep a broad arc of ground.

The history behind it is tidy too. The arrow slit is ancient, but in western Europe it returns to military building only at the end of the twelfth century: at Dover and Framlingham in England, and at Richard the Lionheart's Chateau Gaillard, in Normandy.

Over the next century the slit grows arms. A horizontal cut across the middle turns it into a cross, and a notch at the foot opens the view downward. The cross is usually explained as a way to serve both the longbow and the crossbow from the same wall.

Viollet-le-Duc's 35 degrees 1:41

The shape got its most quoted numbers from Eugene Viollet-le-Duc. At the castle of Carcassonne, he wrote, the wall is thinned to seventy centimetres by the recess. The opening is splayed inside and very narrow outside, to command the ground beyond at an angle of thirty-five degrees.

Then came the test. He tried shooting through these slits, not with a crossbow, which he called as easy as with a musket, but with a bow. The sides of the slit, instead of hindering the shot, act as a sight, he wrote, and make it surer than aiming at something in the open air.

That sentence carries the weight of the man. Viollet-le-Duc restored Carcassonne, Pierrefonds, and Notre-Dame in Paris. His dictionary, ten volumes published between eighteen fifty-four and eighteen sixty-eight, drew the medieval castle as the nineteenth century came to imagine it. When he said the slit was a sight, the slit became a sight.

White Castle and its offset crosses 2:40

White Castle stands in the Monnow valley, in Monmouthshire. It was one of three castles that guarded the road from Wales to Hereford, held for long stretches by the Crown. Its round towers and gatehouse went up in the thirteenth century, around an inner ward ringed by a deep moat cut out of the rock.

Exactly when is disputed. The conventional date is the twelve fifties and sixties, when the castles belonged to the young Lord Edward. The historian Paul Remfry argues for earlier, in two campaigns between twelve twenty-nine and twelve thirty-nine. This episode does not choose between them.

Its arrow loops are unusual. The two arms of the cross are not level: one side sits higher than the other. One reading says that gave an attacker a smaller target. Another says it let the defenders keep an attacker in sight for longer, down the steep sides of the moat.

The tests of c. 1980 3:38

Around nineteen eighty, Peter Jones and Derek Renn tested those loops with real bows. They published in nineteen eighty-two, in the proceedings of the Chateau Gaillard colloquium, the standing European conference of castle studies, under a plain title: the military effectiveness of arrow loops.

An arrow that snakes 3:57

Here is the first problem. An arrow leaving a longbow does not fly straight at first. It flexes, and snakes from side to side. Through a slit a few centimetres wide, that is enough to strike the stone.

The experiments found that an archer could learn to avoid it, with practice. The easiest way was to stand back from the slit. And that is the catch. Viollet-le-Duc's thirty-five degrees is the angle the recess was cut to cover, measured from the slit itself.

An archer standing back is no longer standing where that angle begins. From one place behind the wall he sees the ground through a narrow gap, and has to move to aim anywhere else. What the recess promises and what one man can reach are two different things.

View, not fire 4:47

The cross does less than it looks, too. In Renn's words, cross-slits substantially improved the field of view, but gave little advantage in the field of fire, from any sort of bow. The arms help you see sideways. They barely help you shoot sideways.

And a loop that plunges steeply downward still leaves the foot of its own wall unseen. Even a cross-slit, Renn wrote, gave no view close up. The ground an attacker most needed to reach was the ground the loop could not watch.

The target that wasn't safe 5:20

The second problem is the half of the story that was supposed to protect the archer. The tests at White Castle found its displaced loops extremely vulnerable to shots coming in. The tiny target, tested, was not the safe one the story promises.

Viollet-le-Duc against himself 5:36

The strangest witness against the belief is Viollet-le-Duc himself. Two pages after his test, he describes the loops of the Laon gate at Coucy, built around twelve ten: an angle barely open, great depth, and no notch at the foot to widen the shot.

Those, he concludes, can only be considered as views over the outside, and as openings for light and air. The recesses have no benches, he adds, and wherever a watchman or a defender was posted, you find the stone bench.

And he gives the reason low loops were given up by the end of the fourteenth century. Mining had improved, and the long slits showed the enemy where the wall was weak. Dig a mine between two of them, he wrote, and you were almost sure to bring down a whole stretch.

A castle that never met a siege 6:24

So what did White Castle's loops ever face? In twelve sixty-two, Llywelyn ap Gruffudd attacked Abergavenny, a few miles away, and the three castles were readied. The garrison of its sister castle, Grosmont, was to be held by every man, and at whatever cost. The threat passed without incident.

In twelve sixty-seven the castles went to Edmund, Earl of Lancaster. Edward the First's conquest of Wales, in twelve eighty-two, took away most of their military purpose. By fifteen thirty-eight White Castle was disused, and in sixteen thirteen it was called ruinous and decayed.

As far as the record goes, those loops never met a siege. Every shot they are famous for was fired in a test, seven hundred years after they were cut.

So what were they for? Derek Renn put it in print, thirty years after the tests. Despite their quality, were the loops only decorative and psychological, threatening more than they could deliver? He left it as a question. So does this episode.

There is a reading on the table that does not need a siege. Historians of castle design note that features like arrow slits became associated with high social standing and authority. At Kenilworth, Renn observed, anyone coming from the town faced an array of loops before anything else.

How to read an arrow loop 7:52

So read an arrow loop like this. Measure the slit: most castle slits are at least two inches wide, enough for a fletched arrow to pass. In the great tower at Kenilworth some are barely half an inch. Look for a bench in the recess, and for the notch at the foot. Then ask what the loop faces: the approach, or the inside of the castle.

Viollet-le-Duc's test still counts for something. It proves the shot is possible. The White Castle tests prove it is not easy, and not wide. A loop can be a real weapon and a statement at once, and the wall alone will not tell you which it was on a given day.

One siege shows what shooting from a wall cost when it really happened. In twelve twenty-four, King Henry the Third spent eight weeks before Bedford, ordered forty-three thousand three hundred crossbow bolts, and lost, by one chronicler's count, seven knights and over two hundred men. Then he had the castle pulled down. Next time: what the ground at Bedford still keeps of that siege.

Description and sources

The belief: an arrow loop is a tiny target for the enemy and a wide field of fire for the archer inside. Its learned version has an author and a date — Viollet-le-Duc, Dictionnaire raisonné, vol. VI (1863): a 35° field at Carcassonne (p. 387), and his own test with a bow, in which the sides of the slit "act as a sight" (p. 392).

Two problems, both from the tests Peter Jones and Derek Renn ran at White Castle, Monmouthshire, around 1980 (published 1982). A longbow arrow flexes and snakes as it leaves the bow; to keep it off the stone, the archer did best standing back from the slit — away from the point the splay was cut from. Cross-slits improved the field of view, but gave little advantage in the field of fire (Renn, 2011–12). And White Castle's offset cross loops proved extremely vulnerable to incoming shots.

The strangest witness is Viollet-le-Duc himself: two pages after his test he calls the loops of Coucy's Laon gate mere views and openings for light and air, and explains that low slits were given up because they showed miners where a wall was weak. White Castle's own loops, as far as the record goes, never met a siege. The episode leaves Renn's question open: were they weapons, statements, or both?

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.

  • John Bointon — 2005 | Wikimedia Commons | CC BY 2.0
  • Eugène Viollet-le-Duc — 1863 | Wikimedia Commons | Public domain
  • Eugène Viollet-le-Duc — 1863 | Wikimedia Commons | Public domain
  • Adrian Farwell — 2013 | Wikimedia Commons | CC BY 3.0
  • Eugène Viollet-le-Duc — 1863 | Wikimedia Commons | Public domain
  • Richard Bernard Godfrey — 1784 | Wikimedia Commons | CC0
  • Quinn Norton — 2005 | Wikimedia Commons | CC BY 2.0

... and 8 more plates, each credited on screen in the video and listed in this project's manifest.

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

Every image is public-domain or CC BY material from Wikimedia Commons, credited on screen — nothing is AI-generated. The diagrams and set-type quotations are code: drawings of the argument, not measured surveys. The arrow loops shown at close range are at Corfe and Framlingham, not White Castle, and are labelled as such: no photograph of White Castle's own loops met this channel's licence.

#castles #medievalhistory #archery