The quiet clause that can sink a patent: The disclosure

There’s an article in the European Patent Convention (EPC) that looks harmless, almost an administrative requirement. Article 83 EPC simply says a European patent application must disclose the invention “sufficiently clear and complete” for a person skilled in the art to carry it out.

In real life, that one line can behave like a trapdoor.

Because “sufficiently clear” isn’t a vibe. It’s a legal threshold. And if not fully observed, this threshold can determine the revocation of a granted patent.

The skilled person: A legal fiction with real consequences

The “skilled person” is not a named engineer in a lab. It’s a legal construct: competent, practical, armed with common general knowledge, but not expected to invent the missing pieces.

That last part matters. The European Patent Office (EPO) will usually allow the skilled person to fill routine gaps based on their common general knowledge. But if the patent forces them to run a research programme, to guess what works, to discover key parameters, to try endless combinations, then the disclosure may be deemed insufficient.

And once you’ve filed, you can’t fix it by “adding” technical teaching without risking added-matter problems and hurting a fellow article in the same EPC: Article 123.

So what does this look like in practice?

Four appeal cases, all involving granted European patents that were opposed on the grounds of Article 83, show how sufficiency of disclosure can unravel late, and that means too late to repair.

Four recent cautionary different stories that had a common result: patent revoked

1. Big promise, thin roadmap: T 1026/23 (Nestlé)

This case concerned compositions administered to animals to enhance neurogenesis, built around combinations that included unsaturated fatty acids, nitric oxide-releasing compounds, B vitamins, and antioxidants. The difficulty wasn’t the ambition; it was the reach.

The board concluded the patent didn’t provide enough practical teaching to achieve the promised effect across the full breadth of the claims without undue burden. In other words: the skilled person would have to find the working embodiments through substantial experimentation.

That’s not “carrying out” an invention; that’s discovering it.

The take-away lesson: If you claim broad biological effects, the description must do more than inspire confidence: it must enable results across the scope you’ve claimed.

2. A key concept left in the fog: T 1489/23 (General Electric)

Here, the invention aimed at closed-loop control in additive manufacturing, using high-speed imaging of a “spark plume” during welding and applying statistical process control to detect defects in real time.

The weakness wasn’t the use of cameras or control charts. It was the patent’s failure to anchor a central feature - the “spark plume” - with enough technical clarity and operational guidance. If the skilled person must spend serious effort working out what the signal is, how it behaves, and how to set meaningful control limits, the disclosure can tip into insufficiency.

The take-away lesson: If your invention depends on a phenomenon, signal, or measurement, define it in a way that is operational, something the skilled person can reliably identify, measure, and use.

3. A parameter you can’t reliably hit: T 1495/23 (Daicel)

This patent involved a tow band for e-cigarette tips, defined using denier ranges and a breaking strength ratio parameter (F/TD) aimed at preventing fragmentation and melt holes.

The board agreed that the patent did not teach the skilled person how to achieve and adjust that critical parameter as required by the claims. It’s a classic sufficiency problem: the claim is written around a target number, but the description doesn’t show the reader how to consistently arrive there.

The take-away lesson: If a claim lives or dies by a parameter, the description must teach the how, not just the what.

4. One example trying to carry a whole claim: T 1842/23 (Wavefront Technology)

This case concerned optical products for reproducing 3D images using non-holographic elements “correlated” with a 3D object’s surface, including gradients and orientations, for security applications.

The board found the disclosure effectively described only one route (mimicking surface slopes) while the claims ran far wider. Add a term like “correlate” that is broad and ambiguous, and the skilled person is left without a clear, repeatable method across the claim scope.

The take-away lesson: Broad functional language is only safe when the description supplies enough concrete routes to make that breadth real.

The pattern: When “enablement” quietly turns into “experimentation”
These cases rhyme. Different technologies, same fault line:

• Claims are wide
• Teaching is narrow or vague
• The skilled person is left to bridge the gap with extensive trial-and-error

At that point, Article 83 stops being theoretical. It becomes outcome-determinative.

A practical drafting checklist to reduce Article 83 risk
If you want your patent to survive its worst day (opposition/appeal), bake sufficiency in early:

• Match claim breadth to your teaching. If you claim a broad class, show representative examples across that class.
• Define key terms operationally. “Spark plume”, “correlate”, “effective amount” - tell the reader what it means in practice.
• Explain how to achieve critical parameters. Include controllable variables, manufacturing steps, and measurement conditions.
• Distinguish essential vs. optional features. Don’t force the reader to guess what really matters.
• Include measurement protocols. If performance depends on test conditions, state them.
• Don’t outsource the invention to “common general knowledge.” It’s support, not scaffolding.

Closing thought
Sufficiency is often underestimated because it doesn’t feel dramatic. But it’s one of the few requirements that can invalidate a patent not because the invention wasn’t new - but because the patent didn’t teach it well enough.

And that, ultimately, is the heart of Article 83: patents are a bargain. Monopoly rights in exchange for usable knowledge. If the knowledge isn’t usable, the bargain collapses.

This is a co-published article, originally published on WIPR.