In brief
A search result can establish that a library contains relevant material without establishing that it contains the passage a study needs. The latest research workflow now keeps those two tests separate. A blocked commission stays blocked until its named evidence is available, and the public project analysis chooses the next eligible study.
When a useful match is still insufficient
The Alexandria to Syene commission needs an identifiable Cleomedes edition and passage, together with source-bound evidence for the candidate stadion lengths used in its calculation. General matches for Greek measures do not meet that requirement. The library can contain relevant works while this particular study remains unready.
The repair gives the source requirement authority over the readiness result. Generic catalogue hits remain useful finding aids, but they cannot clear the explicit block. The prerequisite check, default selection and progress display now recognise the same blocked state. This prevents a familiar failure: one part of the workflow says that evidence is missing while another presents the work as ready.
A completed study shows what the gate protects
The released Petrie survey study followed a different path. Its source pages were read before drafting, its four side lengths were checked with independent arithmetic, and the account separates reconstructed dimensions from a proposed ancient unit count. Dividing a base by an assumed number of cubits is a conditional calculation; it cannot independently prove the assumed count.
That distinction belongs in the article itself. The technical lesson belongs here: a catalogue record, extracted text and a successful calculation each answer different questions. The source register locates material, rendered pages control the historical claim, and an independent calculation checks what follows from the stated inputs.
Let a blocked record yield without disappearing
The source block records what is missing and how to resume. Default selection skips the blocked record and moves to eligible work instead of repeatedly selecting an unavailable task. Regression checks cover generic matches, an incorrectly queued record, clearing a block, interrupted work and dependencies. These checks protect the workflow; they do not certify an unread historical passage.
The latest published editorial count is seventeen enrichments out of 230 commissions. Accepted individual unit-graphic upgrades remain zero out of 191. Neither fraction substitutes for the other. The next substantive commission is Ancient Measures for Historical Fiction; Alexandria to Syene remains a research dependency until its specific sources are checked.
The Build Log therefore records both a completed release and an honest stop. Keeping the stop visible makes the next action useful without making the evidence look more complete than it is.