HARD PROBLEMS · 2026 · EVIDENCE TRAIL · NO VERIFIED FORMULA OR DISCOVERY
The candidate list became smaller; the number of verified new formulas or textual discoveries remained zero. Variant formula names and passages were organized into a reviewable search set, but the full corpus, prior restoration literature and expert readings have not closed the identification.
Candidate names, manuscript images and the unpublished comparison table stay out of this draft. The subject is the verification path that must separate a search hit from a historical claim.
Candidate extraction
internal list exists
Full-corpus comparison
incomplete
Expert reading
none
Verified discoveries
0
A search hit begins the philological reading
Classical texts combine orthographic variation, variant characters, damaged passages and differences introduced through copying. A matching string does not prove that two passages name the same formula or carry the same meaning. A spelling difference does not rule out a relationship either.
The first pass grouped variants and reduced them to a list a reader could inspect. That list is a queue for returning to editions and contexts. Reducing the search space is internal progress, not external evidence of a discovery.
The largest gap was prior restoration work
A related passage may already be documented in another edition, while a source outside the searched set may contain an earlier use. Without a corpus-wide and prior-work comparison, both novelty and identity remain unstable.
An expert reading is still required because one misread character can change the candidate relationship. There is no external validation, and the internal shortlist cannot be presented as a catalogue of findings.
A well-reduced candidate list creates work that can be checked; it does not make the historical judgment for the checker.
Four checks come before disclosure
The exact holding institution and edition, the digital object’s reuse terms, candidate-level prior scholarship and an expert reading of the passage all need to be established. Visible online access does not itself grant republication rights.
Until those checks are complete, the shortlist and source images remain private. The bounded result is a candidate-reduction workflow whose historical claims are still open.
Project record
- Period: 2026
- Activity: formula-name and passage extraction, variant grouping and prior-work comparison design
- Internal evidence: candidate list and search flow
- External validation: none
- Final outcome: a narrower review set; zero verified new formulas or textual discoveries
- Not established: candidate identity, novelty, reading, corpus coverage or image-reuse rights