Last verified 19 August 2026.
The three kinds of number on a datasheet
Most property tables contain at least two of these, frequently unlabelled.
A specified minimum. Set by a standard or a purchase specification. The supplier is contractually committing that material sold to that specification will meet it. It is not a prediction of what your plate will do — it is a floor. Real material usually sits above it, sometimes well above.
A typical value. What the supplier’s production usually produces. Useful for thermal design, for weight, for a first pass. It carries no commitment and no stated distribution — “typical” is not a defined statistical term, and two suppliers can mean different things by it.
A statistically derived allowable. A value computed from a characterised population to a defined confidence — the A-basis and B-basis allowables in MMPDS being the best-known example in aerospace. This is the most rigorous of the three and the least commonly available, because establishing one requires a lot of tested material.
Under the E-ladder, the first and third are E5. The second is E3 — real, useful, and not a design allowable. Which of the three you can obtain at all depends on the source you are reading; we list ours and their limits.
Why the gap matters more than it looks
The gap between a specified minimum and a typical value is not a safety bonus you get for free. It is variability that has already been spent.
Design to the minimum and your margin is a margin against the worst material the specification permits. Design to a typical value and your margin is a margin against average material, which means that on the day you receive a heat at the low end of the permitted range — entirely legitimate, entirely conforming — part of your margin has quietly gone.
The failure is rarely dramatic. It shows up as a component that passes qualification on the first batch and fails on the fourth, and by then the datasheet tab is closed and nobody remembers which column the number came from.
The case that shows it best: one alloy, two standards
Duplex stainless 2205 (UNS S32205 / EN 1.4462), minimum proof strength:
| Value | Standard | What it is | Evidence |
|---|---|---|---|
| 450 MPa (65 ksi) | ASTM A240/A240M | specified minimum | E5 |
| 500 MPa | EN 10088-2 | specified minimum | E5 |
Same alloy. Both correct. Eleven per cent apart.
Neither is a typical value — these are both floors, set by two standards bodies that made slightly different choices. An engineer who carries “2205 yields at 450” between projects and lands on a drawing governed by EN is not wrong about the material. They are working under a document they have not read, which is harder to notice.
And a mill datasheet that prints one of these without naming the standard has handed you a number that cannot be checked. That is the difference between E5 and E3, and it is created entirely by what the record says, not by what the material does.
Source: Outokumpu Forta range datasheet, May 2015, which gives EN 10088-2 minima as primary and ASTM A240 imperial equivalents alongside.
The other end of the range
316L (UNS S31603), minimum specified proof strength: 170 MPa (25 ksi) to ASTM A240.
Put that next to 2205’s 450 MPa and the comparison that matters becomes visible: for a strength-driven part the two are not close, and a substitution made on corrosion grounds alone changes the structural problem by a factor of more than two.
Source note, and it is deliberate: the ASTM standard is paywalled and was not consulted directly for this page. Two independent secondary sources agree on the figure. Under the TRACE demotion rule, a value whose claimed level cannot be verified from the record alone drops one level, so we record it as E3, not E5, and we say so. Confirm against the standard before it sizes anything.
How to tell which one you have
Six questions, in the order they are quickest to answer:
- Does the table name a standard, with an edition? No standard, no specified minimum. It is a typical value regardless of what the column heading says.
- Does it say “min” or “typical”? When both appear in one table, treat unlabelled rows as typical.
- Is the product form stated? Specified minima vary by product form — plate, bar, forging, tube. A minimum with no product form is incomplete.
- Is the condition stated? Annealed, cold worked, solution treated and aged. Without it the number belongs to no material state in particular.
- Is there a temperature? Room-temperature values are frequently printed without one, which is fine until somebody uses them at 400 °C.
- Could you reach the clause? If you cannot get to the table in the standard, you are holding a secondary report of a specified minimum — useful, not the same thing.
If the answer to 1 is no, you have an E3 value. Fine for screening, fine for a first pass, not what you size a part from without recording that you chose to.
What to do about it
For screening: typical values are appropriate and there is no reason to be precious about it. Most of the funnel is E3 and always has been.
For sizing: use the specified minimum from the standard your drawing invokes — and check which standard that is, not which one you are used to.
For qualification: the mill test certificate for your actual heat supersedes both. That is the point at which an E3 datasheet value can be retired and replaced with an E5 certificate value, and it is worth recording the swap rather than letting it happen silently. See reading a mill certificate.
For the record: whichever you used, write down which one it was, which document it came from and when you read it. That sentence costs thirty seconds now and is the entire answer to a question you will otherwise have to reconstruct.
What this page does not cover
- How to choose a design factor. That is governed by your design code, not by this distinction.
- Statistical derivation of allowables. A-basis and B-basis values have a defined methodology; this page only places them on the ladder.
- Elevated-temperature values. The specified-versus-typical distinction holds, but the standards involved and the data availability are different problems.
- Non-metals. Polymer and composite datasheets carry a related but differently shaped problem, driven by processing and conditioning rather than by heat chemistry.
- Whether the standard itself is right. Standards are revised, and occasionally for a reason.