# Build handoff: T560 Built **Masur–Veech volumes of Abelian strata**, https://numberdb.org/T560, permanent slug `Masur_Veech_volumes_of_Abelian_strata`. The draft was claimed empty, then filled once with 25 entries. It is unpublished and has not been offered, reviewed, or published by this run. The controller must arrange the independent critique, argument-selection sidecar and any repair before offering it. No independent critique acceptance is asserted here. ## Scope and policy Family numberdb-data#224 was read before work. The live corpus screening found no duplicate family. The expected H(2) volume matches T53's zeta(2,2) row by a mathematical identity; it is linked and is not declared a repeated construction. T459 is linked as another moduli-space volume table, using its fetched slug. The coordinated canonical skill manifest is `agents/runs/20261027T201044Z-build-skills/manifest.json`. The archived local and live copies were both read, hashed and compared. Each is SHA-256 `e37efb7e2ededf6c1d7134c673f152f648f26e0267c556aebdf3ec65765d8b4c`. `skill-evidence.json` records both versions. The source snapshot's canonical file had the same digest. The argument justification is `argument-selection.md`. Every positive partition of 2g-2 for g=2,3,4 occurs, with only its classified components: 18 profiles and 25 rows. The finite cutoff favors complete low-genus breadth across minimal, principal and intermediate strata and all five component labels. Genus/profile are discrete indices; component is intrinsic. There are no continuous reference inputs or probability parameters. Default enumeration, parameter order (g, mu, component), stored order and completeness were checked. ## Conventions and decisions All zeros are labelled, including equal-order zeros. The integral relative-period lattice has covolume one, area is one, phase is retained, and hypersurface volume is 2D times cone volume, where D=2g+n-1. No regular marked points are added. Even/odd spin components exclude the hyperelliptic component. A factor-two discrepancy was investigated rather than chosen by appearance: Eskin–Zorich's 2015 displayed two-zero hyperelliptic formula has prefactor 4, whereas the labelled formula in CMSZ section 6.5 has 8. AEZ Remark 1.2 explicitly explains the 2! from labelling the two zeros and gives H(1,1)=pi^4/135; EMZ Table 1 agrees. Prefactor 4 gives the volume after forgetting those two labels. The stored values and displayed formula use 8. See `sources.json`, `AEZ.txt`, and `CMSZ.txt` for the source trail. This resolves the numerical discrepancy for the chosen convention, without assuming the two differently written formulas agree. The `flat surfaces` tag remains a proposal. Current existing tags `volume` and `period` are used. A search of live tags returned no flat-surfaces tag. The three potential users are this table, **Area Siegel–Veech constants of Abelian strata**, and **Kontsevich–Zorich sum of Lyapunov exponents**, the three proposals in #224. No singleton tag was created. ## Numerical evidence All exact rational coefficients of pi^(2g) were visually checked against EMZ Table 1 (page 28; archived rendering `EMZ-table1.png`). Arb evaluates the exact coefficients at 100 significant digits with 64 guard bits. The least measured relative accuracy is 393 bits. The dry run passed exactness and prose checks; 25 entries have a longest working value of 138 characters and a 6.25 KB measured block including comments. Independent mathematical calculations cover nine distinct rows: all six hyperelliptic values from the AEZ/CMSZ formulas and all six minimal-component values from CMSZ total/spin generating series, overlapping in three rows. The checker uses Python rational coefficient recurrences and Lagrange inversion, not the generator's transcribed coefficients. Controls include nonzero known genus-one volumes and exact exponential/inverse-series coefficients. Twenty-three coefficients were also checked against surface-dynamics' separate zeta-normalized transcription. That code cites EMZ, so this is not an independent mathematical derivation. Principal-stratum ratios for all three genera match EMZ Table 2; these also cover the two rows absent from the software table. General nonhyperelliptic multizero volumes remain exact published-source transcriptions. This limitation is explicit in the table's rigour note and numerical evidence. All 25 written decimal intervals were checked by a separate 512-bit MPFI calculation, including strict finiteness/width tests and a rational enclosure control. The same checks and all displayed identities were repeated on values read from the database. `verify(sample=None)` reports 25/25 matched, zero missing, differing or extra. See `stored-verification.json` and `fill.log`. The site audit returned `findings: []`, `clean: true`; see `audit.json`. ## Reproduction and independent review The draft attachments contain `generate.py`, `checks.py`, `check.py`, `independent.py`, `reference-coefficients.json`, `argument-selection.md`, `numerical-evidence.json`, `sources.json`, `skill-evidence.json`, and the post-fill `stored-verification.json`. The exact source PDFs and source-code snapshot, hashes, rendered table, logs and working scripts remain in this run's artifact directory. No source, skill, pipeline, Git repository, supervisor state, or completion receipt was edited. Run `sage -python checks.py` locally with the attached files. For `--stored` and generator verification while the table is private, provide its authorized NumberDB API key. In this campaign use `agents/sage.sh`, with the relevant files mounted, as shown by the working wrapper scripts. No computation remains running. The controller handoff `agents/runs/20261027T201044Z-build.table.json` names only T560. A read-only argument fingerprint is recorded separately; it is not an independent argument-review sidecar or an approval. The critic must check the live selection and all source dependencies again in its own run.