Sources
Primary sources
This page lists the source records this atlas cites, with exact URLs and the read dates. Six transaction records carry the funding, consolidation, leak deposit, leak spend, reward and #135 story; two BitcoinTalk threads carry the public discovery and the authorship claim; one public dump is kept as a cross-check. Transaction identifiers and URLs are reproduced exactly as published, because a truncated or re-spaced hash is a different object. A source link shows where a claim can be read. It does not prove an identity, and a dump is not a primary record.
How sources are used here
Three habits. First, chain rows carry a block height, a UTC timestamp and an exact transaction URL; the explorer is the current state, and any figure quoted from it is a dated observation. Second, forum rows carry a thread or message permalink, and the label Forum statement — a post is evidence that an account posted. Third, dump rows are cross-checks: they are useful for spotting ranges and published keys, and unreliable for dates, because the same dump may print a 2019 leak date where a 2023 spend date belongs. Where a dump and a transaction disagree, the transaction wins. Links are cited as published; if a page later changes, disappears or shows a stale count, that is a property of the third-party page, not evidence about the puzzle.
Chain records
Six transaction records, each an exact mempool.space URL. These are the on-chain basis of the timeline on /history/ and of the cases on /cases/.
2015-01-15 — Genesis transaction
txid 08389f34c98c606322740c0be6a7125d9860bb8d5cb182c02f98461e5fa6cd15. Block 339085, 2015-01-15 18:07:14 UTC. 256 outputs, 32.896 BTC. Read 2026-08-27 on blockstream.info; the URL below points at mempool.space, and the transaction is the same object on both explorers. This is the funding event the whole puzzle descends from.
2017-07-11 — Consolidation
txid 5d45587cfd1d5b0fb826805541da7d94c61fe432259e68ee26f4a04544384164. Block 475240, 2017-07-11 05:00:53 UTC. 97 inputs, 109 outputs, 103.527 BTC total. Community records describe the destination range as #53–#160 and the emptied range as #161–#256; that grouping is a labelled account, not a per-output verification.
2019-05-16 — 1,000-sat deposits
txid 7c432398c7631600af01695c9767eff109cbfae4f7ecccaff388043a474d4f1e. Block 576251. 21 outputs of 1,000 sat: puzzle #1 plus the twenty leak-series addresses (every fifth from #65 to #160). A receive does not leak a public key, so this transaction is the setup, not the leak.
2019-06-01 — public-key reveal
txid 17e4e323cfbc68d7f0071cad09364e8193eedf8fefbcbd8a21b4b65717a4b3d3. Block 578732. 21-input spend; the scriptSig of each legacy P2PKH input carries a compressed public key. Twenty inputs are the catalog leak series — the twenty addresses, every fifth from #65 to #160. The 21st input is 1PvaqLqRAivje7CactLR55xQBYvBeaDrXN, which is not puzzle #1 and not part of the 2015 256-output set. This is the transaction that made bounded-interval work possible; the “21 leak addresses” shorthand is wrong.
2023-04-16 — reward increase
txid 12f34b58b04dfb0233ce889f674781c0e0c7ba95482cca469125af41a78d13b3. Block 785620, 06:29:48 UTC. 1 input, 85 outputs, 872.19 BTC moved. After this transaction the catalog unit becomes N × 0.1 BTC. The total moved is not a prize figure and not a live balance.
2026-07-28 — #135 spend
txid 817535430c600810080662dd425a2e335f771effd950d28fc443bce4a92fa812. Block 959946, 2026-07-28 08:20:16 UTC. The subsequently published key matches the catalog compressed public key and the P2PKH address 16RGFo6hjq9ym6Pj7N5H7L1NR1rVPJyw2v. The chain part of that statement is the transaction; the key match is arithmetic; the attribution that accompanies it is a separate claim.
Forum records
Two BitcoinTalk threads. Both are public records of what accounts posted. Neither carries a signature, so neither binds a handle to a key.
BitcoinTalk #1306983 — public discovery thread
topic=1306983, https://bitcointalk.org/index.php?topic=1306983.0. The primary forum record of the December 2015 discovery: how the 256 unlabeled outputs were noticed, which low puzzles were already spent, who reported what. It is not a catalog and not an authority on dates: its opening post displays “December 28, 2015, 04:12:47 PM” in the forum clock, while the English timeline carried 2015-12-27. Precursor thread #1305887 (https://bitcointalk.org/index.php?topic=1305887.0) is linked from that opening post and is a precursor, not a replacement.
BitcoinTalk #5201811 — authorship argument
topic=5201811, https://bitcointalk.org/index.php?topic=5201811.0. The thread that argues saatoshi_rising is Satoshi Nakamoto and quotes the 27 April 2017 paragraph about consecutive keys from a deterministic wallet. Cited so it can be read, not endorsed: username resemblance and funding scale are not evidence of identity, and the claim is handled on /theories/ as speculation. Message permalink for the quoted material: https://bitcointalk.org/index.php?topic=1306983.msg18765941.
Reference dump
One dump is kept as a cross-check. It is not a primary record and its dates are not authoritative.
privatekeys.pw puzzle table
privatekeys.pw/puzzles/bitcoin-puzzle-tx, https://privatekeys.pw/puzzles/bitcoin-puzzle-tx. A public dump of addresses, ranges and published keys, useful for cross-checking what was published and when a claim first appeared. Read it with two cautions: its dates are sometimes 2019 leak dates presented next to spends, and its tables are not evidence that a solve happened or that a name is real. Where it conflicts with a transaction, the transaction is authoritative.
Known gaps in the source list
Stated plainly rather than padded. Third-party aggregator pages exist for this puzzle, but their headline totals go stale and are not cited here as counts; if you read one, note the date on the page and treat its totals as that day's snapshot, not as a chain fact. No direct permalink to the saatoshi_rising post renders as an isolated post view: fetching the message anchor returns the surrounding thread page, so the quote is anchored to the thread plus the message anchor. The precursor thread #1305887 is linked from the discovery thread's opening post but was not body-read for this page. Several attribution claims rest on posts quoted second-hand rather than on the original, which is recorded per handle on /people/. A citation here means a claim is traceable, not that it is verified.