A dispute over evidence that arrived later
G 2/21 arose from a case concerning insecticidal compositions: can inventive step rely on effects demonstrated by research submitted after the filing date? The Enlarged Board of Appeal answered on 23 March 2023. The dispute exposes a distinction between the date of an evidential document and the technical content of the invention originally disclosed. [1]
Editorial inference: for a research team, the issue also concerns organization. Experimental timetables rarely align perfectly with a filing plan. Teams nevertheless need to show what was described at a particular date, what was merely confirmed later and which assumptions connect the two stages. Without this chronology, a favourable individual report can be difficult to assess properly.
The two principles in G 2/21
The Board held that evidence of a technical effect cannot be disregarded solely because it was not public before filing and was submitted later. However, reliance on an effect depends on what the skilled person, using common general knowledge and the original application, would derive from its technical teaching: the effect must be encompassed by that teaching and embodied by the same originally disclosed invention. [1]
This is not a general permission to add any justification later. G 2/21 also distinguished inventive step from sufficiency of disclosure; later evidence is not a universal remedy for deficiencies in the original documentation. [1]
What the subsequent T 116/18 decision illustrated
In T 116/18 of 28 July 2023, the Technical Board applied G 2/21 to a composition’s synergistic activity against Chilo suppressalis. It held that the original application need not always contain experimental proof of the particular effect or an express verbal statement of it. The Board examined the technical teaching as a whole and the skilled person’s knowledge, then dismissed the appeal. This describes that 2023 decision; it does not mean every later experiment will be accepted. [2]
Editorial inference: distinguish three assessments. A document may be present in the proceedings, may support a fact requiring proof, and may concern an effect that can be relied on for inventive step. A positive answer to one question should not automatically replace answers to the others.
Connecting experiments with the problem–solution approach
The EPO Guidelines organize inventive-step assessment into three stages: identifying the closest prior art, formulating the objective technical problem, and asking whether the solution would have been obvious to the skilled person. Article 56 EPC establishes the non-obviousness requirement itself. Evidence of an effect therefore forms part of a larger argument rather than replacing a comparison with the prior art. [3] [4]
Methodological proposal: attach a chain of reasoning to a report: distinguishing feature, experimental conditions, result, inferred effect and technical problem. Identify the date, document and limitations at every link. If an experiment compares unsuitable variants or changes several parameters simultaneously, explain why it can attribute the result to a particular feature.
Applying the study to research-project organization
Editorial inference: a useful patent-analysis package includes the application version, experimental protocols, definitions of reference materials, unfavourable and favourable results, and justification for extrapolation. This allows an assessment of whether later data confirm the same concept or require a new assumption. The aim is reproducible reasoning rather than the largest possible collection of charts.
When reading decisions, also record which question each answers and its date. The term “plausibility” alone is not a complete test. G 2/21 provides an exercise in constructing an argument from documents, dates and technical features while separately examining the admissibility of evidence and sufficiency of disclosure.
Additional documentation proposal: a protocol should explain the choice of comparator and the scope of uncertainty. Retain drafts and unsuccessful experiments alongside positive results. This helps identify whether a favourable observation is repeatable and whether the conclusion extends beyond the conditions actually investigated.
Lessons from the analysis
- Separate the date of evidence from what the original application disclosed.
- Connect an effect to a specific feature, comparison and technical problem.
- Do not treat later data as a universal way to repair an application.
Limits of interpretation
G 2/21 concerns the EPC assessment framework. T 116/18 is discussed by reference to the decision of 28 July 2023; its interpretation is not presented as the only possible approach in every later case. Documentation proposals are editorial inferences.
Sources and further reading
- 1EPO — Enlarged Board of Appeal decision G 2/21 ↗
G 2/21; ECLI:EP:BA:2023:G000221.20230323; OJ EPO 2023, A85
Primary source, 23 March 2023; order and reasons, particularly points 77 and 93–95.
- 2EPO — Technical Board decision T 116/18 of 28 July 2023 ↗
T 116/18; ECLI:EP:BA:2023:T011618.20230728
Primary source; particularly reasons 11.10–11.14 and 17; an application of the principles to a specific composition.
- 3EPO Guidelines — Problem-solution approach ↗
Guidelines for Examination 2026, G-VII, 5
Official description of the three stages of the problem–solution approach; checked 16 October 2026.
- 4European Patent Convention — inventive step ↗
EPC, art. 56
Official text of the provision establishing the basis for the non-obviousness assessment.