EPO G 1/26: How the Coated Steel Strips Case Tests Article 123(2)

At Solve Intelligence, we build AI tools purpose-built for patent prosecution and opposition workflows. European teams, including cross-Atlantic firms like HG, use Solve Intelligence to stay ahead of EPO doctrine shifts as they happen. In this article, we cover the three referred questions in G 1/26, the practical implications for prosecution and opposition teams, and what to do before the decision lands.

Key takeaways

• G 1/26 asks whether G 1/24's rule to consult the description extends from patentability assessment to Article 123(2) added matter.
• The underlying case (T 873/24) turns on whether "ratio of titanium to nitrogen in excess of 3.42" means a weight ratio or a molar ratio.
• The referral poses three sets of questions: admissibility under Article 112(1), the reach of the description in claim interpretation, and the consequences for Article 123(2).
• If the Enlarged Board extends G 1/24 to Article 123(2), description-based narrowing applies to added-matter assessment across all European proceedings.

What is G 1/26?

Less than a year after G 1/24 reshaped how examiners and opponents read European patent claims, a Technical Board of Appeal has referred a fresh batch of questions to the Enlarged Board under the name G 1/26 "Coated steel strips". 

The new referral asks whether the description must also be consulted when assessing added subject matter under Article 123(2) EPC, and it puts the gatekeeping test for referrals themselves back on the table.

The dispute arises from a near-mundane technical feature: the ratio of titanium to nitrogen in an aluminium-coated steel strip. Is "3.42" a weight ratio or a molar ratio? The answer determines whether the granted claim adds matter beyond the application as filed. The Enlarged Board’s response will reach far beyond metallurgy. 

The patent and the missing units

T 873/24 sits at the tail end of a long ArcelorMittal patent family on aluminium-coated steel strips for hot stamping, the kind of steel that goes into automotive structural parts. The patent under appeal is EP3587104 B1, a divisional that traces back to a 2006 priority filing. The Opposition Division had maintained the patent in amended form, rejecting most added-matter attacks. The opponent appealed. The appeal sits with Technical Board of Appeal 3.3.05.

The contested feature in claim 1 of the Main Request reads:

"wherein the ratio of titanium to nitrogen is in excess of 3.42".

The parent and grandparent applications expressed the Ti/N ratio explicitly "by weight", and paragraphs [0058] and [0062] of the description in the divisional do the same. The granted claim of the Main Request, however, drops the unit. Read in isolation, "ratio" usually defaults to a weight ratio in metallurgy. Read across the description as a whole, the term could plausibly encompass either a weight ratio or a molar ratio.

That ambiguity has real teeth. Under Article 123(2) EPC, an amended claim must be directly and unambiguously derivable from the application as filed. If "ratio" in the granted claim must be construed broadly enough to cover molar ratios, the granted claim reaches beyond what was originally disclosed and the Main Request falls. If, by contrast, the description’s "by weight" can be read into the claim, the term narrows, and the added-matter objection collapses.

How the ambiguity became a referral

The Board’s preliminary opinion in October 2025 gave no hint of a referral and pointed towards revocation. It was a late submission by the proprietor invoking G 1/24 that put a different question on the table at the oral proceedings on 3 February 2026: 

“Does the G 1/24 mandate to consult the description when interpreting claims apply when the question is added matter, not patentability?” 

The Board concluded that the answer was not clear, that the case turned on it, and that the issue was of fundamental importance. It referred.

So the case poses a deceptively narrow question and produces a sweeping one. The narrow question is whether one ArcelorMittal divisional survives opposition. The sweeping one is whether G 1/24 governs claim interpretation across the EPC, or only within the patentability provisions where it was decided.

What G 1/24 changed

G 1/24 settled, at least for patentability, that the description should always be consulted when interpreting the claims. The Enlarged Board rejected the older "primacy of the claims" approach that treated the description as a tie-breaker available only when claim language was unclear on its face.

But G 1/24 was decided in the context of novelty and inventive step. It did not say in terms what should happen when the same interpretive question arises under Article 123(2) EPC, under Article 84 EPC, or in sufficiency. Boards have since divided on whether G 1/24’s holding is a general rule of construction or a doctrine confined to patentability.

T 873/24 is the case that forces the issue.

What are the three questions referred in G 1/26?

The referring Board grouped its questions into three.

1. When is a referral "required"?

Question 1 puts the admissibility test for referrals themselves on the agenda:

"May a decision be considered to be ‘required’ for the purposes of Article 112(1) EPC, if the referring Board demonstrates that the point of law in question arises out of the context of the case pending before it and, in the circumstances of the proceedings, it is reasonable for the Board to examine it and decide on it next?"

The traditional reading of Article 112(1)(a) is that a referral is only "required" if the answer would actually change the outcome of the pending case. Question 1 asks whether that bar can be lowered. If the Enlarged Board says yes, future Boards will have more room to flag systemic issues even where the immediate appeal could be decided without doing so.

2. The reach of the description in claim interpretation

Questions 2(a) and 2(b) revisit the heart of G 1/24:

"Does the fact that the claims are the starting point and the basis for assessing the patentability of an invention generally preclude a feature which is only disclosed in the description or the drawings of a patent from being read into the meaning of a granted claim, in particular if this leads to a restrictive reading of terms used in the claim?"

"If the answer to question 2(a) is no: is claim interpretation the result of both reading the claims and consulting the description and drawings as a unitary process, and does the claim being the starting point and the basis for assessing the patentability rule out only those interpretations which can be derived from the patent as a whole but would clearly contradict the general technical understanding of the terms used in the claim?"

This is G 1/24 stress-tested at its edge. The question is whether the description can do more than disambiguate. It is whether the description can actually narrow a claim term to something less than its plain meaning.

3. Carrying the rule into Article 123(2)

Questions 3(a) and 3(b) ask whether the same interpretive method applies when checking compliance with Article 123(2) EPC. If a description-narrowed claim is the correct starting point, opposition divisions and applicants will have to assess added subject matter by reference to the narrowed reading, not the literal claim text alone. That would mark a meaningful shift in how Article 123(2) objections are formulated and answered.

Why this matters in practice

The downstream effects of G 1/26 are not confined to coated steel strips.

Drafters routinely choose between a clean, short claim that leans on the description for context and a longer claim that bakes every restrictive detail into the wording. If the description can restrict claim meaning for the purposes of Article 123(2), the cost of a sparse claim falls. If it cannot, drafters must remain disciplined about putting every load-bearing limitation in the claim itself.

Opponents will read this in the opposite direction. A restrictive description-based interpretation is a familiar fallback for patentees fighting novelty attacks. It is a poor friend when the same patentee needs the claim construed broadly for infringement. G 1/26 will affect how this "Angora cat" problem is managed before grant and at opposition, and how it lands in parallel national or UPC proceedings.

Examination practice is likely to shift too. Examiners citing Article 123(2) objections will need to consider whether the description supports a narrower construction that defeats the objection. Applicants will respond accordingly.

The post-G 1/24 referral cluster

G 1/26 does not arrive in isolation. It is the second referral following G 1/24, after G 1/25 "Hydroponics" asked whether and to what extent the description must be amended to conform with allowed claims. The EPO’s 2026 Guidelines for Examination have already absorbed G 1/24 and the parallel G 1/23 decision on prior art; G 1/25 and G 1/26 will require another round of updates. 

Taken together, this cluster is reshaping the relationship between claim and description across the full lifecycle of a European patent. Drafting, examination, post-grant amendment, opposition, and ultimately litigation each touch a different point of that relationship, and each will need to be reconsidered once G 1/25 and G 1/26 are decided.

What practitioners should do before G 1/26 is decided

The Enlarged Board has not yet issued a preliminary opinion or scheduled oral proceedings in G 1/26. Amicus curiae briefs are likely to be invited in due course, and the case will attract submissions from across European industry and the profession. 

Patentees and opponents should map their pending Article 123(2) arguments under both possible outcomes now. The cost of building a fallback is small compared with the cost of replaying an opposition strategy mid-procedure.

How Solve Intelligence helps prosecution teams navigate G 1/26

Solve Intelligence is an AI co-pilot for patent attorneys, built around the European, US, and PCT workflows practitioners use every day, from prosecution and EPO oppositions through to UPC and national-court litigation. Our platform surfaces case-law shifts like G 1/24, G 1/25, and G 1/26 as they happen, drafts and reviews Article 123(2) arguments grounded in current EPO doctrine, and supports the same analysis whether you are responding to an office action, building an opposition, or preparing a revocation attack or defence.

Book a demo to see how prosecution teams stay ahead of EPO referrals without the manual reading load.

FAQs

How does G 1/26 differ from G 1/24?

G 1/24 held that the description should always be consulted when interpreting claims for the purposes of patentability, meaning novelty and inventive step. G 1/26 asks whether the same rule extends to added subject matter under Article 123(2) EPC, and whether the description can do more than disambiguate, that is, whether it can actually narrow the meaning of a claim term.

When will the Enlarged Board decide G 1/26?

No date has been set. Recent referrals of comparable scope have taken between twelve and twenty-four months from referral to final decision, and amicus curiae briefs are expected to be invited.

What should drafters do in the meantime?

Treat the description as load-bearing. Make sure every term that could be read in more than one way, including units, ratios, ranges, and percentages, is either tightly defined in the claim or unambiguously supported in the description. The cost of belt-and-braces drafting is far lower than the cost of an Article 123(2) attack at opposition.