EPC Article 54: Understanding Novelty Requirements

Novelty is the first substantive hurdle every European patent application must clear. Article 54(1) EPC states it plainly: an invention shall be considered new if it does not form part of the state of the art. The rule sounds simple, but most novelty objections in practice trace back to a misunderstanding of two things — how broad the state of the art actually is, and how strict the comparison between a claim and a single prior disclosure must be.

This guide sets out what Article 54 covers, how the European Patent Office assesses novelty, and how to evaluate it yourself before committing to drafting and filing costs. It is written for patent professionals, technology-transfer officers, and in-house IP teams who need to make a filing decision with confidence rather than hope.

What counts as the state of the art

Article 54(2) defines the state of the art as everything made available to the public, by means of a written or oral description, by use, or in any other way, before the filing date of the European patent application — or before the priority date, where priority is validly claimed.

Every word of that definition does work. There is no geographic limit: a disclosure in any country counts. There is no language limit: a thesis in Korean or a patent published only in Portuguese is as damaging as an English-language journal article. There is no form requirement: a conference talk, a poster session, a product demonstration, a sales brochure, a publicly accessible repository commit, or prior commercial use can each destroy novelty just as effectively as a granted patent.

The operative test is availability, not actual access. A document sitting in a public library that nobody ever read is still prior art. Conversely, information shared under a genuine obligation of confidentiality has not been made available to the public and does not count — which is why well-drafted non-disclosure agreements matter so much before filing.

The single-reference standard

Novelty under the EPC is assessed against each piece of prior art individually. To anticipate a claim, a single disclosure must directly and unambiguously disclose every feature of that claim, read through the eyes of the person skilled in the art using their common general knowledge.

Combining two documents to assemble all the claimed features — mosaicking — is not permitted for novelty. If feature A appears in one paper and feature B in another, the claim is novel over both. Whether the skilled person would obviously combine them is a different question, answered under Article 56 as inventive step. Keeping the two analyses separate is essential: a claim can survive Article 54 comfortably and still fall under Article 56.

The directly-and-unambiguously test is strict in both directions. A vague or ambiguous passage that might be read as disclosing a feature generally does not anticipate. But the disclosure does not need to use the same words as the claim: synonyms, equivalent technical expressions, and what the skilled person would inevitably understand all count.

Article 54(3): the invisible prior art

One category of prior art cannot be found by any search at your filing date. Under Article 54(3), the content of European patent applications with an earlier filing or priority date, but published on or after your filing date, forms part of the state of the art — for novelty only.

These earlier-filed, later-published applications are unpublished and therefore invisible when you file. They count for the novelty assessment but are expressly excluded from inventive step, so they cannot be combined with other documents against you. The practical consequence is that drafting should always include fallback positions: if an Article 54(3) document later surfaces, you need claim features to retreat to that the conflicting application does not disclose.

Article 55: narrow exceptions, not a grace period

Article 55 excuses a disclosure in only two situations: where it resulted from an evident abuse in relation to the applicant — for example, a breach of confidence — or where the applicant displayed the invention at an officially recognised international exhibition. In both cases the disclosure must have occurred no earlier than six months before the actual filing of the European application, and the exhibition route requires a declaration and supporting certificate at filing.

These exceptions are read narrowly, and the six-month period is counted from the actual European filing date rather than the priority date. The EPC has no general grace period of the kind available in the United States. For European purposes, the working assumption must be that any pre-filing disclosure by the inventors is fatal.

Feature-by-feature analysis in practice

A rigorous novelty assessment is mechanical in structure even if it requires judgement at each step. First, decompose the independent claim into its individual technical features. Then take each prior art document in turn and ask, feature by feature: is this feature directly and unambiguously disclosed here, explicitly or implicitly? A document anticipates only if the answer is yes for every feature. One missing feature means the claim is novel over that document — though that document may still anchor an inventive-step attack.

Implicit disclosure deserves care. A feature is implicitly disclosed if it is the inevitable consequence of carrying out what the document teaches, even if never stated. A document describing a process that necessarily produces a particular structure discloses that structure. What is merely probable or plausible, however, is not implicit disclosure.

The most common trap is self-disclosure. The state of the art does not exclude the inventor's own work. A conference abstract, a master's thesis deposited in a university library, a grant report posted online, a YouTube demonstration, or an earlier patent application published before the new filing date are all prior art against the inventors themselves. For university tech-transfer teams, screening the research group's own publication record is often the single highest-value step in the whole assessment.

A practical pre-filing workflow

A structured novelty screen before drafting costs a fraction of the prosecution it prevents. Drafting around known prior art produces stronger claims; discovering that art at examination produces expensive amendments, narrowed scope, or abandonment after the fees are sunk.

A workable pre-filing routine looks like this:

• —Write the broadest claim you intend to pursue and decompose it into discrete features, marking which features carry the inventive contribution.
• —Search with synonym expansion, not just the inventors' own vocabulary — different fields name the same mechanism differently, and the most damaging document often uses none of your keywords.
• —Use patent classification codes (CPC or IPC) alongside keyword queries to catch documents that text search misses.
• —Cover non-patent literature: journals, conference proceedings, theses, standards documents, and preprint servers, which dominate the prior art in many academic fields.
• —Screen the inventors' own publications, talks, posters, and public deposits against the intended filing date.
• —Record, for each candidate document, which claim features it discloses and where — a feature-by-feature matrix, not a gut impression.
• —Decide: file as drafted, redraft around the closest art, or stop before spending more.

Reading the result honestly

The output of a good novelty screen is not a verdict of patentable or not patentable. It is a map: which features survive contact with the closest art, which are already disclosed, and where the genuine contribution sits. That map drives drafting strategy — the independent claim should be built on the surviving features, with the disclosed ones moved into the preamble or dependent claims.

It also calibrates expectations. If every document discloses some features but none discloses all of them, the novelty case is sound and the real fight will be inventive step. If a single document covers everything, no amount of clever drafting fixes it, and finding that out before filing is the cheapest possible outcome.

How Patentopia runs this analysis

Patentopia operationalises exactly the workflow described above. A submitted claim is decomposed into individual features, labelled A, B, C and weighted by importance. Each feature is then checked against every retrieved document — patents and non-patent literature — applying the Article 54 single-reference standard, with quoted disclosure locations so each finding can be verified rather than taken on trust. The result is a per-feature novelty matrix showing precisely which features each document discloses, and human-in-the-loop review gates let a patent professional inspect and approve each stage before the analysis proceeds.

The legal standard does not change with the tooling; what changes is the cost of applying it rigorously before filing instead of discovering the answer during examination.