Duarte da Silva: The R950bn safe South Africa already has the key to

In 1890, a tank of cyanide saved the Rand. A hundred and thirty-six years later, the same chemical can't reach up to 420 tonnes of gold locked inside pyrite in the Witwatersrand dumps. At today's price, that's about R950bn. Dr Duarte da Silva argues the key was made right here. BIOX is a process in which bacteria break down pyrite. Gencor developed it; it has run at Barberton's Fairview mine since 1986, and it now works in 14 plants on four continents. Both halves of the job are proven: separating out the pyrite, then letting the bacteria loose on it. Nobody has yet done the two together at dump scale. So do we build the first plant, or pay Helsinki royalties on our own invention?.By Dr Duarte da Silva*.In 1890 a Glasgow chemist, two doctors and a tank of dilute cyanide arrived on the Rand and overturned the verdict that the field was finished; the MacArthur-Forrest process went on to build the deepest industry on earth. This paper is the sequel, a hundred and thirty-six years on. Cyanide still dissolves nine ounces in ten of the world's gold, and the tonnage of it consumed rose again last year. Its monopoly is nonetheless ending, not by decree but by succession: bacterial oxidation that unlocks what cyanide alone never could, lixiviants that take the poison out of the social-licence equation where the licence is thin, and flowsheets that turn refractory waste into plant feed. The succession matters for one reason above the environmental ones: the locked tonnage. On the most recent peer-reviewed count, up to 420 tonnes of the gold left in the Witwatersrand's surface dumps sits inside pyrite, beyond the reach of the old flowsheet at any price. The chemistry that succeeds cyanide is the chemistry that opens it. And the succession, like the original coronation, is substantially a South African story — invented here, proven here, exported for four decades, and waiting to be pointed at home..THE LONG REIGN I. The tank that built the basin No industrial chemical has ever mattered more to one country than cyanide has mattered to South Africa. By 1890 the Rand was dying in its cradle. The oxidised surface ore was cutting out; the pyritic ore beneath it defeated mercury amalgamation, which recovered between forty-five and fifty-five per cent of the assayed gold and left the rest in the sand; and the field stood ready to be abandoned four years after George Harrison found it. John Stewart MacArthur, a chemist, and the brothers Robert and William Forrest, physicians, working in Glasgow, had taken British patents in October 1887 and July 1888 on the dissolution of gold in dilute potassium cyanide and its precipitation on zinc. The process was demonstrated to the Rand's mine owners in 1890 and ran first at the Robinson mine that year at a treatment cost of ten shillings a ton, recovering as much as ninety-six per cent of the gold in exactly the ore that had beaten every other method. Within two years the field's output had risen from 40,000 to 100,000 ounces a month. Everything that followed — the deep levels, the mining houses, the migrant labour system, the thousand-tonne years, Johannesburg itself — stands on that tank. To say the basin was built by chemistry is not a figure of speech..The reign has been extraordinary by any standard: a hundred and thirty-six years as the world's dominant gold process, refined but never replaced. The great refinements were South African too. Carbon-in-pulp, the technique of capturing dissolved gold on activated carbon rather than on zinc, ran first at Homestake's plant in South Dakota in August 1973, and after the first full-scale South African installations at the end of the 1970s more than a dozen Rand plants followed between 1980 and 1985, and the Anglo American research laboratories of that decade wrote the textbook the world still uses. Those were improvements in how to capture gold from cyanide solution, not departures from cyanide. About ninety per cent of the world's primary gold still takes the cyanide route today. The world consumed some 1.26 million tonnes of sodium cyanide in 2025, eighty-six per cent of it in mining, and the forecast for 2026 is 1.31 million; Orica paid 640 million dollars for the North American producer Cyanco in 2024 to secure 240,000 tonnes of capacity. Nothing in this paper disputes what the molecule achieved, and nothing in the consumption figures says it is retiring. This paper is about the two bills the molecule left unpaid — one social, one metallurgical — and about the successors, largely bred in this country, that are paying them on the specific ground where they fall due. The argument runs in four steps. Up to 420 tonnes of the gold left in the Witwatersrand's dumps is sealed inside pyrite, where cyanide cannot reach it at any price. The proven way to unseal it is to float the pyrite out of the sand and let bacteria oxidise it, a process South Africa invented and has run at Fairview since 1986. Each of those two steps has worked on its own for decades; they have never been run together on dump sand, and the one attempt to reach the locked gold without oxidation, at Ergo in 2014, failed. The next move is therefore neither a new invention nor a higher price but a pilot, and the country that owns the process, the sand and the operators should run it before the process is licensed back to it from abroad.Everything — the deep levels, the mining houses, the thousand-tonne years, Johannesburg itself — stands on a tank of dilute cyanide. This paper is about the two bills the molecule left unpaid, and the successors paying them. EXHIBIT A · THE SOCIAL BILL II. The licence cost of a poison The first unpaid bill is the social licence, and it was priced for the industry, permanently, in a single night. At about ten o'clock on 30 January 2000 a tailings dam at the Aurul retreatment plant outside Baia Mare in Romania — a joint venture between Esmeralda Exploration of Australia and the Romanian state company Remin, built to re-treat old tailings — breached and released about 100,000 cubic metres of water carrying between fifty and a hundred and twenty tonnes of cyanide into the Lăpuș, the Someș, the Tisza and the Danube. Cyanide in the Someș reached seven hundred times the permitted level. Hungarian authorities recovered 1,240 tonnes of dead fish; some 320 kilometres of the Tisza catchment were damaged; the plume travelled more than a thousand kilometres through Romania, Hungary and Serbia to the Black Sea; and towns along the way lost their water. European commentators called it the continent's worst environmental disaster since Chernobyl, and the comparison stuck. The industry's response was the International Cyanide Management Code: drafted from a Paris workshop of May 2000, completed in 2002, administered by an institute from 2003, with its first fourteen signatories in November 2005. By the end of 2025 it had 248 signatory companies — sixty one miners, thirty-seven cyanide producers, a hundred and fifty transporters — and 323 certified operations in forty-six countries, of which 115 are mines. Pan African's Fairview BIOX plant and Barberton tailings plant were certified in March 2025. The Code is real and it works: auditable standards for transport, storage, use and decommissioning that the certified part of the industry signs and is inspected against. It is also, read plainly, an admission. A process whose failure mode requires a treaty-grade management regime is a process whose social licence is rented, not owned, and 115 certified mines is a fraction of the mines that use the molecule. The rent rises jurisdiction by jurisdiction, and the ban map is specific enough to list. Montana's voters outlawed cyanide heap and vat leaching for new open-pit mines by ballot initiative in 1998, two years before Baia Mare, and reaffirmed it in 2004; the Czech Republic legislated a ban in 2002; Wisconsin in 2001; Costa Rica by moratorium in 2002 and presidential decree in 2010; seven Argentine provinces between 2003 and 2008; Hungary, nine years after the Tisza ran poisoned, in December 2009. The European Parliament voted on 5 May 2010 for an EU-wide ban by the end of 2011 and the Commission declined to act on it. Each ban is local. The signal is not: every new gold project on earth now carries the molecule as a permitting risk, a community flashpoint and a financing question, and the provenance regime now being written in London adds a documentary dimension. The London Bullion Market Association's tenth Responsible Gold Guidance, to be finalised in December 2026 and implemented through 2027, attaches a footprint declaration to the origin record, and a flowsheet that uses less of the molecule will have less to declare. None of this retires the molecule; the installed base is vast and the Code makes it insurable. But a plant designer specifying cyanide beside a township in 2026 is designing against the current, defensibly but consciously — and a designer specifying the successors is designing with it. A process whose failure mode requires a treaty-grade management regime holds its social licence on rent, not freehold — and the rent rises jurisdiction by jurisdiction. The Aurul plant that poisoned the Tisza was a tailings retreatment plant.EXHIBIT B · THE METALLURGICAL BILL III. The gold cyanide cannot reach The second unpaid bill is the one that matters most here, because it is denominated in ounces. Cyanide dissolves gold it can touch. In a large fraction of the world's remaining gold — and in a large fraction of the Witwatersrand's tailings — the gold cannot be touched, because it sits inside sulphide minerals, encapsulated in pyrite and arsenopyrite as particles too fine and too armoured for any leach solution to reach. Run that material through a conventional plant and the cyanide washes over the sulphide like rain over a safe. The count is peer-reviewed. In February 2023 Chingwaru, von der Heyden and Tadie of Stellenbosch University published, in Scientific Reports, a mineralogical census of the gold left in the Witwatersrand's surface tailings — six billion tonnes of them, on the standard estimate — by goldfield. In the Klerksdorp dumps sixty-five per cent of the remaining gold is hosted in pyrite; in Carletonville seventy-eight per cent; in Evander eighty-five per cent. The heavy-mineral fraction of the tailings runs from thirty-five to seventy-six per cent sulphides and carries from 0.71 to 10.12 grams of gold a tonne, against a bulk grade of two-tenths of a gram. Conventional cyanidation, they found, recovers about thirty per cent of what remains, and their conclusion was that up to 420 tonnes of gold — the university's press office later put the total invisible gold at 460 — sits inside detrital and arsenian pyrite in the dumps, beyond the reach of carbon-in-leach at any price. At 4,287 dollars an ounce, the close of 23 September 2026, and a rand at 16.35, 420 tonnes is 57.9 billion dollars — about 947 billion rand — of metal that is present, assayed, at surface, already crushed, already owned, and invisible to the flowsheet that built the industry. Handley's 2023 accounting in the Journal of the Southern African Institute of Mining and Metallurgy puts the gold sent to the Witwatersrand's tailings between 1886 and 2005 at 2,110 tonnes, give or take 590, at an average discarded grade of 0.36 grams a tonne; about forty per cent of that has since been retreated, leaving about 1,600 tonnes in the dumps and 48,100 tonnes, give or take 9,000, in the basin as a whole. The retreatment operators — DRDGold, Pan African, Harmony's surface plants — are working, overwhelmingly, the free and cyanide-accessible portion of that 1,600 tonnes. The locked portion — up to 420 tonnes, about a quarter of Handley's 1,600 overall, and two-thirds or more of what remains in the three sulphide-rich goldfields Chingwaru sampled — sits behind a chemical door the industry has walked past for a century, on ground already mined, already milled, already at surface, already permitted. Open the door and the surface endowment does not grow by a margin; it grows by a multiple of many operators' current reserves. The key to the door is not a price. The price is here. The key is oxidation — cracking the sulphide armour so the leach can reach the gold — and the story of oxidation is where this paper comes home to Barberton..The locked gold is present, assayed, at surface, and to the old flowsheet invisible — up to 420 tonnes behind a chemical door the industry has walked past for a century. The key is not a price. The price is here. EXHIBIT C · THE ROADS NOT TAKEN IV. Roaster, autoclave, and the one attempt at dump scale The industry always knew the locked gold was there, and its pre-biological answers are instructive because their costs explain the dumps. The first answer was the roaster: float the sulphides out of the sand, burn the concentrate at five to seven hundred degrees until the armour oxidises, then leach the calcine. Roasting works, and the Rand ran it at scale. When Anglo American commissioned ERGO on the far East Rand in 1977 it built a flotation circuit ahead of the gold plant to pull pyrite from 1.6 million tonnes of tailings a month, roasted the concentrate, and fed the sulphur dioxide to two Lurgi double-contact plants with a daily output of 1,500 tonnes of sulphuric acid; over twenty-five years the complex treated 890 million tonnes, recovered 8.2 million ounces of gold and 5.5 million pounds of uranium, and sold the acid. That is the working precedent for everything Section VII proposes, and it closed in 2005 with gold at 445 dollars. The old roasters paid for their gold in stack emissions, and on arsenopyrite feeds in arsenic that settled over everything downwind; the modern ones — Goldstrike's 6.6-million-tonne roaster in Nevada, the Syama and Gidji plants — capture the sulphur, fix the arsenic and run under emission regimes the old stacks never met. The route survived; what changed was its emission regime and its capital cost. The second answer was the autoclave: pressure oxidation of the concentrate in titanium-lined vessels at heat and pressure. Homestake ran the first commercial gold autoclave at McLaughlin in California in 1985; Goldstrike's five-million-tonne circuit, Lihir, Pueblo Viejo and Macraes followed. It is clean and it is effective, and it is so capital-hungry that only the largest orebodies can carry one. McKinsey's 2021 survey of refractory gold put the capital cost of a pressure-oxidation plant forty-eight per cent above a conventional one, and the 2026 example is i-80 Gold's refurbishment of the Lone Tree autoclave in Nevada — 430 million dollars for a 2,250-tonne-a-day vessel, construction from the fourth quarter of this year, first gold in late 2027. Between the dirty road and the unaffordable one the marginal sulphide tranche of the Witwatersrand tailings simply waited: too lean for an autoclave, too visible for an old roaster, too locked for the leach. There has been one attempt to open that tranche at dump scale, and it belongs on the record. In the 2014 financial year DRDGold commissioned a flotation and fine-grind circuit at Ergo — float the pyrite out of the retreatment stream, grind the concentrate ultra-fine to liberate the gold inside it, leach the product — with the stated target of the gold in pyrite and a hoped-for lift of sixteen to twenty per cent in recovery. The circuit reached full operation in January 2014. By the first quarter of that year production had fallen fourteen per cent and unit cost had risen twenty-four; the gold was reporting to solution but not to the carbon, the thickeners and power draw misbehaved, and on 4 April 2014 the company suspended the high-grade section. Components of the circuit were halted for evaluation in the first quarter of the 2020 financial year and it has not been mentioned in the company's operating results since, beyond boilerplate. At its 2026 results the company's process pilot is an up-flow reactor "showing good promise", not a sulphide circuit. The one operator with the sand, the plant and the cash tried the mechanical key to the safe, at the wrong price, and put it down. The lesson is not that the locked gold is unreachable. It is that fine grinding was the wrong key — fine grinding liberates; it does not oxidise — and that the right key was, and is, biological. Two clocks explain the wait. Through the long bear market the locked gold was doubly dead — unpayable at the price and unreachable by the flowsheet — and the industry was right, then, to leave it. Ergo's flotation and roasting were shut with gold at 445 dollars; DRDGold's fine-grind circuit was suspended with gold at 1,300. The price clock has since moved to 4,287. The flowsheet clock moved decades ago at Fairview. What has not moved is the verdict, still filed where a roaster, a ruined grind and a bear market left it. Ergo floated pyrite from 1.6 million tonnes a month, roasted it and sold 1,500 tonnes of acid a day for twenty-five years. DRDGold's fine-grind key broke in 2014. Between the dirty road and the unaffordable one, the locked gold waited for the biological one. EXHIBIT D · THE SOUTH AFRICAN KEY V. BIOX: the successor bred in Barberton The door-opening technology was invented in South Africa, commercialised in South Africa, and has been running in South Africa for forty years — a fact that ought to be famous and is instead a footnote. In the early 1980s researchers at Gencor's process laboratories in Johannesburg developed a biological answer to the sulphide problem: instead of roasting refractory concentrate at great cost and worse emissions, or pressure-oxidising it in autoclaves at greater cost still, let bacteria do it. Naturally occurring microorganisms — the same acid-loving, sulphide-eating genera that generate acid mine drainage, in villainous form, in the flooded basins — will, in engineered tanks, oxidise the pyrite and arsenopyrite around the gold, dismantling the armour crystal by crystal, and leave the gold exposed for a conventional leach. The process was named BIOX. In 1986 it went commercial at the Fairview mine outside Barberton, the world's first commercial biooxidation plant for refractory gold, since expanded to about sixty tonnes of concentrate a day, on a mine that had been working the same stubborn ores since the 1880s and had roasted them for most of that time. The working organisms — Acidithiobacillus, Leptospirillum, Sulfobacillus and their consortium relatives — make their living oxidising iron and sulphur. In a BIOX circuit they are given exactly what they want: finely ground sulphide concentrate, air, warmth around forty degrees, and dilute acid to live in. They eat the crystal from the outside in, converting pyrite and arsenopyrite to soluble sulphate and freeing the gold the crystal enclosed, in residence times of four to six days. No pressure vessels, no seven-hundred-degree burn, no stack. The arsenic that made old roasters infamous reports to solution and is precipitated as a stable ferric arsenate, the treatment the United States Environmental Protection Agency lists as best available; the acid the organisms generate is the process reagent, made on site by the feed itself. Fairview's gold recovery from the oxidised product has run at about 98.8 per cent. It is the domestication of the basin's own destructive chemistry: the reaction that poisons the West Rand's rivers, put in a tank and given a job. It never stopped. The Fairview plant has now run for four decades — through every gold price from 252 dollars in 1999 to 5,589 in January 2026, through Gencor, Billiton, Gold Fields and Pan African — and it is still used as the training facility for BIOX plants worldwide. The technology has since been licensed and built at fourteen plants on four continents, treating from twenty to more than two thousand tonnes of concentrate a day: São Bento in Brazil and Harbour Lights in Australia in 1992, Ashanti's Sansu plant at Obuasi in 1994 — for years the largest, on one of the great refractory orebodies of the world — Tamboraque in Peru in 1999, Fosterville in Victoria and Suzdal in Kazakhstan in 2005, Jinfeng in China in 2007, Kokpatas in Uzbekistan in 2009 at 2,137 tonnes a day, Runruno in the Philippines in 2016, and most recently Cam and Motor in Zimbabwe. More than thirty six million ounces of gold have been recovered from BIOX product. The process passed through Gold Fields' ownership to Biomin, to Outotec in 2015, and to Metso, which has since added a thermophile stage, MesoTHERM, running at sixty-five degrees, demonstrated at Fairview itself over twelve months, that cuts the cyanide consumed on the oxidised product from nineteen kilograms a tonne to eight — a cyanide-reduced flowsheet within the cyanide-succession story — and a new reactor design in 2024 that trims agitation power by a fifth. The organisms, however, remain resolutely local: the founding cultures were raised on Barberton concentrate, and their descendants are dismantling sulphides in Kazakhstan and Uzbekistan. It is one of the most successful technology exports in South African history — and, in a country that has given away most of what it built in gold, one of the few it still owns where ownership counts: the organisms, the operators and the training plant are here. The succession chemistry was born on the oldest working goldfield in the country. Sheba began producing in 1885, a year before Harrison's discovery on the Rand; Fairview and New Consort have worked the refractory ores of the De Kaap valley for more than a hundred and thirty years; the mining right runs to 2051; and in July 2018 the Makhonjwa Mountains above the mines were inscribed as a World Heritage site for rocks three and a half billion years old. It was because Barberton's gold was locked that Barberton had to become clever first. Today the operator is Pan African Resources, whose results for the year to June 2026, published on 16 September, show 47,677 ounces from Fairview, 24,320 from Sheba and Consort and 12,932 from the tailings plant on the same site, inside a group total of 272,310 ounces at an all-in sustaining cost of 1,867 dollars an ounce against an average price received of 4,235. The technology that opens sulphide-locked gold is therefore not speculative, not foreign, and not new. It is a forty-year-old South African export, running some three hundred kilometres from Johannesburg's dumps and two hundred from Evander's, operated by a listed company that owns both ends of the flowsheet. What stands between the 420-tonne tranche and a plant is not invention. It is the decision to point existing capability at a category everyone has agreed to call waste. The technology that opens the sulphide safe is a forty-year-old South African invention, running in fourteen plants on four continents and still trained at Fairview. What stands in the way is not invention. It is category.EXHIBIT E · THE WIDER SUCCESSION VI. The lixiviants after the king, scored on the record None of the new leach chemistries opens pyrite by itself: the locked gold needs oxidation first, and the leach that follows can stay cyanide. The new lixiviants answer the social bill, not the metallurgical one, and their record is easy to overstate. They have moved from laboratory curiosity to commercial fact and, in the most famous case, back again. The landmark was thiosulfate. Barrick commissioned a full commercial calcium-thiosulfate circuit at Goldstrike in Nevada in November 2014, at a cost of 620 million dollars, expressly to treat some four million ounces of stockpiled double-refractory ore that cyanide could not economically process, and doubled its throughput to 12,150 tonnes a day. It was the first industrial-scale demonstration that the world's largest gold companies would build cyanide-free chemistry when the orebody demanded it. It ran for eight years. Then, in the first quarter of 2023, Nevada Gold Mines spent thirty million dollars converting the thiosulfate resin circuit back to standard cyanide leaching with carbon-in-leach, and its 2025 technical report records the conversion in a single sentence. The thiosulfate existence proof stands as chemistry. As commerce it has been withdrawn by the operator that banked it. The livelier front is glycine — the simplest amino acid, cheap, stable, recyclable and biodegradable, developed as a gold and base-metal lixiviant by Jacques Eksteen's group at Curtin University in Western Australia and carried to market by Draslovka, the Czech cyanide producer, as GlyCat. The first commercial heap leach ran at Bayan Airag in Mongolia from August 2023: a thousand-tonne trial heap, seventy-eight per cent recovery, cyanide consumption down sixty-five per cent, reagent cost down seventy. Barrick adopted it at Bulyanhulu in Tanzania in February 2024 with an eighty per cent cut in cyanide and tailings free of weak-acid-dissociable cyanide; Draslovka and Curtin signed a three-million-dollar research collaboration in November 2024 and a Western Australian tailings partnership that August. Since then, on the company's published record to September 2026, no new named deployment. And the process is cyanide-reduced, not cyanide-free: glycine leaches with a catalytic dose of the old reagent, which is exactly what makes it bankable and exactly what the marketing sometimes omits. Alongside it run halide systems, Dundee's hypochlorite-hypobromide CLEVR process at demonstration scale in Quebec, a CSIRO cyanide-recycling technology at laboratory readiness, and a family of proprietary reagents of varying credibility — a crowded, unsettled field of the kind that precedes a technical transition, and that has already produced a graveyard of "cyanide killers". This paper does not bet on a winner and does not need to. It needs only the field-level fact, which the scorecard above leaves intact: for the first time since 1890 a gold operator designing a plant has credible chemistry choices, and for low-grade, environmentally sensitive, community-adjacent material like urban tailings, the alternatives' selling points read like a description of the feedstock. The successors are not better than cyanide everywhere; Goldstrike proved that at 620 million dollars. They are better than cyanide exactly where the Witwatersrand's remaining gold lies. The succession, properly understood, is not one chemistry replacing another but a portfolio replacing a monopoly: oxidation biological, pressurised or thermal; leach chemistry cyanide, glycine-catalysed or halide; each matched to feed, jurisdiction and file. The plant designer of 1990 had one flowsheet. The designer of 2026 has a menu..The menu after cyanide. Every route is scored against its published record; the fit column is the author's judgement for Witwatersrand surface tailings. The successors are not better than cyanide everywhere — Goldstrike proved that at 620 million dollars. They are better than cyanide exactly where the Witwatersrand's remaining gold lies: low grade, sensitive, city-adjacent sand that the old king could never touch politely. EXHIBIT F · WHERE CHEMISTRY MEETS THE DUMPS VII. The flowsheet of the second life The chemistry of this paper is the missing middle of the surface case, and it assembles into a three stage flowsheet. Stage one is running today: hydraulic re-mining of the dumps, conventional leaching of the free gold, redeposition in engineered facilities — audited economics, at a fifty-eight per cent operating margin on 0.19 grams a tonne in DRDGold's year to June 2026. Stage two is this paper. As each dump's free gold is taken, its sulphide fraction — today either returned to the residue or, at best, floated and stockpiled — becomes feed for oxidation, biological where the Barberton precedent fits, and the locked tranche begins converting to recoverable ounces on ground already permitted and plumbed. The arithmetic of concentration is what makes it plausible: if the heavy fraction of the sand carries 0.71 to 10.12 grams a tonne against a bulk 0.2, flotation is doing what it did at Ergo for twenty five years — upgrading a lean stream by an order of magnitude or more into something a sixty-tonne a-day tank can afford to treat. The bacteria never meet the sand. They meet the concentrate. Stage three closes the loop, and it is where the environmental case and the commercial case stop being two cases. The oxidation of sulphides is, chemically, the controlled version of the reaction that generates acid mine drainage in the wild — pyrite plus oxygen plus water yielding sulphate, acid and dissolved iron — and the Witwatersrand has been running that reaction uncontrolled at basin scale since the Western Basin began decanting acid water at Randfontein in 2002. The state's answer has been treatment: a Central Basin plant at Germiston rated at seventy-two megalitres a day and run at up to eighty-four, an Eastern Basin plant at about 106, a Western Basin plant that treats only part of its decant, and a long-term programme costed at ten to twelve billion rand in 2016. Every tonne of pyrite left in a dump is a future tonne of acid for those plants to neutralise, at public expense, forever. The second-life flowsheet does deliberately, in tanks, with the acid captured and the sulphate and arsenic fixed, what the dumps have been doing destructively, in the open, for a century. Ergo proved the commercial half of that sentence in 1977 by selling 1,500 tonnes of acid a day; the bacteria supply the version that needs no stack. The plant recovers gold and pre-empts the pollution by consuming its cause. That is the sense in which the greenest route and the most bankable route converge, stated precisely. Removing the sulphide from the residue removes the engine of acid generation; recovering the sulphur as acid or sulphate products monetises what the state now carries as a perpetual liability; and the gold pays for the plant that does both. The environmental case and the commercial case are not aligned by coincidence or by offset accounting. They are the same flowsheet, read from opposite ends. The oxidation plant does deliberately, in tanks, with the acid sold and the arsenic fixed, what the dumps have done destructively in the open since the Western Basin decanted in 2002. The environmental case and the commercial case are the same flowsheet, read from opposite ends. EXHIBITG · THE WORLD'S LOCKED GOLD VIII. A global problem, and who solves it first The locked tranche is not a Witwatersrand eccentricity; it is the direction the whole world's gold is heading. The free-milling deposits — the ores that gave up their gold to a simple leach — were found and mined first, everywhere, because they were easy. What remains is increasingly refractory. McKinsey's 2021 survey put twenty-four per cent of the world's gold reserves and twenty-two per cent of its resources in refractory ore, and found that the refractory ore was richer — 2.25 grams a tonne against 1.21 for free-milling — which is the geological reason the industry will follow it however hard the chemistry. Obuasi, where BIOX first ran at scale in 1994; the Carlin-type ores of Nevada that drove Barrick to build and then abandon thiosulfate; the double-refractory stockpiles that justify a 430-million-dollar autoclave at Lone Tree; the pyrite of every mature goldfield on earth. The gold industry's future is, chemically, a locked-gold future, and the succession chemistry of this paper is not a niche but the field's coming mainstream. The South African question follows. The country that invented the leading biological answer, that trained the operators of thirteen plants abroad, and that holds one of the largest above-ground refractory endowments in the world in its own dumps is positioned — by accident of history and chemistry — at the head of a global queue. Every legacy goldfield that follows the Witwatersrand into its second life will need this flowsheet: Kalgoorlie's, Nevada's, the Urals', Ghana's. The first jurisdiction to run flotation-to-biooxidation at dump scale writes the manual, trains the crews, licenses the cultures and exports the service, and sells a service as well as a metal. The queue forms behind a tank of bacteria that were South African before they were anything else. The world's gold future is, chemically, a locked-gold future: a quarter of reserves, and the richer quarter. The first jurisdiction to run the answer at dump scale writes the manual, trains the crews and exports the service.THE BEAR, MET IX. What the sceptic rightly takes The bear's first blow: cyanide is not going anywhere, and pretending otherwise is green theatre. Ninety per cent of primary gold takes the cyanide route; consumption rose to 1.26 million tonnes in 2025 and will rise again; the Code makes it insurable; and the flagship cyanide-free plant on earth was converted back to cyanide in 2023 for thirty million dollars after eight years and 620 million of capital. Conceded, almost entirely, and conceded harder than most advocates of the successors are prepared to. This paper has not argued that cyanide dies; it has argued that cyanide's successors win specific ground — refractory feed, sensitive jurisdictions, city-adjacent tailings — and that the specific ground they win happens to be where the Witwatersrand's remaining gold sits. The claim is a targeting claim, not an extinction claim, and Goldstrike's reversal, on high-grade Nevada ore with an autoclave and a roaster next door, does not touch a proposal for lean sand beside Soweto. The bear's statistics about the installed base show that the successors will have to earn their ground feed class by feed class, not that the ground is unwinnable. Second blow, and the one with teeth: biooxidation at tailings-scale volumes is unproven. BIOX runs on high-grade concentrate at twenty to two thousand tonnes a day, not on megatonnes of lean sand, and the one operator that tried to concentrate and liberate the pyrite gold from a Witwatersrand retreatment stream at scale — DRDGold, in 2014 — lost fourteen per cent of production and added twenty-four per cent to cost inside a quarter and shut the section on 4 April. Conceded as the engineering question of the whole case, and it is why the flowsheet of Section VII runs concentration first and oxidation second and why fine grinding is not on the card. Flotation of pyrite from Rand tailings is established practice: Ergo did it for twenty-five years. Bacterial oxidation of pyrite concentrate is established practice: Fairview has done it for forty. What has never been done is the two in series on 0.2-gram sand with a modern liner at the end, and the status is that every unit operation in the chain is individually proven within three hundred kilometres of the dumps, while the integrated chain at dump scale is the detailed-study gap — the kind that must be run before capital, not after. The bear defines the study. He does not close the door; he defines its hinges. Third: the skills. Biooxidation and alternative lixiviants need process chemists and metallurgists, and the South African pipeline of both has been thinning for twenty years. Conceded and doubled — but the remaining skill concentrates in exactly the right place: the operators, laboratories and consultancies of the South African gold rump, including the Barberton plant that trains the world's BIOX operators. The succession chemistry is one of the few fields where the national skills story runs the right way: the world licenses a South African process, and the founding expertise still lives a morning's drive from the feedstock. Use the capability while its carriers can still teach it — and there is a commercial reason to hurry, because Metso, not Gencor, owns the marque now, and the royalty on the Witwatersrand's second act will flow to Helsinki unless the operators here build the reference plant first. Every unit operation in the chain is proven within three hundred kilometres of the dumps; the integrated chain at dump scale is the detailed-study gap. The bear defines the study. He does not close the door; he defines its hinges. THE VERDICT X. The succession, claimed or exported South African gold has a pattern with its own technologies: invent, prove in first production, then abandon with the strategy the technology served, the frontier left where the retreating company dropped it. The chemistry of this paper is the same national asset at a later, luckier stage: invented here, commercialised here, exported successfully for forty years — and now needed at home, by the surface retreatment industry, against the locked quarter of the gold left in the dumps. The question the pattern asks is whether the country that bred the bacteria will point them, at last, at its own dumps — or whether the second life of the Witwatersrand will one day be processed under licence from Helsinki, with the royalty flowing to whoever bothered to own what Barberton built. The programme. Commission the study Section IX defined: flotation-to-biooxidation on representative dump material from Klerksdorp, Carletonville and Evander, where Chingwaru's deportment numbers say the prize is largest, run by the operators who already own both ends of the flowsheet, costed with upper and lower estimates and with the 2014 fine-grind failure as its starting document rather than its embarrassment. Protect the skills: the process chemists and biooxidation operators of the Barberton school are a national asset with a retirement date, and the window in which they can still train their successors is closing. Put the succession into the certified file: a cyanide-reduced, bacteria-driven, sulphur-capturing flowsheet is exactly the process biography the London file is beginning to reward as its tenth version is implemented through 2027, and it should be documented as such from the first pilot campaign. And write the locked tranche into the national endowment conversation explicitly — 420 tonnes is not a rounding item; it is nearly five years of national production at the 2025 rate of about ninety tonnes, sitting behind a door whose key the country already cut. .Read more:.Duarte da Silva: SA's richest goldmine is the bin — and almost nobody's mining it.A roaster and a BIOX tank run the same reaction — one at seven hundred degrees under a stack, the other at forty degrees in open tanks, with the acid captured and the arsenic fixed —and the difference between them is the whole of this paper. A hundred and thirty-six years ago imported chemistry saved this basin in its infancy. The chemistry that opens its second life needs no importing; it has been running up the road the whole time, waiting — like the dumps themselves — for the category to change and the capital to notice. A chemist, two doctors and a tank of cyanide built the first Witwatersrand. The second is available to the first operator who reads Fairview's forty years for what they are: the longest-running, least-celebrated proof in world mining that South Africa already owns the key to its own locked gold. .The chemistry that opens the locked gold was invented here, proven here, and has run for forty years a morning's drive from the dumps. Cyanide saved the Basin in 1890 and the world uses more of it every year. Its successors open the sulphide safe — four hundred and twenty tonnes behind a door the industry walked past for a century..*Dr Duarte F da Silva • Capital Markets Strategist • Managing Director, Northbound Processing • Germiston, September 2026.A NOTE ON THE FIGURES Every figure in this paper was checked against its source in the week of 14 September 2026 and re-checked for later releases on 24 September 2026; the sources follow. MacArthur-Forrest: British patents of October 1887 and July 1888; first Rand operation at the Robinson mine, 1890, at 10s a ton; amalgamation recovering 45–55% of assayed gold in 1890; cyanide recovery up to 96%; monthly output from 40,000 to 100,000 oz within two years (MacArthur papers, Balliol College archive; Prometheus, "Discovery, development and diffusion of the cyanide process, 1887–1914"). Carbon-in-pulp: Homestake, Lead, South Dakota, August 1973; first South African full-scale installations at the end of the 1970s (sources give 1978 and 1980 for Modderfontein); more than twelve South African plants 1980–85 (Fleming and McDougall, SAIMM literature). ~90% of primary gold by cyanidation: standard industry references. Sodium cyanide 1.26 Mt (2025) and 1.31 Mt (2026), mining 86.1% of volume: Mordor Intelligence, 2026; Orica's US$640m acquisition of Cyanco, 2024 (240 kt/yr): Orica. Baia Mare: 30 January 2000; ~100,000 m³; 50–120 t of cyanide; ~1,240 t of fish; ~320 km of Tisza catchment damaged; cyanide >700× the limit in the Someș; Aurul as a joint venture of Esmeralda Exploration and Remin (Baia Mare Task Force report, December 2000; ReliefWeb). International Cyanide Management Code: Paris workshop May 2000, Code 2002, ICMI 2003, first 14 signatories November 2005; 248 signatories, 323 certified operations (115 mines, 47 production, 161 transport) in 46 countries at end-2025 (ICMI newsletter, Q1 2026); Fairview BIOX plant and BTRP certified 17 March 2025 (ICMI directory). Cyanide restrictions: Montana I-137 (1998, reaffirmed 2004), Wisconsin (2001), Czech Republic (2002), Costa Rica (2002 moratorium, 2010 decree), Argentine provinces (2003–08), Hungary (December 2009), European Parliament resolution of 5 May 2010 (E&MJ; EUR-Lex 52010IP0145). Tailings mineralogy — gold deportment to pyrite 65% (Klerksdorp), 78% (Carletonville), 85% (Evander); heavy fraction 35–76% sulphides at 0.71–10.12 ppm Au; conventional cyanidation ~30%; up to 420 t Au in detrital and arsenian pyrite; 6 billion t of tailings: Chingwaru, von der Heyden and Tadie, Scientific Reports 13, 22 February 2023 (Stellenbosch University press release of April 2024: ~460 t). Basin ledger — 50,200 t produced 1887–2019; 2,110 ± 590 t sent to tailings 1886–2005 at 0.36 g/t; ~1,600 t remaining in tailings; 48,100 ± 9,000 t remaining in the basin: Handley, Journal of the SAIMM 123(4), April 2023. ERGO: commissioned 1977; ~1.6 Mt/month; flotation for gold, uranium and pyrite; two Lurgi double-contact acid plants at 1,500 t/day; 1.5 Mt/month CIL plant from 1985 (Bosch, "Extractive Metallurgy of Gold", SAIMM, chapter 12); 890 Mt treated, 8.2 Moz gold and 5.5 Mlb uranium over 25 years; closed by AngloGold in 2005; DRDGold joint venture June 2007 (Mining Review; SEC 6-K of 7 June 2007). DRDGold flotation and fine-grind: full operation January 2014; Q1 2014 production −14%, cost +24%; high-grade section suspended 4 April 2014 (SEC 6-K, April 2014); components halted Q1 FY2020 (20-F, 2020); up-flow reactor pilot (FY2026 results, 19 August 2026). Pressure oxidation: McLaughlin 1985 (SME, "Gold pressure oxidation circuit design: a 40-year historical review", 2022); i-80 Gold Lone Tree autoclave US$430m, 2,250 t/d, construction Q4 2026, first gold by end-2027, ~40% of capital committed (update of 28 July 2026); Nevada Gold Mines Carlin Complex technical report of 14 March 2025 (autoclave 5.0 Mt/yr; roaster 6.6 Mt/yr; "a $30M conversion from the calcium thiosulphate RIL process to standard cyanide leaching with CIL was completed in Q1 2023"). Goldstrike TCM: US$620m; first doré November 2014; 12,150 t/d; ~4 Moz stockpiled double-refractory ore (Ausenco; Canadian Mining Journal). McKinsey, "Refractory gold ores: challenges and opportunities", 23 March 2021 (24% of reserves, 22% of resources; 2.25 g/t vs 1.21 g/t; POX capital +48%). BIOX: developed by Gencor Process Research; commercial at Fairview 1986 (some references date the full scale expansion to 1988), since expanded to ~60 t/d; recovery ~98.8%; Fairview as global training facility (Metso BIOX brochure; Pan African integrated annual report 2023); 13–14 plants, >25–36 Moz, 20–2,137 t/d (Metso; International Mining, 27 November 2024, including the 14th at Cam and Motor, Zimbabwe); plant dates and capacities from Metso and van Aswegen, van Niekerk and Olivier, "The BIOX process for the treatment of refractory gold concentrates" (Springer, 2007); MesoTHERM at ~65°C demonstrated at Fairview, cyanide 19–20 to 8–10 kg/t (Metso Outotec, 2020–21); OKTOP BIOX reactor, >20% agitation energy saving (November 2024); ferric arsenate as US EPA best demonstrated available technology; consortium organisms and 4–6 day residence per the technical literature. Barberton: Sheba from 1885; mining right to 2051; Makhonjwa Mountains inscribed 2 July 2018 (UNESCO); Pan African summarised audited results for the year to 30 June 2026, published 16 September 2026 (group 272,310 oz; Fairview 47,677 oz; Sheba and Consort 24,320 oz; BTRP 12,932 oz at AISC US$1,564/oz; group AISC US$1,867/oz; average price received US$4,235/oz; net cash US$185.8m; FY2027 guidance 280,000–302,000 oz). DRDGold FY2026 operating margin 57.8% at 0.193 g/t (results, 19 August 2026). Acid mine drainage: Western Basin decant 2002; Central Basin plant rated 72 Ml/d, run at up to ~84; Eastern ~106 Ml/d; Western partial treatment only (TCTA); long-term solution R10–12bn (Department of Water and Sanitation / TCTA, 2016; parliamentary oversight, October 2025). South African gold output ~90 t in 2025 (USGS estimate; Minerals Council, 9 February 2026). Gold: US$4,287.39 close on Wednesday 23 September 2026 (Trading Economics); record US$5,589.38 on 28 January 2026; 1999 low US$252.80; USD/ZAR ~16.35 on 23 September 2026 (Trading Economics; ECB reference 16.3471); 420 t × 32,150.7 oz/t × US$4,287.39 = US$57.9bn ≈ R947bn. Distances: Barberton lies ~300 km from Johannesburg and ~200 km from Evander in a straight line (author's measurement). The framing of the locked tranche as "the sulphide safe", of oxidation as "the domestication of the basin's own destructive chemistry", and of the menu table's fit column is the author's own. Northbound Processing • Germiston, September 2026. The Gold Papers — Paper 25. This article does not constitute investment advice. The maturity of each process route is stated as disclosed by its operators and developers, including the withdrawal of the Goldstrike thiosulfate circuit in 2023; the integrated flotation-to-biooxidation flowsheet of Section VII is a proposal for detailed study, not a claim of demonstrated economics; the 420-tonne figure is a peer-reviewed mineralogical estimate of contained gold, not a reserve. The author operates a business in secondary and by-product metals processing that a Witwatersrand sulphide programme would benefit.

Duarte da Silva: The R950bn safe South Africa already has the key to

Related Stories

Duarte da Silva: SA's richest goldmine is the bin — and almost nobody's mining it

Duarte da Silva: The country we declared explored

Dr Duarte da Silva on environmental arbitrage: The asset in the flood

Da Silva’s Gold Papers: Sibanye's plan to flood the richest reef on earth - at best gold price ever