Ordinals number satoshis by convention, not by consensus

Inherited archive entry · retained from the earlier edition. Its factual, legal and policy claims have not all been reverified for 19 September 2026. Research scope.

A numbering convention, released as software

Version 0.4.0 of the ord software, which implements what its documentation calls ordinal theory, was published on 20 January 2023, according to its GitHub release record. Ordinal theory assigns a sequential number to every individual satoshi, bitcoin's smallest unit, of which one hundred million make a bitcoin, and tracks that number as the satoshi moves between transactions. An inscription attaches content to one specific numbered satoshi, giving that content an identity distinct from the token-style balances this archive covers elsewhere on other chains.

Where the data actually sits

The mechanism is described precisely in the inscriptions documentation: content is stored on-chain inside a taproot script-path spend, pushed as data within an unexecuted conditional branch, an 'envelope' opened with OP_FALSE OP_IF and closed with OP_ENDIF, so the data behaves as a script no-op rather than an executable instruction. A two-step process creates one: a commit transaction locks funds to a taproot output containing the inscription script, and a reveal transaction spends that output, making the content visible on-chain. The inscription's identity is the reveal transaction's id plus an index, nothing more exotic than that.

The boundary: consensus validates the data, an indexer assigns the meaning

This is the distinction the documentation itself draws, and it matters more than the mechanism's novelty: Bitcoin's consensus rules validate that this is an ordinary, correctly formed transaction, and nothing in the protocol enforces ordinal numbering or inscription ordering as such. That work is done by ord, described as an off-chain indexer that tracks satoshi provenance and assigns inscription numbers by convention. As an editorial inference from that indexer-based design, rather than a scenario the documentation itself models: two indexers applying a different counting rule, or disagreeing about which of two conflicting inscriptions came first during a reorganisation, could produce different answers to 'what is inscription number one' from identical chain data, because the numbering is a convention layered on top of validated transactions, not a rule the transactions themselves encode.

Questions to carry forward

• If a claimed inscription number depends on one indexer's convention, what happens to that number if a competing indexer, or a future version of ord, orders inscriptions differently?
• Does a marketplace listing an inscription's provenance cite the reveal transaction directly, or does it rely entirely on an indexer's derived number?
• What happens to an inscription's claimed identity during a chain reorganisation that reorders the block containing its reveal transaction?

The chain enforces that a transaction is valid; it does not enforce that any given inscription is 'the first' or 'number one' of anything. That gap between validated data and assigned meaning is the detail worth carrying into any claim about ordinal provenance.