A note before you begin. This piece deals with slavery, torture, and colonial and apartheid-era violence, and with a site whose descendant communities are living, local, and currently in public disagreement about how it should be remembered. Nothing here is sensationalised, but none of it is comfortable either.
The soldiers who guard the Castle at night have a route they prefer. Rather than pass through the archways in the small hours, they walk the long way round, on the outside of the building. One guard on the two-to-four shift described hearing someone screaming for help, and voices, from the direction of the dungeon and the torture chamber. He went to look. There was nobody there.
Hold that for a second, because it is the most important sentence in this article and almost nobody who repeats it stops to ask the obvious question.
Who is screaming?
Not what — who. The Castle of Good Hope has been standing in the middle of Cape Town since 1679, and the Castle of Good Hope ghosts are as well documented as any roster on the continent. Lady Anne Barnard, who lived at the Castle after arriving at the Cape in 1797 as the colonial secretary’s wife, is seen in a white dress, in the rooms where she entertained. People report her laughter. People report catching rose perfume, which was the scent she wore. She is said to appear when dignitaries visit, still doing the job she did in life. Governor Pieter Gysbert van Noodt condemned seven alleged ringleaders to death; contemporary records say four were ultimately hanged on 23 April 1729, and Noodt was found dead in his chair later that day. The curse at the gallows belongs to the legend rather than the administrative record. A tall figure was seen on the battlements in 1915 and again in 1947, over a period of weeks. The Lady in Grey runs through the building weeping, and members of the royal party, Princess Margaret among them, are said to have seen her during the 1947 tour.
Names. Dates. A dress. A scent. A death certificate’s worth of detail.
And then, from the Donkergat, the dark hole, the windowless chamber under the walls: screaming. That’s it. That’s the entire contribution to the Castle’s ghost literature from the people the Castle was actually built to control. No name. No face. No clothes. No story. A sound in the dark that some soldiers would rather walk around than walk past.
The Dutch East India Company built its Cape colony on enslaved labour, and enslaved people were among those forced to build this fortress between 1666 and 1679. Prisoners were interrogated in the adjoining torture chamber. Under Roman-Dutch criminal procedure, courts could authorise torture to obtain a judicial confession when the evidence had reached the required threshold but was considered insufficient for conviction. On any reasonable accounting of suffering, they are among the Castle’s dead.
They are not its ghosts.
I want to be careful here, because there is a lazy version of this observation and it isn’t the one I’m making. The lazy version says the ghost stories are racist. That is too easy, and it also isn’t quite true. Nobody sat down and decided who would be permitted to haunt. What happened is stranger and harder to argue with.
To become a ghost, you need a biography. You need someone to have written down what you looked like, what you wore, how you laughed, where you stood. Lady Anne Barnard is one of the Castle’s most vivid ghosts because she is also one of its best-documented residents: letters, journals, portraits, a room named after her. The people in the Donkergat left nothing, because nobody with a pen thought their interior lives were worth a line. Two centuries later, the haunting sorts itself along exactly the contours of the archive, and it does so without anyone intending it.
And here is the part that stopped me. A later account attributed to Lady Anne describes winter floodwater surging into the dungeon and convicts drowning while chained to the walls. I have not found that passage in a digitised primary edition of her letters, so it should be treated as a repeated tradition rather than a verified quotation.
So the woman who is the Castle’s most-seen ghost is the reason we know anything at all about the people who aren’t. The same act of writing that made her a character made them a footnote in her paragraph. There is no villain in that. There is just a record, and who was permitted to keep one.
The episode goes further into the archive than the article can — what the VOC actually wrote down about the people it enslaved, and what it didn’t. Follow The Dark Atlas on Spotify and Season 1 arrives as each location is mapped.
The Place
The first thing to understand about the Castle of Good Hope is that the sea has left it.
It was built on the shoreline of Table Bay, a five-pointed star fort with a moat, positioned so that its guns covered the water and its walls met anything that came ashore. Land reclamation in the 1930s and 1940s pushed the coast back to make room for the Foreshore and the harbour, and the Castle now sits inland, marooned in traffic. Its moat holds rainwater. On the seaward side there is a dual carriageway, a railway station, and the long grey rectangle of the Grand Parade.
The shape survives intact, and it is the most legible thing about the building. Five bastions, each with its own garrison, magazine, storerooms and workshops, named Leerdam, Buuren, Katzenellenbogen, Nassau and Oranje after the titles of the Dutch Prince of Orange. Walk the walls and the geometry does what geometry was designed to do in the seventeenth century: no attacker gets a blind approach, every face is covered by the fire of two others. The geometry is an argument about controlling space, and it was never once fired on from the sea.
Inside, it is quieter and stranger than the plan suggests. A defensive wall crosses the interior and divides it into two courtyards, and against that wall is the Kat balcony, a broad ceremonial staircase in Cape Dutch plaster. Announcements were made from it to soldiers, enslaved people and burghers. Sentences were read out there. Today the balcony leads into rooms holding the William Fehr Collection of paintings and furniture, administered by Iziko Museums, and there is a café, and there are school groups.
Below all that is the other Castle: prison cells, the adjoining torture chamber and the Donkergat, or Dark Hole. The Castle’s own visitor map treats the torture chamber and Dark Hole as connected but distinct areas. The Donkergat is a windowless chamber below the walls, remembered in the site’s folklore as a place of confinement and terror.
The other thing to know is that this is not a ruin, and it is not only a museum. It remains a military site administered by the Castle Control Board. Many of the most detailed modern ghost reports came from guards and other people working there at night.
The History
There was a fort here before there was a castle. Jan van Riebeeck arrived in 1652 to establish a replenishment station for Dutch East India Company ships on the run to the Indies, and threw up an earth-and-timber structure called the Fort de Goede Hoop, on ground that is now the Grand Parade, a few hundred yards from where the stone fortress went up later. That distinction is going to matter.
The first stone of the Castle was laid on 2 January 1666, and the work was finished in April 1679. The materials were local: granite cut from Signal Hill, blue slate and shells brought across from Robben Island. The labour was soldiers, volunteers, enslaved people, and Khoi being punished.
That last category deserves a moment, because it is doing a great deal of quiet work in a tourism bullet point. Building this fortress was a sentence. The wall you photograph was, for some of the people who raised it, the punishment itself.
Between 1652 and 1808, an estimated 63,000 enslaved people were imported to the Cape, mainly from the African continent, India, Madagascar and Indonesia. The traffic ran through VOC-sponsored voyages, return fleets, private sales and, after the British occupation, other channels. Some captives were held at the Castle; many others were confined in the Slave Lodge a few streets away, which is now also a museum.
And this building was where the colony’s violence was administered. The Donkergat — the Dark Hole — and the adjoining torture chamber are distinct spaces, though modern retellings often blur them together. Under Roman-Dutch criminal procedure, a judicial confession carried special evidentiary weight. When the evidence met a prescribed threshold but was still deemed insufficient for conviction, a court could authorise torture to obtain that confession. The violence was not merely an unofficial excess; it was built into the legal procedure.
People were executed in the courtyard. Restoration work has periodically opened up rooms that had been bricked shut, and architects who studied the fabric concluded there are probably more.
The Castle’s prisoners include names that never appear among the Castle of Good Hope ghosts. Cetshwayo kaMpande, king of the Zulu, was held here after his capture in the Ngome Forest in 1879. So was Sekhukhune, king of the Pedi. During the South African War of 1899 to 1902 part of the building was used as a prison, and those cells are still there. Through the apartheid decades it remained a military headquarters, and it is now the subject of academic work on how survivors of that period experience the site as somewhere their own trauma is stored.
Which brings us to Krotoa, and to the argument the Castle is currently having with itself.
Krotoa was Khoi, a niece of the interpreter Autshumao, and was taken into Jan van Riebeeck’s household at about ten or eleven to work as a servant and childminder. She learned Dutch and Portuguese and became an important interpreter and intermediary, including during negotiations after the first Dutch-Khoi war. Baptised as Eva in 1662, she married the Danish surgeon Pieter van Meerhof in 1664 and was widowed after he was killed on a slaving expedition to Madagascar. She was later imprisoned and banished to Robben Island. She died there on 29 July 1674, aged about thirty-two.
In August 2016, as part of the Castle’s 350th anniversary, the Department of Defence held a memorial service for her here. A wreath was laid, a bench was unveiled in her honour, and her descendants attended. The Department also commissioned statues of four African leaders who fought dispossession.
Two objections followed, and both are worth reporting without my deciding between them. The first is historical: the Castle was not finished until five years after Krotoa died in 1674 while the stone Castle was still unfinished. The church in which she was buried belonged to the earlier Fort de Goede Hoop on what is now the Grand Parade, and there is no firm evidence that she entered the completed Castle. The second came from Khoi community voices who objected that she was being memorialised at a site where she was imprisoned and broken. Against both, there is the position of the descendants who came, and of a defence ministry trying to convert a colonial fortress into something a democratic country can use. I am not the person to settle that. I can only note that it is live, and that it is being argued about by people with a great deal more standing in it than me.
The Legend: The Castle of Good Hope Ghosts, Named and Unnamed
The introduction laid the roster out. Now look at how it holds up.
Start with van Noodt, because the most-repeated story has accumulated contradictions. Some versions date the execution to 1728, but contemporary records place both the hangings and Noodt’s death on 23 April 1729. Seven alleged ringleaders were condemned; four were ultimately hanged. The famous curse from the scaffold appears in later retellings, not in the surviving administrative record.
It is a small discrepancy, but it tells you a lot. The story has been copied from page to page for decades, often without comparison to the surviving record. When a story travels that far without its contradictions being examined, that is worth knowing before you weigh what else it claims.
It gets worse in the recent retellings. There is now a widely circulated version supplying the exact words the condemned man shouted, and the expression on van Noodt’s face when he was found, and the conclusion of the doctors who examined him. None of it is sourced. All of it reads as though it were, and it will be in the next twenty articles about this place.
Then the bell. A soldier is said to have hanged himself in the bell tower with the bell rope, after which the tower was walled up, and the bell is still heard ringing on its own. A black dog is reported in the courtyard, approaching visitors and then not being there. A female skeleton uncovered at the old sally port during excavation work has been attached to the Lady in Grey, which is the closest thing in the Castle’s ghost literature to physical evidence of anything.
Now set the roster against itself, which is the thing nobody does.
The Castle of Good Hope ghosts who have names are: a governor of the Cape, and the wife of a colonial secretary. The ghosts who have a story but no name are: a soldier, a soldier, a woman in grey, a tall figure on the battlements, a dog. And the people who have no name, no face, no clothing and no story — who appear in the literature purely as noise from beneath the walls — are the enslaved and the imprisoned.
The correlation is not with how badly a person suffered here. Van Noodt did not suffer at all; he died comfortably in a chair. It is not with how long ago it happened, or how violent the death was. It tracks most closely with one thing: whether anyone wrote you down.
The woman at the top of the roster left volumes of letters and journals. The men in the Donkergat left a legal category. Two hundred years later, one of them has rose perfume and the others have a sound.
The Skeptics: What’s Really Behind the Castle of Good Hope Ghosts
Begin with who is making the reports, because it is not who you’d assume.
At most sites in this series the witnesses are paying customers on a night tour, primed by a guide with a commercial interest in something happening. Here they are soldiers. They are at work, on a rostered shift, in a building they know better than anyone, and they gain nothing by saying they heard screaming. Several have said plainly that they would rather not talk about it. That is a different evidential class, and any honest sceptical account has to start by conceding it.
Then it has to point out that the two-to-four shift is the single worst window in the twenty-four hours for a human nervous system to assess anything. Core temperature bottoms out. Vigilance collapses. Auditory thresholds shift, and the brain, starved of input in a dark, echoing, largely empty structure, starts filling the gap. Anyone who has done night work in an old building knows the sensation of hearing a voice in what turns out to be plumbing.
And this is an exceptional building for producing sound with no visible cause. Three-hundred-and-fifty-year-old stone, chambers cut below the waterline, vaulted rooms that carry noise between spaces with no line of sight, a moat holding water, and drainage of unknowable antiquity. Outside the walls, a dual carriageway, a railway line, and the wind that comes off Table Bay hard enough to have its own name. A windowless chamber under the ramparts is not a silent place. The city presses on all sides of it, and the stone carries everything.
The individual stories thin out badly on inspection.
Take the bell tower. Nobody names the soldier. Nobody dates it beyond a vague three centuries ago. A story that will produce no death record, no name and no year, but will produce a walled-up tower and an unexplained ringing, is a story built backwards from an architectural feature. Bells in exposed towers ring in high wind, which is the one thing Cape Town reliably supplies.
The Lady in Grey has bones behind her, which is more than the others can say. At a fortress with three and a half centuries of burials, executions and demolition, finding human remains is close to expected. The bones are real. The inference that they belong to a weeping woman seen running through corridors is not an inference at all. That is a story reaching for the nearest available object.
Lady Anne Barnard appears at functions. Functions happen at night, in period rooms, with wine, among guests who have all been told the story before they arrive, in a room named after her. If you wanted to design conditions for a group of people to produce a shared apparition, you would design that.
Van Noodt’s death is the clearest case. A forty-eight-year-old man died suddenly at his desk, which is unremarkable enough that it happens somewhere in the world every few minutes. What made it a story is that a hanging occurred the same morning. Nobody records the days when a governor was cursed by a condemned man and then lived another decade, because on those days nothing happened. And the year being wrong for three centuries tells you nobody in the chain of transmission ever went back to the record at all.
The corpus is also actively getting worse. Recent long-form retellings supply the dialogue and the forensic detail flagged earlier: none of it sourced, all of it written in the register of report. That material will be quoted as fact within a few years, because it already reads like fact. Whatever was originally true in these accounts is being buried under invention faster than anyone can dig it out.
So the sceptical account is strong. Bad hours, resonant stone, unfalsifiable stories, coincidence retained and non-coincidence discarded, and a literature contaminating itself in real time.
And it leaves one thing completely untouched.
Every explanation above accounts for why people at the Castle experience something. Not one of them accounts for the shape of what they experience. Fatigue and acoustics predict that a guard on a night shift will hear something inexplicable. They do not predict that when the same building produces a visual apparition, she will be a named eighteenth-century Scotswoman in a white dress, and when it produces a sound, the sound will be anonymous. Suggestion explains the phenomenon. It does not explain the casting.
That’s not a supernatural residue. It is a sociological one, and it is the part of this that no amount of debunking touches, because it isn’t a claim about ghosts at all.
My Thoughts
I don’t think anything walks at the Castle of Good Hope. I think the roster is an index, and what it indexes is the colonial archive with unusual precision.
If you wanted to know who at the Cape was considered a person with an interior life worth recording, you could throw away the Company’s papers and read the ghost stories instead, and you would arrive at almost the same list. The woman with the perfume and the piano. The governor with the death date. And beneath them, a great weight of human beings compressed into a noise.
That correspondence is too clean to be accidental, and once you see it you can’t unsee it in any haunted place. Ghosts are made of documentation. To come back, somebody has to have thought you were worth describing, and the people doing the describing owned the building.
Which means being haunted by somebody is, in a strange way, a form of remembering them. It is a rotten form, and Lady Anne Barnard is its proof: reduced to a dress and a scent, when the actual woman, who was a serious writer and a sharp observer of a society she was complicit in, would probably find her afterlife insulting. But it is more than the people in the Donkergat get. They are not remembered badly. They are not remembered at all, except as weather.
Here is where I stop being able to make it come out right.
I can’t fix this by naming them. I don’t have the names. The Company kept records, and there are names in those records, but a name in a ledger is not a biography, and turning one into the other means inventing a person. This series has already watched what happens when someone does that: at the Myrtles, a woman named Chloe was invented to give the plantation’s enslaved dead a face, and the invention didn’t rescue anybody. It replaced them. The made-up woman now occupies the space where the real ones should be, and she’s easier to sell.
So the choices at the Castle appear to be: leave the silence exactly where it is, or fill it with something I made up. And I’ve come round to thinking the silence is the more honest of the two — but I want to be clear that I’ve talked myself into that, and that it is a very comfortable conclusion for a person with no stake in it to reach. The people arguing about this building in Cape Town right now are arguing about whether silence is honest or just easier, and they are arguing about their own ancestors. I’m arguing about a research problem.
What I keep returning to is the question the guards raised without meaning to.
They hear someone calling for help. Every account repeats it, none of them pauses on it, and the reason nobody pauses is that the answer feels unavailable — the way weather is unavailable. But it isn’t a weather question. Somebody was in that room. The room is still there. The screaming, if it is anything at all, is the last remaining trace of a specific human being who had a name, a language, a family somewhere across the Indian Ocean, and a reason for being in a windowless chamber under a wall they may have helped build.
So the question isn’t whether the Castle of Good Hope ghosts are real, and it isn’t whether the guards are mistaken. I think they probably are.
The question is the one the story has been asking since the first soldier walked the long way round rather than pass the archway, and it has never once been put to anybody who could answer it, because there is nobody left who can.
Who was screaming?
Visitor Information
The Castle’s official address is at the corner of Castle and Darling Streets, beside the Grand Parade and a short walk from Cape Town station. It is central enough to reach on foot from much of the CBD; use normal city precautions and check local transport arrangements if visiting near closing time.
The Castle’s official visitor page currently lists daily opening from 09:00 to 16:00, with the last ticket sale at 15:15, and reduced admission for pensioners, children and booked school groups. It also lists guided tours during the day. Check again before travelling: this is a functioning Defence Force facility as well as a heritage site, and ceremonies, parades and government functions can close parts of it at short notice. Published hours are not a guarantee of access to any particular room.
If you go for the reasons this article is about, go for the lower levels: the garrison cells and the Donkergat. Access varies and is sometimes only on a guided tour, so ask at the desk rather than assuming. The William Fehr Collection and the Kat balcony are the showpieces and worth the time, but they are the Castle’s account of itself, and the cells are the other one.
Two things to pair it with. The Slave Lodge on Adderley Street, a few minutes away, is where the Cape’s slave history is properly told, and it does what the Castle by its nature cannot. And the Grand Parade immediately outside is worth standing on for a moment: the original Fort de Goede Hoop was there, which makes it the actual ground of the Cape’s first decade, now a car park and a market.
Staying overnight in Cape Town? Compare places to stay in Cape Town. If you book through this link, Shin Stories may earn a small commission at no extra cost to you.
On conduct, one thing only. Serving personnel work here and descendant communities visit. Both are more likely to be present than at any other location in this series, and neither is part of the attraction.
Before You Go
The full record for this location is on the page you’re reading: the 1666-1679 construction and who was made to build it, Krotoa’s life and the dispute over how she is memorialised, the Donkergat and the legal machinery that required it, the van Noodt date that has been wrong in print for decades, the sources below, and the practical guide to visiting.
This is location ten of twenty in Season 1 of The Dark Atlas, the halfway mark. The episode takes a different route through this material: it stays underground longer, and it spends most of its time on a single question the article can only ask at the end.
Continue through the archive with the Mackenzie Poltergeist at Greyfriars and the truth behind the Myrtles Plantation’s “Chloe” story, or explore every mapped location in The Dark Atlas.
Sources
Books and Academic Works
Barnard, Lady Anne. The Cape Journals and The Cape Diaries (Van Riebeeck Society editions). Primary eyewitness material for life at the Castle. The dungeon-flood story is widely attributed to Barnard in later accounts, but I have not verified the claimed wording in a digitised primary edition.
Elphick, Richard. Khoikhoi and the Founding of White South Africa (Ravan Press, 1985). Standard scholarship on Khoi society, the early Cape and Krotoa’s position within it.
Leibbrandt, H.C.V. Precis of the Archives of the Cape of Good Hope, Journals 1662-1670 and 1671-1674 (W.A. Richards & Sons, 1901-02). The published VOC record for the period of Krotoa’s life.
South African History Online, “The Early Cape Slave Trade”. Overview of slave-import routes and Robert Shell’s estimate of roughly 63,000 people imported between 1652 and 1808.
Dutch National Archives VOC record for 23 April 1729. Contemporary administrative record used to check the Van Noodt chronology and the number ultimately executed.
“Crime and punishment: the role of the courts and the law at the Cape”, Fundamina. Scholarly context for confession and torture under Roman-Dutch criminal procedure.
Resilient Apartheid Survivors and Their Navigation of Historical Trauma at the Castle of Good Hope, Cape Town (master’s thesis, University of Cape Town, 2020). Empirical work on how survivors of the apartheid period experience the site.
Journalism and Long-form Reporting
Independent Online (IOL), ‘Ghosts haunt guards at Cape Town Castle’ (2001). First-hand accounts from serving guards, and the origin of most later reporting on the night-shift experiences.
Mail & Guardian, ‘The history of Van Riebeeck’s slave Krotoa unearthed from the masters’ view’ (September 2016). Reporting on the memorial service and the questions surrounding it.
News24, ‘Spirit of Krotoa returned to the Castle of Good Hope’ (August 2016), and its accompanying Krotoa timeline.
Official and Institutional Sources
Castle of Good Hope, including the Castle’s history, visitor information and anniversary material.
Iziko Museums of South Africa: the William Fehr Collection at the Castle, and the Slave Lodge.
Cape Town Tourism, construction facts and visitor information.
South African History Online, entry for the Castle of Good Hope.
Further Reading
Camissa People, ‘Krotoa: Drawing the Longbow in the Fort 1642-1674’. The clearest statement of the historical objection to the Castle memorial, and essential for anyone who wants the counter-position in full.
Cape Town Museum’s Krotoa entry, including the burial-site question and the Camissa river course.
Can the Fort of Good Hope be Decolonised? (I SEE YOU, 2024). On the transformation debate and the Indian Ocean slave trade at the Cape.
The Original Shoreline project, on the pre-reclamation coastline and the Castle’s original position on it.
Discover more from Shin Stories
Subscribe to get the latest posts sent to your email.