SINKSTONE

PUBLIC REGISTRY · ENHANCED ROCK WEATHERING

Eleven claims.
Every one of them checkable.

Sinkstone is a public registry for enhanced rock weathering. Every tonne is linked to the field, the sample batch, the lab, and the retirement record. If any of that is missing, the tonne does not list.

LIVE ON THE LEDGER

41,208 t

retired, irreversibly

0 t

issued forward, ever

1,140 t

voided by us, published

01 / THE MANIFESTO

Eleven claims, each one falsifiable.

Each claim links to the record that would prove it wrong. If you find one that does not hold, the challenge goes on the register with our answer beside it.

01

No tonne is issued before the soil samples are in. We have never issued a forward credit and never will.

See the data →

02

Every certificate links to the raw lab data, including the runs that disagreed with each other.

See the data →

03

Retirement is final and public. You cannot quietly unretire a tonne here, because the ledger does not have that function.

See the data →

04

Every field has a name, a boundary polygon and an operator we can put you in a room with.

See the data →

05

Sampling depth, date and method are recorded per batch. If a batch is resampled, both results stay visible.

See the data →

06

We publish the lab, the instrument and the run identifier. Anonymous chemistry is not evidence.

See the data →

07

Weathering is modelled with published parameters, and the model version is stamped onto every tonne it issued.

See the data →

08

Uncertainty is carried through to the certificate. We publish the deduction, not only the headline number.

See the data →

09

Our fee is flat and published per transaction. It does not scale with the price you agree to pay.

See the data →

10

Any tonne can be challenged in public. The challenge and its outcome stay on the record either way.

See the data →

11

When we get it wrong we void the tonnes and say so at the top of the page, permanently.

See the data →

02 / MEASUREMENT

How a tonne is measured, from field to certificate.

Six steps. Each one leaves an artifact that outlives the people who made it, and each artifact is readable without asking us for permission.

STEP 1

Field registered

Boundary polygon, soil series, baseline cores and the operator agreement are filed before any rock moves.

BASELINE-CORE-SET

STEP 2

Rock applied

Tonnage, grain-size distribution and application date logged against the field, per spreading pass.

APPLICATION-LOG

STEP 3

Samples taken

Fixed depth, fixed grid, fixed method. Every batch carries its own chain-of-custody record.

BATCH-CUSTODY-CHAIN

STEP 4

Lab run

Named lab, named instrument, run identifier. Duplicate and disagreeing runs are published too.

RAW-LAB-EXPORT

STEP 5

Model applied

Published parameters, versioned. The version that issued the tonne is stamped on the tonne.

MODEL-V4.2

STEP 6

Certificate issued

Uncertainty deduction applied, then the tonne lists — linked back through every step above.

CERT-PDF + JSON

03 / THE LEDGER

The retirement ledger, publicly readable.

Retirement is a one-way write. There is no unretire function to call, no admin key that restores a tonne, and no private view of this table that differs from the one you are reading.

CERTIFICATE

RETIRED

TONNES

HOLDER

STATUS

SK-2026-0418-A

2026-04-18

2,400 t

Northwind Logistics

RETIRED

SK-2026-0402-C

2026-04-02

860 t

Halden Insurance Group

RETIRED

SK-2026-0311-B

2026-03-11

1,150 t

Marrow & Fen Foods

RETIRED

SK-2026-0227-A

2026-02-27

318 t

Cartwright Steel

RETIRED

SK-2025-1119-D

2025-11-19

1,140 t

Basalt Field 07 — sampling error

VOIDED

Read the whole ledger →

04 / REFUSALS

What we refuse to list, and why.

We do not list avoided emissions, cookstoves, forestry offsets, or anything where the counterfactual is an argument rather than a measurement.

The test is simple and we apply it the same way every time: if the tonne only exists because of a story about what would otherwise have happened, it is not a tonne we can hand you evidence for.

Avoided emissions

The counterfactual is unobservable. There is no sample to take.

Cookstoves

Usage estimates drift, and nothing in the field can be re-measured later.

Forestry offsets

Reversal risk is real and the buffer pool is an accounting device, not a measurement.

Forward credits

A promise is not a tonne. Sample first, issue after.

Unnamed labs

If we cannot name the instrument and the run, we cannot defend the number.

05 / VERIFICATION

Third-party verifiers, and the conflicts they carry.

Nobody in this industry is unconflicted. Pretending otherwise is the failure mode, so we publish each engagement, who pays for it, and what the verifier had to recuse themselves from.

VERIFIER

SCOPE

DISCLOSED CONFLICT

Meridian Soil Assay

Engagement file →

Independent re-sampling on 20 percent of batches, unannounced, at our cost.

Also runs paid analysis for two operators listed here. Both engagements are named in the file.

Kolbe & Ash Metrology

Engagement file →

Instrument calibration audits and blind duplicate submission across all partner labs.

A partner holds equity in an aggregator that buys tonnes here. Recused from pricing work.

Delta Weathering Review

Engagement file →

Model parameter review and annual re-derivation of the uncertainty deduction.

Authored two of the parameters they now review. Disclosed since 2024, second reviewer required.

06 / PRICE

Price, spread, and where the money goes.

Tonnes clear between 174 and 240 dollars. Our take is a flat 4 percent, published per transaction. 82 percent reaches the field operator.

$174

LOWEST CLEARING PRICE, LAST 90 DAYS

$240

HIGHEST CLEARING PRICE, LAST 90 DAYS

82%

Field operator — spreading, haulage, sampling labour

14%

Lab work, model review and independent re-sampling

4%

Sinkstone, flat, published on every transaction

07 / FAILURES

Failure cases we have published.

In 2025 we voided 1,140 tonnes from one site after a sampling error we caught ourselves. The write-up is on the registry, top of the page, permanently.

NOVEMBER 2025

1,140 tonnes voided

A sampling error at Basalt Field 07 put the cores at the wrong depth for two batches. We caught it ourselves, voided every affected tonne, and refunded at the original clearing price.

Read the write-up →

JUNE 2025

Lab disagreement published

Two labs returned cation results 19 percent apart on the same batch. We published both, held issuance for eleven weeks, and re-ran under a third lab before anything listed.

Read the write-up →

FEBRUARY 2024

Operator delisted

An operator could not produce chain-of-custody records for four batches. We delisted them, voided the unretired tonnes and named the fields involved.

Read the write-up →

08 / OPEN ACCESS

Open the registry. Check the eleven.

No account, no sales call, no gated data room. Every field, batch, lab run and retirement is readable right now — including the ones that went wrong.

SINKSTONE · ENHANCED ROCK WEATHERING · EVERY TONNE, EVERY SAMPLE, ON THE RECORD