Receipts

Pre-registration.

Before any surface was queried, we recorded what would be measured, on what population, by what instrument, and under what rule a result would count. That record was sealed, timestamped, and signed to a permanent append-only ledger.

It cannot be edited. It cannot be edited by us.

The consequence is narrow and it is the entire point: we could not have chosen these findings after seeing them, because the specification existed before the evidence did.

Reproduction — withdrawn.

This page previously stated that the measurement was re-run and reproduced identically, forty times out of forty. That claim was not supported and it has been withdrawn. The figure came from a re-score of a different population, and the run behind the published findings carried no seal at all.

What exists instead is the sealed run indexed above. It is a first measurement of a single subject — this estate — under pre-registration 2097415. It is not a reproduction of the findings and nothing here should be read as confirming them.

That run is also not clean, and the shape of it is stated here rather than summarised away. The sealed counting rule specified three repeats. Four were taken. The third was abandoned after 14 of its 49 prompts and the fourth is missing one. 160 captures were signed where a complete four-pass grid would be 196. Every capture that exists settled, and none was discarded — but a run that reports only the passes that finished is precisely the failure this instrument was built to refuse, so the gaps are on the face of it.

How a third party checks this.

Each record above carries a ledger reference. The ledger is append-only and hash-chained: an entry cannot be altered without breaking every entry recorded after it.

Verification requires no relationship with us and no permission from us. It requires the reference and the chain.

The ledger above is ours. This is not.

The pre-registration for the CAT-A run — the sealed method, fixed before any evidence existed — has SHA-256 a445d8ad3d1f3885a03573d05d1844fe454a12d0e2b33419530b045a2140cd14, the same value recorded on our ledger as event 2167336. That hash is committed to the Bitcoin blockchain via OpenTimestamps, in block 962642, mined 2026-08-15T23:49:56Z.

The CAT-B pre-registration — SHA-256 d0951b1d6e2d72042e912c6d89ba4a330972d32c86644ff5d48ca1c1f8dff75e, ledger event 2168013 — is anchored the same way, in block 962820, mined 2026-08-17T02:43:46Z. Between them the two pre-registrations cover all twenty-three readings.

Download the proofs: xam-prereg-XAMA-001.a445d8ad.ots · xam-prereg-XAMB-001.d0951b1d.ots

Verify it yourself. The proof commits to a SHA-256 digest. You do not need anything from us except the proof file and the digest printed above — both are on this page.

Download a proof, then run, with any OpenTimestamps client:

ots verify -d a445d8ad3d1f3885a03573d05d1844fe454a12d0e2b33419530b045a2140cd14 xam-prereg-XAMA-001.json.ots

and for CAT-B:

ots verify -d d0951b1d6e2d72042e912c6d89ba4a330972d32c86644ff5d48ca1c1f8dff75e xam-prereg-XAMB-001.json.ots

The -d flag verifies the timestamp of a digest rather than of a file. The client checks the proof against the Bitcoin blockchain and tells you the block. Nothing in that exchange involves us.

Why we do not publish the pre-registration file itself. The proof attests the digest of a file, and that file names every subject in the cohort and its domains. Publishing it would identify all twenty-three companies measured — and the entries on this page are anonymised on purpose. We are not going to break that to make a verification step shorter. The digest is the public commitment; it is recorded on our ledger as the events listed above, and it is the value the Bitcoin proof attests. A subject who holds its own reading can be given the pre-registration directly and can then run the file-based check, ots verify xam-prereg-XAMA-001.json.ots, with the file beside the proof.

One limitation, stated rather than hidden. The ots command-line client does not run on the machine that produced these proofs, so we exercised the same verification through the OpenTimestamps Python library and read the Bitcoin block headers from two independent public explorers. That is a limitation of our host, not of the proof. These are standard detached .ots files and any working client will read them.

What this does and does not prove. It proves the sealed method existed before that block was mined. It says nothing about whether the method is any good — that is what the rest of this page is for. A timestamp that precedes the data cannot be authored after the fact, and that is the whole of its claim.