Last week we published a long write-up of the first CDI Symposium, where a room of soil scientists picked apart the world’s most-measured enhanced-rock-weathering experiment and landed on an uncomfortable conclusion: the rock dissolves, but the expected carbon-removal signal often doesn’t show up — and whether it does depends heavily on the specific soil. Enhanced rock weathering (ERW), the symposium suggested, is not a universal climate fix but a context-dependent one, and measuring it honestly is the whole ballgame.
This week, a new paper argues exactly the same thing — from the opposite direction. Not from a cutting-edge greenhouse, but from the archive.
The paper is “Accelerating weathering: lessons from a century of soil rejuvenation”, a perspective in Frontiers in Climate by Budiman Minasny (University of Sydney) and Xavier Dupla (ETH Zurich). If Dupla’s name sounds familiar, it should: he’s one of the CDI Symposium speakers, the one who showed that crushed rock improved the crop — less acidic soil, better forage — without producing a clean carbon signal. Here, he and Minasny reach back ninety years to show that the field has been learning, forgetting, and re-learning this same lesson for a very long time.
A “senile” soil and a bag of basalt
The story starts in 1930s Mauritius. The island is built from basalt lava, but by the early twentieth century its tropical soils were exhausted: decades of intense rain, heat, and cultivation had leached the nutrients out and left “acidic, iron- and aluminium-rich soils.” A sugar-cane scientist named Octave d’Hotman de Villiers called them “senile” — soils weathered all the way to the end of their useful life.
His fix was elegant and, in hindsight, ahead of its time: if the climate had weathered the good minerals out of the soil, put fresh minerals back in. He spread finely crushed basalt to reintroduce primary minerals, replenish base cations, and restore the soil’s buffering capacity — betting that Mauritius’s warm, humid climate would weather the added rock quickly. Crucially, his goal was agronomic — rejuvenating dead dirt — not capturing carbon. Nobody was talking about CO₂ removal in 1938.
And it worked. In his 1938 Belle Rive trial (basalt at 85 t/ha), treated cane yielded about 13% more than the controls. In the bigger 1947 Hermitage experiment he tested staggering rates — 225 and 450 tonnes per hectare — and got roughly 9% and 18% annual yield gains respectively, with the effect growing in later ratoons, often past 20%. Side experiments were wilder still: soybeans grown at 370 t/ha reportedly yielded over 400% more. The rock was, by any agronomic measure, a triumph.
Then the thread was lost — the way scientific threads often are. De Villiers retired in 1951, the research station was handed to industry, a cyclone flattened the facilities, and his successor died suddenly in 1963. The trials were abandoned, and the results faded into the literature for half a century.
The line the modern gold rush forgets
Here is the sentence in the paper that every ERW investor, buyer, and founder should tattoo somewhere visible:
“A positive agronomic response does not necessarily imply net carbon removal.”
De Villiers proved, decisively, that crushed rock can bring a tired soil back to life. That is a real, valuable thing. But it says nothing, by itself, about climate. A bigger sugarcane harvest is not a tonne of CO₂ removed. The two can travel together, but they don’t have to — and today’s carbon market is built on quietly assuming they do.
This is the exact fault line the CDI Symposium landed on. Dupla’s own symposium talk showed the agronomic wins (acidity down about a third, better forage) sitting right next to a missing carbon signal. Mike Kelland gave it a mechanism — “variable-charge buffering,” where the soil grabs the weathered calcium and magnesium and holds them on newly created exchange sites, so the rock can dissolve without capturing any carbon at all. The 1930s and the 2020s are describing the same gap between dissolution and drawdown, ninety years apart.
The replication that didn’t replicate
The paper’s sharpest modern data point is a 2024 attempt to redo de Villiers’ experiment under contemporary conditions, in Queensland, Australia: crushed basalt at 50 t/ha/yr for five years on sugarcane. The result?
No significant yield improvement over five years. The basalt did change the soil chemistry — pH up, magnesium and silicon up — but the crop didn’t care, and direct carbon removal was minimal.
Why the opposite of Mauritius? Two site-specific reasons, and both are the whole point:
- The Queensland soils weren’t nutrient-starved. De Villiers’ magic depended on “senile” soils desperate for exactly what basalt provides. Give the same rock to a soil that isn’t limited by those nutrients and you get… a more expensive soil chemistry and a shrug.
- The dissolution was driven by strong acids, not carbonic acid. This is the geochemical heart of it: rock can be dissolved by acids that have nothing to do with atmospheric CO₂, in which case you’ve weathered the rock but removed no carbon. As the authors put it, “the stoichiometric representation of ERW as a simple geochemical reaction does not capture the complexity of soil systems.”
Same rock, different soil, opposite outcome. Which is why the paper’s headline number is so sobering: across rock types and field conditions, measured CO₂ removal varies by up to four orders of magnitude — from around 8.6 tonnes of CO₂ per hectare per year in some trials to essentially nothing in others. That is not a rounding error. That is the difference between a climate solution and a boondoggle, and it hinges on where you do it.
And, again, the soil-carbon wildcard
If you read last week’s symposium piece, you’ll remember the most unsettling open question: the change ERW might cause in the soil’s own carbon stock could be big enough to swamp the tiny weathering signal everyone’s chasing. The paper says the same thing, in almost the same breath. Shifting a soil’s pH and nutrients can either fire up microbes that respire carbon away or grow more plant matter that adds it — and those effects, the authors note, are “one to three orders of magnitude larger than the CO₂ capture usually expected from ERW.”
A century of soil science and the world’s largest ERW greenhouse independently arrive at the same warning: you cannot credibly account for the small thing (weathering-driven CDR) until you can measure the enormous thing it’s hiding inside (soil carbon). Nobody can, yet.
The material reality nobody puts on the poster
There’s also the unglamorous logistics the historical rates make vivid. De Villiers used 85 to 450 tonnes of crushed rock per hectare. That rock has to be quarried, milled to a fine powder, hauled, and spread. The paper is blunt that the net climate benefit “depends not only on weathering rates, but also on rock availability, mining intensity, grinding energy, haulage distance, transport mode, fuel source, local infrastructure and the carbon intensity of the electricity and transport sectors.” Weather the rock in the wrong place, with the wrong grid and the wrong trucks, and the life-cycle emissions eat the removal.
The most important sentence for our field
The line I keep coming back to isn’t about chemistry at all:
“Each new wave of interest in rock weathering has tended to restart the conversation with new terminology and motivations, while revisiting the same assumptions, constraints, and unresolved questions.”
That is a quiet indictment, and it’s fair. ERW has been “discovered” more than once — as soil rejuvenation in the 1930s, as a geoengineering idea in the 2000s, as a carbon-credit product in the 2020s — and each wave has tended to sprint past the same three walls: it needs enormous amounts of rock, it behaves completely differently from site to site, and the field evidence keeps lagging the models.
But here’s the optimistic reading, and it’s the one Captain Drawdown holds. The current wave doesn’t have to restart the conversation. It can finish it. The CDI Symposium and this paper are the same argument from two directions — the greenhouse and the archive — and they converge on a genuinely useful playbook:
- Treat ERW as a context-dependent intervention, not a universal solution. Target the soils where it actually works: highly weathered, nutrient-poor ground where plants and microbes generate enough acidity to drive dissolution, and where the carbon can be independently verified.
- Keep agronomy and carbon in separate columns. Sell the soil benefit as a soil benefit; only claim the tonne when you can measure the tonne.
- Make measurement (MRV) the product, not an afterthought — because without it, four orders of magnitude of uncertainty is exactly the room a market needs to fool itself.
None of that is a reason to walk away from enhanced rock weathering. It’s the opposite: it’s the difference between a method that earns trust and one that quietly spends it. A scientist in 1938 already knew that a better harvest isn’t a climate outcome. Ninety years later, with a greenhouse full of instruments and a century of hindsight, we’re finally in a position to prove — site by measured site — when it actually is.
Sources: Minasny, B. & Dupla, X. (2026), “Accelerating weathering: lessons from a century of soil rejuvenation,” Frontiers in Climate. The historical Mauritius trials are drawn from that paper; the modern Queensland replication is Holden et al. (2024), as cited therein. Companion reading: our write-up of the first CDI Symposium.