Method
How we verify
A wrong torque figure is not a typo. Here is the process, and where it stops.
Two ways a chart can be right
Some tables are definitions. American Wire Gauge is a geometric series anchored at two diameters, and a fractional drill size is the fraction, so every figure can be recomputed from first principles. For those, our build recomputes each value and fails if any one drifts. 3 of the charts here are published on that basis.
Most tables are not definitions. Bolt torque, thread pitch, O-ring dimensions and NEC ampacity are published documents that have to be read and transcribed. Those are published only after a person has read the standard back against the table, line by line, and signed the date. 0 charts are published on that basis today, and 10 are drafted and held waiting for it.
Sources are mapped to columns, not to tables
A wire table takes its diameters from one document and its resistance from another. One citation under both would be an overstatement about one of them, so each column names its own source, and the source block on every page lists which columns each document covers. The build fails if a column has no source, or names one the page does not define, which is the only reliable defence against a column being added later without one.
What the machine checks catch
- A column with no source behind it.
- A citation nobody uses, which reads as coverage the table does not have.
- A value that breaks a monotonic run, such as torque falling as diameter grows.
- Inch and millimetre columns that disagree.
- A gap in a series, and a control on the page with no data behind it.
What they cannot catch, and why it matters
A draft of our O-ring chart converted inches to millimetres perfectly on every row and still disagreed with AS568 on every row: the dash numbers had shifted by one, so each size carried the dimensions of its neighbour. Every check we had passed. Consistency is not correctness, and no amount of automation reaches this, which is why a transcribed table waits for a person rather than for a test.
Every standard we cite
14 published documents, with the charts that draw on each one. The link on every designation goes to the publisher's own page, which is the page that stays correct about which revision is current. Where a source has no publisher page, the row says so rather than leaving a gap.
| Standard | Publisher | Used by |
|---|---|---|
| ANSI/ASME B94.11M (1993) ↗ | ASME |
|
| ASME B1.1 ↗ | ASME |
|
| ASME B18.2.2 ↗ | ASME |
|
| ASTM B258-18 ↗ | ASTM International | |
| AWG (Brown & Sharpe) for aluminium sheet No publisher page. | n/a |
|
| ISO 235 (2016) ↗ | ISO |
|
| ISO 261 (1998) ↗ | ISO |
|
| Manufacturers Standard Gauge No publisher page. | n/a |
|
| NBS Circular 31, Copper Wire Tables (3rd edition, 1914) ↗ | National Bureau of Standards (now NIST) | |
| NEC 240.4(D) ↗ | NFPA |
|
| NEC Chapter 9, Table 8 ↗ | NFPA | |
| NEC Table 310.16 ↗ | NFPA |
|
| SAE AS568 ↗ | SAE International |
|
| SAE J429 (J429_201405) ↗ | SAE International |
|
Standards are copyrighted documents, so we cannot host them. Most publishers sell access; NFPA offers free read-only access to the National Electrical Code through its own site. We link, quote the clause we relied on, and record the date we read it.
Dates
Every chart shows when it was last checked. None shows when it will be checked next: a future date is accurate only until the day it arrives unmet, and then it says the opposite of what it was put there for. The date of the last check is a fact and it is the one we publish.
When the manufacturer disagrees with us
Follow the manufacturer. A reference sheet gives you the general figure for a class of fastener or conductor; the instructions that came with the part account for the joint, the material and the assembly. Where they conflict, theirs governs.