What this course does not know

The position, stated plainly

For most of its life this course could not consult the UK National Annexes, and said so on every page. The documents were then obtained in two stages: most in July 2026, and BS EN 1993-1-8 and BS EN 1090-2 with their National Annexes shortly after. All 47 design values have now been read from their published source, and each one carries the clause it came from.

The position taken while the documents were out of reach is the point of this lesson, and it does not stop being the point now that they have been read. There were three possible responses, and only one of them was honest.

Use plausible values quietly. Many of the numbers in general UK circulation are almost certainly right. Using them without saying so would have produced a course that looked complete and could not be audited — and a reader would have had no way of telling which numbers were checked.

Omit the topics entirely. This would teach less than is teachable. The mechanics is unaffected by a national choice: the buckling behaviour of a column does not depend on which partial factor a National Annex adopts.

Teach the behaviour and refuse the number. Which is what the course did. Where a design value could not be verified, it was recorded as requiring verification, and any calculation depending on it was marked teachable but not usable.

That discipline paid, and this is the argument for it. Reading the documents found four errors, one of them unconservative. It found them because every value had a clause reference against it to check. A course that had quietly used plausible numbers would have had the same four errors and no way to discover them.

One thing has not changed. Verified means each number can be traced to a clause. It does not mean this course is fit for producing design values for construction: none of the work here has had independent design review, and Module 19 is about exactly why that distinction matters.

What the verification found

Reading the documents was not a formality. It found four things.

η was wrong, and unconservatively. The course held the recommended European value of 1.20 for the shear-area factor. The UK National Annex to EN 1993-1-5, NA.2.4, sets it to 1.0 for all steels. The course had been overstating the shear resistance of a stocky web by 20%.

Two different factors are both called γ_M2, and they had been conflated. EN 1993-1-8 gives 1.25 for bolts and welds. EN 1993-1-1 §6.1 gives one for a cross-section in tension to fracture, and the UK sets that to 1.10. The course was using the bolt factor in a member check — which is why Module 6 now spends a paragraph on the distinction, and why its allowance moved from 3% of area to 15%.

The plastic sway threshold was 15; the UK gives 10, and 5 for portals under gravity loads only. Module 4 was rewritten around it.

Steel density is a range, 77.0 to 78.5 kN/m³, and the course held the lower bound without saying so.

None of these was found by re-reading the calculations. They were found by opening the source — which is the last lesson's point about independence, applied to this course.

The register, and why it is generated

The register that follows is produced by reading the design-standard profile at the moment the page is built. Nothing in it is written by hand.

That is deliberate, and the reason is worth stating: a hand-written register drifts. The first time a value is added to the profile without the document being updated, the register becomes wrong — and a register that has drifted is worse than no register at all, because it is believed.

Generating it means the two cannot disagree. If a value's status changes in the code, the register changes with it; if the course ever verifies the National Annexes, the register empties itself without anybody editing this page.

That is now the best evidence this page can offer, because it is exactly what happened. An earlier version of this lesson made that claim as a prediction, with eight entries still outstanding — seven from BS EN 1993-1-8 and one from BS EN 1090-2, the two documents not then in the set. Those documents were obtained and read. Nobody edited the register. It is empty because the profile changed underneath it.

As it stands: 0 of 47 design values require verification. Read the denominator, not the zero. A register that reports only its outstanding entries becomes meaningless at the moment it succeeds — "nothing outstanding" tells you nothing about whether the profile holds four values or four hundred, or whether anyone ever populated it at all.

The register is the weaker claim of the two on this page, and it is worth being clear about which is which. It says every number is traceable to a clause. It does not say the numbers are right for your project, that the clause was the applicable one, or that anyone independent has looked. Those are different questions, and Module 19 is about them.

A cleared register is not a finished job. It is the point at which the interesting question changes from do we know where these numbers came from? to did we choose the right ones?

The end of the course

Twenty modules, and the argument they add up to is smaller than the material.

Steel is a material whose failures are mostly about stability, and stability is a question about geometry and restraint rather than about strength. That is why so few of the remedies in this course involve a stronger steel: a deflection problem wants depth, a buckling problem wants restraint, a fatigue problem wants a better detail, and none of them cares what the yield strength is.

The rules are a mixture — some derived, some calibrated, some chosen by a national committee — and telling them apart is not pedantry. It tells you which ones you can reason from and which you have to look up.

And the decisions that matter most are taken earliest, with the least information, and then stand. Which is an argument for spending real effort at the beginning, and for writing down why — including what you did not know at the time.

The last of those is the one this course has tried hardest to practise, and the register is the evidence.