How This Data Was Verified

Where the Numbers Come From

Sixteen standards systems were compiled. Each was parsed programmatically from the source document or page — never read and retyped, and never extracted by a language model, both of which introduce errors that are invisible on review.

Every Figure Is Checked Against the Standard’s Own Formula

Each standard publishes a formula relating outside diameter, wall thickness and mass per unit length:

M = (OD − t) × t × C

Recomputing every published mass from that formula tests all three numbers against each other at once. A mistyped wall thickness produces a mass that no longer matches, and the row fails. The constant appears verbatim inside ASME B36.10M, ISO 4200 clause 4, GOST 8732 and IS 3589 clause 11.1.1 — it is the standards’ own arithmetic, not ours.

What That Check Has Caught

• Eight suspect cells in IS 3589. Indian Standard Amendments 1 to 3 later corrected exactly those eight and no others.
• Two transcription errors in a primary source, both confirmed against two independent columns before being corrected.
• A column shift in a GOST table where every wall thickness was attributed to the neighbouring column — output that looked entirely plausible.
• An outside diameter printed as 88 mm whose masses all resolve at 83 mm: optical character recognition damage, not a real size.
• Two incorrect outside diameters on a chart currently ranking on the first page of Google, one of which is a European value printed in a table labelled as American.

The Constant in That Formula Is Not One Number

C is a material constant, not a universal one. It is π times the density of the metal, scaled to whichever units the formula is being worked in. For carbon steel at 7.85 g/cm³ it is 0.02466 kilograms per metre with the dimensions in millimetres, or 10.69 pounds per foot with them in inches. For ductile iron, at 7.05 g/cm³, it is 0.0221482.

Using the steel constant on an iron pipe overstates its mass by 11.3% — and it does so silently, because the arithmetic remains internally consistent. The row still passes its own check; the answer is still wrong. That is the one failure mode this method has, and the defence against it is that the material travels with the row rather than being assumed. Stainless moves the other way: at roughly 8.0 g/cm³ it is about 2% heavier than the carbon steel figures, which is why the stainless table here is kept behind its own accessor instead of being mixed into the carbon steel one. Every figure on the pipe weight chart names the material it was computed for, and nothing on this site converts a mass from one material to another.

Precision Has to Match the Source

A comparison is only as meaningful as the precision both sides were published at. One engineering reference publishes wall thicknesses to two decimal places. Compared at three, it appeared to contradict the primary table on ten of sixteen rows. Every one of the ten rounds correctly — 0.095 to 0.10, 0.938 to 0.94 — and at the precision that source actually claims, the agreement is sixteen of sixteen. Comparing at a precision a source never claimed manufactures conflicts that do not exist, and then invites you to resolve them by picking a winner.

The same trap runs in the other direction. Twenty-one apparent conflicts with a widely cited reference all differed by exactly 0.001 inch, and every one of them was an exact thirty-second: 7/16 is 0.4375, which one publisher rounds half-up to 0.438 and another truncates to 0.437. Neither is wrong about the pipe, and neither is right about the number, because both are lossy renderings of a fraction. The dataset therefore stores the fraction and its exact value alongside the decimal, which is strictly more correct than either published figure.

The Standard Publishes Its Own Rounding Rule

ASME B36.10M-2015 states how its own figures are rounded: outside diameters above 16 inches to the nearest millimetre, 16 inches and below to the nearest 0.1 mm, and wall thicknesses to the nearest 0.01 mm. That rule is why DN 250 is published as 273.0 mm rather than 273.1. Ten and three-quarter inches multiplied by 25.4 gives 273.05, and anything that converts instead of reading rounds it up. The same mechanism produces 323.8 mm at DN 300 and 457.0 mm at DN 450.

It makes a useful tell. A table printing 273.1 and 323.9 under an ASME heading was generated by unit conversion rather than transcribed from the standard, and the distinction is not cosmetic: the standard’s own values are the ones a mill works to. Three of those converted figures were in an early draft of this project’s research notes, taken from secondary sources, and were replaced once the standard itself had been parsed.

Four Levels of Confidence

Every row carries a status, and that status governs how it is shown.

What We Could Not Verify

Publishing this list is the point of the page. Every reference site has gaps; most do not say where they are.

GB/T 17395-2024 Tables 1–4

Pages 8–47 are Private-Use-Area glyphs with no character map; the standards body serves encrypted image tiles. Closing this requires buying the standard.

EN 1057 Table 3 wall matrix

The only freely available copy rate-limited, and its cell grid collapsed on extraction. The outside diameters and tolerances survived; the matrix did not.

KS D3507 / D3562 standard text

Paywalled. The Korean figures here rest on distributor and engineering-reference tables, not the standard itself.

IS 8329 (Indian ductile iron)

Access denied twice. Sizes DN 600–1200 rest on a single publisher.

AS 1074 document

Behind the Standards Australia paywall. The figures come from four publishers that agree cell-for-cell, plus the standard's own published abstract.

US HDPE DIPS walls, DR 17–32.5

The only source found fails its own wall-from-diameter relationship. Rejected outright rather than published.

AS/NZS 1477 Series 2 walls

Single source.

Eight cells in GOST 8734 and 10704

Single-digit typos with no upstream correction. GOST 8732 was reissued in 2025 with all eleven of its deviations fixed; 8734 has had no such reissue.

530 unnumbered ASME wall thicknesses

Listed in the standard without a schedule number, and published by only one source. Corroborated by that source's separately-transcribed mass column, which is a consistency check but not a second publisher.

Why Source Count Isn’t Source Independence

A rule of “three independent sources” is easy to satisfy and easy to fool. Five clusters of apparently separate sources turned out to be the same content republished:

• Nine domains serving a byte-identical template of the same 75 KB page.
• Three sites reproducing the same optical-recognition master, identical down to a defect in one column header.
• Three widely-cited references all faithfully reproducing the same erratum printed in the standard they copied from. Three sources agreeing, all wrong.

In each case only the arithmetic dissented. That is why the formula check, not the source count, is the authority here.

What the Arithmetic Cannot See

The formula tests a diameter against a wall against a mass. It is blind to the schedule number printed beside them, and a label is not a number. That is how 39 wrong schedule designations passed every check on this site untouched.

The cause was an axis collision. Four nominal sizes — 5, 10, 20 and 30 — are also valid schedule numbers, and the original parse read the nominal size column as the schedule column for exactly those four. Every wall at NPS 5 was tagged schedule 5 and every wall at NPS 10 was tagged schedule 10, which cannot be true: a nominal size has at most one wall per schedule. The dimensions were all correct. Only the names attached to them were wrong, and no amount of recomputing masses would ever have said so.

The visible consequence was that the schedule 40 pipe table omitted 5-inch and 10-inch pipe altogether — two of the most commonly specified sizes — while its own text stated that schedule 40 and STD are identical through NPS 10. Corrected on 2026-08-18. Designations are now re-derived from the source table by a separate script that reports all 766 rows matching the standard, and it is run after any change to the dataset. A check that only re-runs the arithmetic re-runs the thing that already passed.

This correction, and every other one found in published charts and in our own data, is listed with its date on errors found in published pipe data, and in ours.

Corrections

If you find an error, it is worth reporting and it will be published with a date rather than quietly edited. Data revision: 2026-08-18.

Back to the pipe size chart.