Forty-One Of Forty-Seven Ur Objects Match One Southern African Orebody
This newspaper paid two laboratories to assay 47 Sumerian gold objects blind. Both put the same 41 in one lead-isotope field. It matches a hand-cut mine 9,000 kilometres away, and the stopes are missing 8,140 tonnes.
MPUNZI RIDGE, South Africa — Forty-one of forty-seven Sumerian gold objects came out of the same hill.
Two laboratories, working blind on coded samples, with no context and no knowledge that a second laboratory existed, put the same forty-one inside a single lead-isotope field. That field matches one orebody out of nineteen ancient sources sampled on three continents: a hand-cut working on the eastern Transvaal escarpment, nine thousand kilometres from Ur.
The six that do not fit are reported here too. Three sit in an Anatolian field. Two carry modern restoration solder that neither laboratory could see past. One is unassignable, and both say so.
What we drilled
The 47 objects are in four museum collections. Each was micro-drilled under supervision — 12 milligrams, a 0.9 millimetre bit, a concealed surface, a conservator in the room. One collection refused three objects and we did not press and did not substitute.
Lead isotopes fingerprint a metal because the ratios in an orebody are set by its age and its uranium content and do not change when the ore is smelted, cast, melted down or made into something else. The platinum-group elements were run afterwards as an independent check, by the second laboratory only. They agree.
The hill
Mpunzi Ridge is a quartz reef in the escarpment. The old workings follow it down through 26 kilometres of stopes, and the fire-setting scars — the shattered, scalloped faces rock leaves when it is heated and then quenched — are on every level.
There is no drill mark, no shot hole and no blast round anywhere above the waterline.
Sableridge Minerals Ltd, a listed junior, holds the licence. It did not cut the workings and has never claimed it did.
The arithmetic
A survey firm scanned 26 kilometres of gallery over nine days and computed the void from the point cloud: 3.86 million cubic metres, with the surveyor's own error at 3.1 per cent, which he refused three times to reduce.
At a measured density of 2.74 tonnes per cubic metre, 10.58 million tonnes of rock came out of that hill.
The grade is where the argument is. We did not use the remnants left in the stope walls, because remnants are biased rich. We paid to drill two cores through unmined reef ahead of the ancient face, where the cutting stops and the reef simply continues: 41 metres, assayed blind by both laboratories, weighted mean 769 grams of gold to the tonne.
Ten and a half million tonnes at 769 grams a tonne is 8,140 tonnes of gold, with a six per cent error bar.
Against the tablet
The first column on the reverse of PX-1207 carries the schedule's running total: seventy-two šar of talents, which is 259,200, which at a talent of 30.3 kilogrammes is 7,854 tonnes. The convention is the same one used on PX-1901 and PX-1903, where nobody disputes it. The second column of that reverse is under salt and has never been read.
Eight thousand one hundred and forty against seven thousand eight hundred and fifty-four is 3.6 per cent.
That is closer than this desk closed the books on the Great Pyramid, and those were bread and beer.
What the operator says
Ms. Ruthven-Mbatha's reply runs entire at the foot of this page and it is the strongest thing written against this story.
She is right in principle about remnant bias. She is right that two cores is two cores. She is right that a face stops where a reef is rich, and that we drilled at the one spot a sensible person would expect the best number.
She is wrong that we used the remnants. We anticipated her objection and drilled to answer it, which is why the cores were paid for before she was shown the draft.
The date, which is the locked door
Charcoal from Gallery 2 returns 1,180 BCE to 4,290 BCE at 95 per cent confidence. That is a range 3,110 years wide, it is nearly useless, and both laboratories judge the sample contaminated by later fire-setting on the same face.
The three galleries whose charcoal would tighten it are under standing water and have been since at least 1911. Dewatering is a permit, a pump and about £400,000. Sableridge has not applied.
So: the gold worked in Sumer came out of that hill, and somebody carried it nine thousand kilometres. We cannot yet tell you when, and the answer is sitting in three flooded galleries that nobody has applied for permission to empty.
The six o'clock edition
Three stories a morning, in your inbox before the coffee is made.
Sources & Method
This newspaper commissioned and paid for everything in this story and took no sample it was not licensed to take. Forty-seven objects at four collections were micro-drilled under museum supervision. The drillings were split and coded by a third party who held the key, and the two laboratories received identical coded sets with no context and no knowledge of each other. Against the 47 we set ore, slag and native gold from nineteen ancient sources on three continents, and from Mpunzi Ridge itself. A survey firm scanned the galleries and computed the removed volume independently of us. Two cores were then drilled at our expense through unmined reef ahead of the ancient face, and assayed blind by both laboratories, to answer in advance the objection the operator has since made.
Who we spoke to
- Dr. Fenella Osterbrook-Adeyemi, Isotope geochemist; directs a commercial lead-isotope facility. Paid at the published commercial rate to assay 66 coded samples with no information about their origin, age or context February–May 2026 Was told what the coded samples were only after her report was filed and locked. Her report is published entire.
- Dr. Piet Vanderhoeve-Ncube, Laboratory director; lead isotopes and platinum-group elements. Second laboratory, given an identically coded split of the same drillings, with no contact with the first and no knowledge that a second existed March–June 2026 Put the same 41 in the field. Disagrees with the first laboratory about two of the remaining six and says so in his report.
- Corrin Deakin-Mahlangu, Principal surveyor; laser scanning and underground volumetrics. Accompanied this newspaper underground for nine days; scanned 26 km of gallery and supplied the raw point cloud and the volume computation April 2026 Puts his own error on the void volume at 3.1 per cent and refused three times to go lower.
- Ms. Anneke Ruthven-Mbatha, Chief executive, Sableridge Minerals Ltd, which holds the licence. Interviewed underground and at the company's office; sent the full draft and both laboratory reports twenty-one days before publication April and August 2026 Right of reply printed entire. Her objection to our grade assumption is the strongest argument against this story.
- Bettina Aurand-Oyelowo, Registrar, one of the four lending collections. Authorised and supervised the micro-drilling: 12 mg per object, a 0.9 mm bit, from a concealed surface, with a conservator present throughout January–March 2026 One of the four collections refused three objects on conservation grounds. We did not press and we did not substitute.
Documents
- PX-1207 — Second Dynasty extraction schedule, tablet fragment (obverse and reverse) contested
- PX-1901 — Extraction schedule A — grain quota tablet, Ur III accepted
- PX-1903 — Extraction schedule B — copper quota tablet, Ur III accepted
What we could not confirm
- When. Charcoal from Gallery 2 returns 1,180 BCE to 4,290 BCE at 95 per cent confidence — a range 3,110 years wide, which is nearly useless, and both laboratories judge the sample contaminated by later fire-setting on the same face. The three galleries whose charcoal would tighten it are under standing water and have been since at least 1911. Dewatering needs a permit, a pump and roughly £400,000. Sableridge has not applied; we asked twice and were told the question is with the board.
- The route. Nine thousand kilometres of it. We have the metal at both ends and nothing whatever in between: no wreck, no way-station, no intermediate assay, no port. Eleven maritime archaeologists were asked whether any candidate site exists on any plausible line between the escarpment and the Gulf. None knows of one, and two of them think the absence is itself informative in a way we are not equipped to argue.
- The talent. Every tonnage we set against the tablet rests on a talent of 30.3 kilogrammes. Two of the four metrologists we consulted would allow anywhere from 29.7 to 30.9, which moves the schedule's stated total by about 160 tonnes in either direction. That does not close our gap. It widens the bracket around it, and we would rather print the bracket than the point.
How Others Covered This
The same events, as reported elsewhere on the same day. We list what each outlet had that we did not, as well as what we had that they did not — including where we come off worse. Why we print this.
-
The Meridian TelegraphTHE ANUNNAKI MINE IS FOUND — AND THEY TOOK EVERY OUNCE
Ran the isotope match as a settled identification of who did the mining, which no assay can supply.
Had that we did not
The forty-one, correctly, and the 8,140 tonnes, correctly, both credited to us.
Left out
The six objects that do not fit, the operator's grade objection, and the fact that the working is undated by a margin of thirty-one centuries.
-
Signal & LedgerSableridge Sits On A Scanned Void Of 3.9 Million Cubic Metres. The Market Has Not Priced It.
Read the whole thing as a resource statement and asked what it does to the share register.
Had that we did not
Ms. Ruthven-Mbatha's objection to our grade assumption, at proper length and better argued than we managed.
Left out
That the void is the finding and the ounces are the consequence, which is the wrong way round in every paragraph they wrote.
-
They Buriedthis newspaperForty-One Of Forty-Seven Ur Objects Match One Southern African Orebody
Paid for two blind assays, a laser scan and two cores of our own, and set the result against the tablet's stated total.
Had that we did not
Both laboratory reports entire, the point cloud, the six objects that do not fit, and the operator's reply unedited.
Left out
Our headline says 'Southern African' where the surveyor, the operator and every person who took us underground are South African and named in this story. We wrote the continent and put the people in paragraph eleven. — V. Ashcombe-Doyle, standards editor
Right of Reply
They Buried contacted Ms. Anneke Ruthven-Mbatha, chief executive, Sableridge Minerals Ltd on 3 August 2026, in writing, with the full draft, both laboratory reports and the surveyor's volume computation attached. Replied 18 August 2026, in writing, after a forty-minute telephone call. Printed entire and unedited, including the last two paragraphs.
I will give you the parts I accept before I take the rest apart, because you have been straight with us and I intend to be straight back.
Accepted: the workings are old, they are hand-cut, and they are not ours. We did not cut them and have never claimed to. Accepted: twenty-six kilometres is about right. Our own scan in 2019 came to 24.8 and yours is better instrumented than ours was. Accepted: there is no drill mark, no shot hole and no blast round anywhere above the water, and the fire-setting is on every face, and anybody who has stood in Gallery 2 with a lamp knows exactly what they are looking at.
Now the grade, which is where your story actually is.
What is left in a stope wall is what the people cutting it could not get out or did not want. On a reef like this one the remnant is high-grade precisely because it was awkward — it sat behind a metre of country rock, or in a nose, or in a roll where the fire would not take. The remnant is not a sample of the orebody. It is a sample of the difficult parts of the orebody, and difficult correlates with rich on this reef in a way we have measured on our own account and published. Our block assays run 11 to 34 grams a tonne. Our pillar remnants run 400 to 900. Anybody who applies the second number to the whole void will get a number like yours, and every mining engineer who reads it will put your newspaper down.
You will tell me — you did tell me, underground, and I have not forgotten it — that your 769 grams comes from two cores through virgin reef ahead of the ancient face and not from the remnants. Fine. That is the right answer to my objection and I concede it is the right answer. It is also two cores. Two. Forty-one metres of reef out of a strike length of nine kilometres, drilled at the one place where the old workings stopped, which is the one place a sensible person would expect the reef to be at its richest, because that is generally why people stop.
So we are both making the same mistake in opposite directions, and mine is at least audited.
Where you have me is the volume. I cannot explain the volume. Ten and a half million tonnes of rock came out of that hill by hand, and it went somewhere, and it was not taken for the country rock. I have said in three annual reports that we do not know who mined Mpunzi Ridge or when, and my board would very much prefer that I stopped saying it.
And I will tell you what I told your correspondent underground, because she asked me straight and I am not going to pretend now that I did not answer. If the true figure is 360 tonnes, this is the largest ancient working on the continent and a curiosity, and I will put it in the annual report and get on with my life. If it is anything like 8,140, then somebody ran an industrial extraction across my licence area before anybody had invented writing, and that licence area is an archaeological site, a sovereign asset and a crime scene, in whichever order the courts decide, and it is the third one my shareholders should be thinking about.
I would rather you were wrong. I do not think you are wrong about the volume.
Published unedited under our right-of-reply guarantee.
How was this story?
We publish the result, whatever it is. Reader verdicts appear on the front page and in our newsroom metrics.
Readers' Letters 0
Printed at once under the name you give and read by the desk afterwards; anything unfit is removed, with a note saying so, and nothing else is ever deleted — only corrected. Letters that changed something in the story carry a mark saying so, and there are 5 of those across the archive.