A second instrument.
A second team.

Independent apparatus, independent judgment.

Use a separate team, balance and location. Record shared software, designs and standards; they can create common errors even when instruments differ. Payment must not depend on the outcome. Agree in advance to publish null results, discrepancies, exclusions and raw data.

Dresden’s published balance range supports a feasibility discussion near 13 nN. Its reported noise and error are not a certificate for Metric Flight’s setup. A provider’s willingness, schedule, uncertainty and price require a written response.

Predict each team’s actual force.

For normal incidence and collinear transmitted/reflected beams, F = (1 + R − T)P/c. For an opaque mirror, F = (1 + R)P/c. Absorbed light transfers momentum too: multiplying 2P/c by reflectivity alone is incorrect. The ideal 2 W value is 13.343 nN; actual power, scatter, clipping and angles require a measured vector momentum ledger.

Compare closure residuals.

For each team, r = Fmeasured − Fpredicted. Require |r| ≤ U(r) and U(r) ≤ 0.15|Fpredicted|, plus all control and repeatability gates. Compare rA − rB using u² = uA² + uB² − 2 cov(rA,rB), with justified coverage. Shared calibration can correlate results. Never use the near-zero residual as the denominator of a relative uncertainty requirement.

The current seven-term allocation gives U = 1.626 nN. Force calibration (0.667 nN standard allocation) exceeds optical prediction (0.400 nN); optical calibration is not the largest term in this budget. Literature nuisance bounds do not transfer automatically to the new instrument.

Freeze before acquiring decisive data.

Both teams must approve geometry, serial numbers, traceability, corrections, covariance, controls, stopping/exclusion rules and the comparison criterion before unblinding. Secondary linearity tolerances remain unresolved. Registration and signed review are prerequisites, not completed outcomes.

Deliver a reproducible record.

Release protocol/code hashes, raw measurements, calibration certificates, environmental records, deviations and an independent report. An empty ledger template is supplied; synthetic calculations are never presented as observations.