How should you check the inconsistent uncertainty figures at 10 N·m in Annex A?
Applies to: ISO 6789-2:2017 · Updated 2026-10-28
Technical review: Alston Hsiao (Chieh-Jen Hsiao), CTO, STAND
Recalculate from the components and formula, then cross-check the tables: the Annex A result at 10 N·m should be W = 1.160 %, rather than the 1.660 % printed in Table A.15. This is the conclusion of our check against the 2017 text, not an official corrigendum issued by ISO.
Identify which quantity is inconsistent
Annex A provides an informative worked example for an indicating torque tool. At the same 10 N·m calibration point, Table A.14 gives w = 0.580 % and W = 1.160 %, while Table A.15 gives W as 1.660 %. The calibration point has not changed, so the difference cannot be explained as two measurements made under different conditions.
Keep the three quantities distinct when reading the tables. Here, w is the combined standard measurement uncertainty, W is the expanded measurement uncertainty, and W′ also includes the mean measurement error and the device measurement error. Confusing columns for different symbols can make a clear numerical inconsistency look like a difference in definitions.
Recalculate the original example from its components
The values below are only those needed to locate this inconsistency in the original example; they are not invented teaching data. Using Table A.13 and the formula for an indicating torque tool, the resolution term must be counted twice:
w = √[(0.150 / 2)² + 2 × 0.029² + 0.304² + 0.396² + 0.092² + 0.255² + 0.080²] %
≈ 0.580024 %.
First round to w = 0.580 %, then apply k = 2 to obtain W = 2 × 0.580 % = 1.160 %. This agrees with Table A.14. Omitting the factor of 2 before the resolution term means the formula no longer matches this indicating torque tool example.
Next, cross-check using the other terms in Table A.15: W′ = 0.654 + 1.160 + 0.100 = 1.914 %, matching the interval reported in that table. Using 1.660 % instead would give a sum of 2.414 %, which does not produce the tabulated W′. This provides a reproducible check beyond simply comparing the two printed values.
Recalculate before adopting any corrected value
Recognizing that a printed value is wrong does not mean that any replacement value is correct. Whatever the source of a proposed correction, recalculate it from the formula and components above and cross-check it against Table A.14 and the W′ sum; adopt it only when all three agree.
If you adopt 1.160 % in an internal calculation record, retain the original table reference, formula, components, and recalculated result so that a reviewer can follow the reasoning. This check covers only the specified calibration point. It does not establish that every final digit throughout the annex has been checked, and this article should not be labeled an official corrigendum.