Corrections
Roughly twenty-five errors turned up across thirteen publishers while this data was compiled — two of them in the source our own ASME B36.10M-2015 rows are parsed from.
Errors Found in Published Sources
Sizes below are given as nominal pipe size (NPS) or DN; the figures at issue are outside diameters (OD), wall thicknesses and masses per unit length. Publishers are described rather than named: the claim is about a number, not about a company.
Some of these are errors inside standards documents, not in charts copying them. That is the whole case for checking arithmetic rather than counting how many sources agree.
| What was wrong | Where | How it was caught | What we publish now |
|---|---|---|---|
| Outside diameter of NPS ⅜ given as 15.9 mm | A chart headed ASME B36.10M, on a site ranking on the first page of Google for the head term | It disagreed with five independent publishers | 17.1 mm |
| Outside diameter of NPS 5 given as 139.7 mm | The same chart | 139.7 mm is the EN 10220 value — a European figure printed in a table labelled American. Five publishers give the ASME one | 141.3 mm |
| Schedule columns shifted one place right at NPS 4 and NPS 5 | A piping-materials reference | ASME publishes no schedule 140 at NPS 4 at all. Adjudicated two to one against the source by two other publishers | Schedule 120 = 0.438 in, schedule 160 = 0.531 in, XXS = 0.674 in |
| Wall thickness of 2.70 mm at NPS 26, schedule 20 | The source the ASME rows here are parsed from | The mass formula: the row’s own 136.30 lb/ft resolves at 0.500 in, which is 12.70 mm | 12.70 mm, carried in the data with a correction note |
| Wall thickness of 1.1250 in at NPS 18 | The same source | The mass formula: 31.75 mm is 1.2500 in, and the row’s 223.82 lb/ft matches 1.2500 | 1.2500 in. NPS 18 also has a legitimate 1.125 in wall at 28.58 mm, so the fix was keyed on both units — a blanket find-and-replace would have corrupted a real row |
| A published mass of 233.72 kg/m for a 402 mm outside diameter with a 24 mm wall | GB/T 17395 itself | The formula and both neighbouring rows give 223.72 kg/m | 223.72 kg/m, with the misprint recorded against the row |
| The bracketed imperial cross-reference read as the Chinese nominal size — 60.3 mm where the standard means 60 | English-language reproductions of GB/T 17395 | 2,485 published mass cells re-derived: 2,462 of them, 99.07%, match the un-bracketed integer. The standard’s own text says the imperial size is not recommended | The un-bracketed value as the Chinese nominal size, with the ISO figure shown as the cross-reference it is |
| EN 10255 DN 150 given as 168.3 mm | An engineering reference back-filling the EN 10220 value | Five sources give 165.1 mm, and the GOST 3262 series independently lands in the same place | 165.1 mm |
| Superseded BS 1387:1985 and DIN 2440 walls — 2.65, 3.25, 4.05, 4.85 mm — printed under an EN 10255 heading | Several UK stockists | Mass arithmetic: a DN 25 medium tube published at 2.41 kg/m resolves at a 3.2 mm wall (2.407), not at 3.25 mm (2.440) | The current EN 10255 walls. Legacy walls are not republished under a current name |
| Every wall thickness in one GOST 10704 table attributed to the neighbouring column | An online copy of the standard | The mass formula. Read cell by cell the shifted output looked entirely plausible | The corrected pairing |
| An outside diameter printed as 88 mm | A GOST table | Every mass in the row resolves at 83 mm — optical character recognition damage, not a real size | Nothing. There is no 88 mm size here |
| Eight single-digit typos in GOST 8734 and GOST 10704 | Online copies of both standards | The mass formula | The rows, with a visible “could not be verified” label. No upstream correction exists |
| A wall thickness of 45.3 mm where the table’s own ratio gives 400 ÷ 26 = 15.3 mm | An engineering reference | Recomputing the wall from the ratio that same table publishes | Nothing. The figure was rejected rather than reconciled |
| US HDPE DIPS walls for DR 17 to 32.5 that fail the source’s own wall-from-diameter relationship | The only source found for that series | Recomputing wall thickness from outside diameter and dimension ratio | Nothing. The series stays out until a second source exists |
| Type L copper at 4 inch given a 0.114 in wall | A widely-mirrored engineering reference | Its own weight column. All three sources print 5.38 lb/ft, and only a 0.110 in wall reproduces it — 0.114 in gives 5.568 | 0.110 in, which two other publishers also give |
| Extra-heavy cast iron given the service-weight outside diameters | A cast iron soil pipe catalogue, on all eight sizes it lists | Geometry rather than a head-count: the majority diameter minus twice the agreed wall reproduces the bore a separate publisher lists, and the dissenting one does not. At 3 inch, 3.50 − 2 × 0.25 = 3.00 against 3.30 − 2 × 0.25 = 2.80 | 3.50 in at 3 inch, and correspondingly at the other seven sizes |
| An HDPE table with its 12-inch row dropped, sliding every wall up a size | A pipe manufacturer’s dimension chart, refetched in 2026 | The dimension ratio. The diameters are right, so bore and mass still reconcile with each other and a diameter check passes — only wall = OD ÷ DR exposes it. Fourteen cells, each 12–14% thin | Rejected outright. The same publisher had already been rejected here once for the same fault |
| A column headed SDR 7.6 containing SDR 7.4 data | An HDPE manufacturer, at all 22 sizes in the column | Re-deriving each column’s actual ratio and comparing it against its own heading — median 7.31. No mass check can catch a wrong label | The numbers are sound and the heading is not. Carried under the ratio they actually have |
| An ISO 4427 table that is a calculator rather than a standard | A plastics supplier | Every cell is exactly diameter ÷ ratio to two decimals — no rounding onto the ISO wall series and no minimum-wall floor, including a 20 mm pipe given a 0.61 mm wall nobody makes | Rejected. It passes the arithmetic gate perfectly while being wrong about the standard |
| A flange pressure class filed under the wrong heading | A flange supplier, on all twelve rows of one class | Its own page. The heading reads Class 600 and a note beneath it opens “From Class 400+…”, which a backwards text search reaches first. Every value was internally consistent and every geometric gate passed | Class 600. Headings are now read from heading elements only, never from nearby prose |
| Eighty-two flange rows missing one table cell each | A flange stockist, across five pressure classes | The bolt-hole cell’s own shape. A dropped cell slides every column after the bore one place left; the surviving 8-0.88 pattern proves the alignment up to it and breaks after | Those rows are read only as far as the cell whose format can be checked |
| ASME B16.9-2018 printing 355.4 mm for a 14-inch end | The standard itself — Table 6.1-11, the 22 × 14 reducer | Arithmetic. 14.00 in × 25.4 = 355.6, and every other appearance of NPS 14 in the same document, including a table three pages earlier, prints 355.6 | 355.6 mm. It is the only ASME fitting row held back on a metric mismatch |
| A standard whose PDF hides its own numbers from anyone extracting it | CISPI 301, Table 1 — the hubless cast iron barrel columns | The glyphs’ fill colour. The table is decimal-aligned with a filler “1” painted in white: invisible on the page, present in the text layer, so an ordinary extraction reads the 1½-inch bore of 1.50 as 11.50 and the 2-inch diameter of 2.35 as 12.35 | The printed values are correct. The corrupted extract was in this repository before the correct one, and glyphs are now dropped by colour — not by a rule about large numbers, since 10, 12 and 15 legitimately begin with a 1 |
The GOST typos are the ones that cannot be closed by looking again. Every accessible copy of GOST 8734 online is the same scan, so re-checking returns the same typo and reads like confirmation. GOST 8732, by contrast, was reissued in 2025 with all eleven of its deviations corrected; 8734 has had no such reissue. Those rows stay labelled rather than quietly changed to a value no source supports.
Corrections to Our Own Data
A corrections page that only lists other people’s mistakes is marketing. These are ours. Most of them produced output that looked entirely plausible: the missing sizes were invisible until the data was checked against the standard itself, and eight publishers agreed with us about rows we had never loaded. That is why arithmetic, not agreement, is treated as the test.
| What was wrong | Where | How it was caught | What we publish now |
|---|---|---|---|
| Four nominal sizes missing entirely — NPS 1¼, 1½, 2½ and 3½, forty rows | Our own parser | Checking coverage against the standard itself. The source writes 1-1/2 (40) with a hyphen and our size pattern had none, so those rows were dropped silently | All four sizes. DN 65 — 2½ inch — carries the largest JIS-versus-ASME divergence of any size, so the hole would have been conspicuous |
| Every wall thickness from one source parsed a thousand times too large — .109 read as 109.0 | Our own parser | A 0% agreement rate against every other source. The number pattern demanded a digit before the decimal point, and that source omits leading zeros | Correct values. This bug announced itself; the missing-sizes bug above did not, which is the argument for checking arithmetic rather than agreement |
| 732 rows silently collapsed to 223 | Our own row keys | Rows were keyed on nominal size and schedule, but 530 ASME walls carry no schedule number at all, so they overwrote one another | Rows keyed on the wall thickness. Every row survives |
| Table figures read by a language model | Our earlier method | Three separate failures: rows returned for sizes the page did not contain; schedule columns mis-mapped; and the ASME PDF read down its wall-thickness column instead of across the schedule row, returning 0.5 in for NPS 24 schedule 40 where the standard says 0.688 | Nothing from that method survives. Every table is parsed deterministically and no language model reads a number |
| 21 conflicts with a cross-check source that were not conflicts | Our own comparison | Every difference was exactly 0.001 in, and every value an exact thirty-second of an inch: one side rounds half up, the other truncates | The exact fraction — 7/16 in (0.4375) — rather than either party’s lossy three-decimal figure |
| 10 of 16 rows appeared to conflict with a two-decimal source | Our own comparison tolerance | Comparing at three decimals against a source that publishes at two manufactures false conflicts. At its own precision all 16 agree | Sources compared at their own published precision |
| Schedule 40 and STD treated as one column | A cross-check source’s combined heading, and our reading of it | Eleven apparent conflicts resolved to a single cause: above NPS 10 the STD wall is not schedule 40. At NPS 12, STD is 0.375 in and schedule 40 is 0.406 in | The two listed separately. STD equals schedule 40 through NPS 10 and XS equals schedule 80 through NPS 8; above those they diverge permanently |
Every Row Carries a Publication Status
Corrections are only half of the policy. The other half is that no row is allowed to appear without saying how far it has been checked. Every row in the dataset carries a publish_as value, and that value governs how the row may be shown. The rule and the standing list of known-unverifiable items travel inside the data files themselves, not in a document beside them.
- verified
- Confirmed against the primary standard, or against two genuinely independent publishers, and it passes the mass formula. Published plainly.
- computed
- The dimensions are verified but the published mass failed the formula. Our computed value is shown and the published mass is never quoted.
- single_source
- One publisher, arithmetically self-consistent. Usable, and labelled as such.
- unverified
- Could not be confirmed. Shown with a visible “could not be verified” label, or omitted. Never stated as fact.
In the main chart, 766 of 772 rows are verified against ASME B36.10M-2015 itself and 6 could not be. The full accounting, including the figures across every other standard that could not be confirmed, is in how this data was verified.
How to Report an Error
Reports are welcome and they are how this page grows. A usable one names the standard and its edition, the table or clause, the size, and the value you believe it should be — what makes a report checkable is stated on what this reference is and who builds it.
Send it to info@pipesizechart.com. Every report is tested the way the original data was: against the standard’s own mass formula, and against the standard’s text where it can be obtained. If it holds, the row changes and the change appears on this page with a date. Nothing is edited silently, and no correction is removed once published. Data revision: .