A better test point. The same obligation to prove it.
Kößling and Tajmar report a linear interval from 10 nN to 300 µN and 11.9% calculated error at 10 nN. Our old 1 W ideal signal falls below that demonstrated interval. The proposed 2 W signal lies inside it, making it a better starting point for a host-lab discussion. This is not a universal 10 nN detection floor, a supplier commitment or a transferable uncertainty certificate. Read the balance paper ↗
Keep the reference experiment honest
Shaw and colleagues used a 1363 nm switched optical arrangement. Their single-reflection table gives a mean differential power ΔPᵦ of 2.31293 W and measured electrostatic force of 16.59 nN. The differential power combines the two switched paths; it is not a single-beam rating. Their optical prediction and instrument-specific uncertainty must not be transplanted into our 1064 nm design. NIST manuscript, Tables 3–4 ↗
Power calibration is necessary, but not sufficient
SP 250-77 describes calibrations at 2 W and above near 1.06 and 10.6 µm. That supports discussing this operating point with a calibration provider. Detector spatial response, vacuum compatibility, aperture, transfer optics and coverage factors still need an actual calibration chain. SP 250-77 ↗ · Current NIST service information ↗
| Standard uncertainty term | Allocation at 2 W | Status |
|---|---|---|
| Optical | 0.400277 nN | Proposed; not demonstrated |
| Calibration | 0.667128 nN | Proposed; not demonstrated |
| Statistical | 0.119523 nN | Proposed; not demonstrated |
| Thermal | 0.150000 nN | Proposed; not demonstrated |
| Electrostatic | 0.080000 nN | Proposed; not demonstrated |
| Gas | 0.080000 nN | Proposed; not demonstrated |
| Mechanical | 0.080000 nN | Proposed; not demonstrated |
The omitted terms change the total
The supplied audit code’s 1.574 nN total includes only statistics, optical power and calibration. Retaining thermal, electrical, gas and mechanical residual allowances gives U = 1.626 nN (12.19%). With an illustrative thermal/mechanical correlation of 0.6, U becomes 1.644 nN. Unknown terms cannot be assigned zero. The 10% stretch objective is not met by these assumptions.
Gas is not automatically negligible
The supplied code computes A p/(4T) correctly but its explanatory notes are 1,000 times too small. For 1 cm² at 300 K, the illustrative slope is 83.333 pN/K at 10⁻³ Pa, and 83.333 nN/K at 1 Pa. Even the lower value is comparable with the 0.08 nN gas allocation for a 1 K gradient. Geometry and surface accommodation matter; measure pressure dependence and thermal transfer instead of declaring a pressure threshold sufficient. Thermal-gradient force study ↗
Seventy blocks are a proposal, not a timer
For equally spaced windows, Dᵢ = (A₁ − B₁ − B₂ + A₂)/2 is the full on/off difference. Seventy independent blocks with 1 nN window noise yield 0.1195 nN statistical standard uncertainty. ABBA cancels linear drift; it does not cancel arbitrary thermal transients or correlated noise. Commission settling, counterbalance ABBA/BAAB and freeze the analysis before decisive data.
A pass needs two conditions
Require signed agreement |Dmeasured − Doptical| ≤ Udifference and Udifference ≤ 0.15 |Doptical|, with all nuisance controls passed. For the ideal 2 W case, that ceiling is 2.001 nN. Use the measured optical prediction and combined covariance in the real test. A large interval cannot manufacture a pass. Repeat across three days and controlled reassembly.
Five ways to challenge a convenient result
Matched-heater transfer; beam-off electrical switching and a grounded dummy; a pressure sweep; pump/tilt/cable tests; and optical null, reversal and power-linearity checks. Each must produce a measured correction or defensible bound. A failure leads to diagnosis or an incomplete result, not a new-field claim.
What the recomputation actually establishes
The attached Edition 08 script contains 22 explicit site-value comparisons; all 22 pass its stated tolerance. It also contains interpretive claims that arithmetic cannot establish. Its score vector gives M3 = 89 before caps and 80 after caps; M4 = 90. No replacement vector supporting 90.5 was supplied. This site does not use a credibility score as experimental evidence. The new kit includes a separate standard-library calculation with 12 numerical and estimator checks.
The next commitments
Appoint a qualified experimental lead and independent reviewer. Commission a paid feasibility review and comparable host-lab/owned-system quotations. Resolve the open design fields, then register the frozen protocol before confirmatory acquisition. The existing owned-system allowances remain visible; the newly quoted host-lab totals cannot be certified without their line items. Neither procurement nor laboratory access is assumed.
| Release item | Current state |
|---|---|
| Experimental lead / independent reviewer | Not appointed |
| Apparatus / experimental dataset | Not qualified / none |
| Protocol | Draft 08; not registered |
| Zenodo DOI / CODECHECK | Not deposited / not certified |
| Source and package integrity | Reproducible calculations and SHA-256 file manifest |
| Hosting | Review Site; external-domain ZIP supplied separately |
Choose the submission from the evidence
NSF 26-511 lists 4 November 2026 and 4 March 2027 as full-proposal deadlines. Phase I requires an official Project Pitch invitation; a named PI alone does not establish eligibility. Instrumentation is a potential fit only if the company can substantiate technical novelty, customer need and program requirements. Official solicitation ↗ IEPC’s organizer lists Pasadena, 12–17 September 2027, with call details forthcoming; the claimed early-2027 call date is not yet confirmed. Organizer notice ↗
A version is not a validation
Edition 08 is a website and commissioning revision. The v0.2 scientific model remains available, with its scope intact. No DOI, preregistration, external review, grant award, conference acceptance or new propulsion result is implied. The status record preserves missing appointments as null rather than filling them with presumed collaborators.