Following G 1/23, products made available to the public are prior art in Europe, even if a skilled person cannot analyse or reproduce the product. The interpretation of G1/23 is clear with respect to novelty. The question becomes whether the product was or was not disclosed before the priority date. The implications of G 1/23 are tricker to predict for inventive step. The decision in T 1719/21 considered whether G 1/23 does or does not permit a non-reproducible product to be the closet prior art in the problem solution approach. Rejecting an interpretation of G 1/23 that it felt would have over-interpreted the EBAs "Coco-Cola example", the Board of Appeal in T 1719/21 unequivocable rejected the blanket exclusion of non-reproducible products as closet prior art. This decision is therefore aligned with the other recent interpretation of G 1/23 in T 1044/23 (IPKat). 

Legal background: G 1/23, solar panels and Coca-Cola

G 1/23 may be considered to have fundamentally changed the EPO approach to non-reproducible prior art. Previously, under G 1/92, products that could not be analysed and reproduced were genereally excluded from the state of the art entirely. G 1/23 rejected this position and confirmed that non-reproducible products disclosed before the priority date can be cited as prior art for both novelty and inventive step.

The case underlying the referral in G1/23 related to the use of non-reproducible prior art in an inventive step attack (T 0438/19, EP2626911). Interestingly, in its detailed reasoning, the EBA included the "Coca-Cola" example of a product that can be purchased but whose formula is secret. The EBA considered whether a skilled person would realistically start from such a product in a problem-solution assessment of inventive step if the invention required a modification of the secret internal structure. For the EBA, if a claimed invention merely involves an external modification, such as adding lemon juice to the beverage to adjust the taste, the secrecy of the original recipe is irrelevant. The skilled person can physically obtain the product and perform this modification, so the secret formula does not inevitably confer an inventive step on the new mixture. However, the analysis changes if the invention requires modifying the product's internal structure, such as creating a sugar-free version that perfectly replicates the original taste. In this scenario, because the starting formula is unknown and non-reproducible, achieving the specific technical effect remains an "unsolved problem," making the secret product a poor starting point for an obviousness attack.

The return of ENGAGE® 8400

The facts of the case in the recent case of T 1719/21 are familiar, relating as it does to same patentee, opponent, subject matter and the same specific legal issue as the referring case in G1/23. The patent (EP 2833415) particularly related to a solar cell sealing material comprising an ethylene/alpha-olefin copolymer. 

Importantly, as with the referring case to G1/23, the main issue on appeal was not novelty in view of a cited prior use, but inventive step. The Opponent (Borealis GmbH) argued that the closest prior art was the commercial polymer ENGAGE® 8400 (the same product as in the G1/23 case). It was undisputed by the parties that ENGAGE® 8400 was a commercial product available before the priority date under non-confidential conditions. However, it was also common ground that the exact synthesis method (specifically the catalyst system) required to reproduce ENGAGE® 8400 was not in the public domain.

Just prior art or closest prior art?

In T 1719/21, the Patentee argued that, even if G 1/23 meant the product was technically "prior art", a non-analysable product could not represent "the closest prior art" for the purposes of the problem-solution approach. The Patentee submitted that it would be "unrealistic and necessarily contaminated by hindsight to assume that a skilled person would start from a non-reproducible polymer". Indeed, argued the Patentee, why would a skilled person try to improve a product if they didn't know how to make it in the first place, especially given that this would require "spending considerable resources on establishing the synthesis conditions" before any modification could even be attempted.

The Patentee's argument relied heavily on the Coco-Cola example from G 1/23. They argued that the EBA had implied that where a modification of a non-reproducible product is necessary, that product cannot form the closest prior art. In other words, whilst the skilled person might add lemon to Coca-Cola (modification of the whole), they wouldn't try to change the secret syrup formula itself.

The Board of Appeal was not convinced by this argument. Aligning with T 1044/23, the Board of Appeal fundamentally rejected the Patentee’s interpretation of G 1/23, finding instead that the EBA's discussion of Coca-Cola served merely as an illustration of potential arguments and not as a rigid rule that excluded non-reproducible commercial products from being the closest prior art starting point in a problem-solution assessment of inventive step. Importantly, the Board of Appeal further distinguished between the election of the starting point and the modification required to solve the problem:

"What needs to be modified, however, is part of the inventive thinking of the skilled person in order to solve the problem addressed, but not a consideration concerning the selection of that starting point." (r. 14.3.4)

The Board of Appeal reasoned that the selection of the closest prior art should thus be based on what is "realistic or promising". For the Board of Appeal, if a commercial product has uses, effects, and properties relevant to the goals of the patent (in this case, solar cell encapsulation), it may be a realistic starting point and the skilled person doesn't need to know the synthesis method a priori to select the product as a candidate for improvement. The Board of Appeal noted, for example, that the skilled person might solve the problem, not by synthesising the polymer from scratch, but by replacing it with another, or modifying other aspects of the composition. For the Board of Appeal, to assume that the skilled person must reproduce the synthesis "would imply that the closest prior art... is selected after having completed the assessment of obviousness of the solution" (r. 14.3.4).

All still to play for in inventive step analysis 

Having rejected the argument that G1/23 should be interpreted as excluding non-reproducible products from being the closest prior art, the Board of Appeal concluded that, in the case in question, the commercial product ENGAGE® 8400 was a valid starting point for the assessment of inventive step. The distinguishing feature between the claim and the ENGAGE® 8400 prior art was found to be the specific content of fluorine in the copolymer. The Board of Appeal accepted that whilst ENGAGE® 8400 likely contained some fluorine residues from the catalyst system, the Opponent had not proven the content was within the claimed range. 

The problem to be solved was thus formulated as the provision of an encapsulating material with higher volume resistivity. The Patentee had provided evidence that keeping the fluorine content low improved this property. For the Board of Appeal, even if the skilled person started from ENGAGE® 8400, there was no suggestion in the prior art that reducing fluorine content to such low levels would improve volume resistivity. The Board of Appeal concluded that the claimed solution was not obvious. The Board of Appeal therefore dismissed the inventive step attack and the patent was maintained on the basis of an auxiliary request limited to the lower fluorine level. 

Final thoughts

In G 1/23, the EBA made the important observation that, whilst non-reproducible products should now be considered part of the state of the art, this does not mean they thereby automatically render subsequent inventions obvious. On the contrary, the EBA emphasised that there are no "strict rules" for how such products are treated in the problem-solution approach. As the Board of Appeal in this case clarified, the appropriate approach to inventive step is a case-specific inquiry. For the EBA in G1/23, the "reproducibility" hurdle was thus primarily about whether the product is prior art. Once that hurdle is cleared, the test for whether it serves as the closest prior art reverts to the standard criteria of inventive step analysis. 

The Board of Appeal's reasoning is this case and the prior decision of T 1044/23 thus avoid a potential loophole to G 1/23 whereby a patentee might have been able to acknowledge a product is prior art but effectively side-line it from the inventive step analysis by claiming it is too mysterious to be a starting point for the problem solution approach. Of course, as demonstrated in this case, the "black box" nature of the product can still save patent invention from an inventive step attack. Therefore, whilst G 1/23 removed the legal fiction of non-existence of non-reproducible products, it didn't remove the practical difficulty of attacking a patent based on a product whose properties cannot be fully elucidated. 

Source : https://ipkitten.blogspot.com/2025/12/non-reproducible-products-can-be.html