Context: The Enlarged Board of Appeal (EBA) of the European Patent Office (EPO) made a decision on claim construction last month that aligns its case law with that of the Unified Patent Court’s (UPC) Court of Appeal (CoA) (June 18, 2025 ip fray article). We will continue to report on EBA decisions that we believe are relevant to EPO and UPC practitioners.
What’s new: In a case over polymers used in solar cells referred to the EBA by a Technical Board of Appeal (TBA) two years ago (G 1/23), the decision came down today (PDF) and formally interprets, but for all practical intents and purposes modifies a decision that was also the first referral to the EBA in a given year, but more than three decades earlier: G 1/92. The analyzability and reproducibility of a product actually sold is irrelevant to its fundamental eligibility as prior art, and as a result, any pre-launch information published on that product cannot be deemed ineligible only for that reason. Such prior art references are novelty-destroying. In the context of the inventive step, analyzability and reproducibility may be relevant in connection with secondary indicia, but will often not be outcome-determinative either.
Direct impact: In the case at hand, it appears that patent holder Mitsui will lose (at least in its granted form) EP2626911 (“Solar cell sealing material, and solar cell module”), a patent challenged by German rival Borealis, which opposed the grant. The case will be remanded, but the EBA decision sounds like the result is a foregone conclusion.
Wider ramifications: This holding has implications not only for opposition proceedings and revocation actions in the UPC or national courts (provided that such courts agree with the EBA’s reasoning) but also for the timing of patent applications. As patent attorney Chris Goddard from the firm that won this appeal (Dehns) noted on LinkedIn, “this is pretty close to an ‘on-sale bar’” (a U.S. patent law concept codified in 35 U.S.C. § 102, which among other things defines when the release of a product by a patent applicant starts to create prior art). In G 1/23, however, the prior art reference of disputed eligibility was a third-party product.
Let’s start with the key holdings (headnotes):
I. A product put on the market before the date of filing of a European patent application cannot be excluded from the state of the art within the meaning of Article 54(2) EPC for the sole reason that its composition or internal structure could not be analysed and reproduced by the skilled person before that date.
II. Technical information about such a product which was made available to the public before the filing date forms part of the state of the art within the meaning of Article 54(2) EPC, irrespective of whether the skilled person could analyse and reproduce the product and its composition or internal structure before that date.
The second sentence was considered to simply be a corollary of the first.
There was also a third question, but it was a contingent one and wasn’t reached based on the answers to the first two questions.
How did the EBA get there? It didn’t perform statutory interpretation on a green field, in which case it could simply have held that enablement is a requirement for patentability but not an eligibility criterion for prior art. Here, however, the starting point was decision G 1/92, which contained a wording interpretable in different ways. The two interpretations that were mainly advocated in this case and also found in divergent decisions of recent decades (spoiler alert: neither convinced the EBA in the end) hinged on whether the pronoun “it” referred to a product or a composition. Grammatically, both interpretations were defensible. But in the case at hand, that would not have helped much, given a practical 1:1 relationship between product and composition.
The 55-page ruling touches on a number of interpretive questions, even on what the “common” in “common general knowledge” means (in other words than the ones you’l find in the decision, the EBA held that it means “commonly available” as opposed to “shared”).
The way the EBA interpreted and effectively vitiated at least a key element (if not the entirety) of G 1/92 rests on the following pillars:
- The referred questions involved both analyzability and reproducibility of the product proposed as a prior art reference.
- In the end, nothing in the world can be made completely of reproducible material, at least not until humans are capable of manufacturing any atoms. Therefore, there is nothing wrong with using a product on the market as a starting point for making something else.
- The distinction between manufacturing methods and properties is key. The notional skilled person may not know how to manufacture something, yet may be able to understand the properties.
- The actual or permanent availability of the product doesn’t matter as long as this kind of knowledge has been gained. Otherwise the eligibility determination would become a predictive (as to future supply) exercise. And yes, products can change, but then the old version and the new one will both be prior art references, possibly involving tricky questions as to what was prior art as per a given priority date.
The decision constitutes an interesting mix between abstract philosophical thoughts (such as that everything is, in the end, made of something irreproducible) and pragmatic considerations such as that the notional skilled person will, of course, look at what is available to it. If anything, availability issues resulting from irreproducibility might adversely affect the motivation to combine something, but for novelty considerations that wouldn’t be an issue at any rate and even in connection with the inventive it’s just part of the secondary indicia.
The party opposing the patent grant argued that it would be absurd not to consider Coca-Cola prior art. It’s been around for ages. Its recipe may be a trade secret, but that doesn’t mean that its properties couldn’t be part of the prior art. The EBA decision gets back to that example at the end of the part that addresses the first referral question:
“[D}isclosed but nonreproducible features of a product may, but need not, flow into the assessment of inventive step. Adding some lemon juice to a glass of Coca-Cola for a less sweet taste may not be inventive simply because the recipe for Coca-Cola is secret and therefore the original taste of Coca-Cola is considered as unattainable. On the other hand, achieving the original taste of CocaCola without sugar or caffeine is probably an unsolved problem at the time of writing.”
The same paragraph then continues with what can be paraphrased as saying that the decision merely deals with fundamental eligibility as prior art, but there can still be an argument over the substance of that prior art for patentability purposes:
“There are no formal and strict rules as to how a non-reproducible but publicly available product or any of its properties can be taken into account when inventive step is examined. The relevant technical teaching that a skilled person would take from such a product is always case specific – it depends on both the product in question and the invention under examination.”
Essentially, the EBA of 2025 finds that the EBA of 1992 said something that was redundant if interpreted reasonably and should, therefore, be ignored going forward.
Panel and counsel
Panel: C. Josefsson (Chairman), I. Beckedorf, T. Bokor, M. Teppey, Y. Rüedi, P. Gryczka and G. Pricolo.
Counsel for appellant-opponent Borealis: Dehns (Neil Campbell).
Counsel for appellee-patentee Mitsui: Hoffmann Eitle (individual(s) not known).