Every patent question eventually returns to one idea, the prior art. Prior art is the body of what was already known before an inventor sought a patent, and it is the measure against which an invention is judged new enough, and inventive enough, to deserve one. It decides whether a patent can be obtained at all, and years later it is the most common reason a granted patent is struck down. The opening article described what a patent is. This article describes the foundation on which every patent stands or falls. As with the rest of the series, this is general information rather than legal advice, and it assumes filing in the United States, with notes where other jurisdictions differ.

A Brief History of the Novelty Requirement

The demand that an invention be new is as old as patents themselves. The Venetian statute of 1474 protected new and ingenious devices, and the English Statute of Monopolies protected new manufactures. The word new did the work then that the term prior art does now. What changed over the centuries is not the principle but its precision. The modern system defines, in statute and in case law, exactly what counts as already known, exactly when the comparison is made, and exactly how close an invention may come to what came before without being too close. The history of patent law since 1474 is in large part the history of sharpening that single comparison.

What Prior Art Is

Prior art is broader than most people expect. It is not limited to earlier patents. It includes earlier patents and published patent applications, but also printed publications of any kind, academic papers, product manuals, conference talks, theses, and web pages. It includes products that were in public use or on sale. It includes anything otherwise made available to the public before the relevant date. A graduate thesis shelved in one university library, a product sold in one country, a forum post, and an expired patent are all prior art if they disclose the invention.

Prior art is also global and language-blind. A disclosure in any country, in any language, counts. An inventor in one country can be anticipated by a document published decades earlier on another continent that no one involved has ever read. This is why the prior art can never be fully known. It is too large, too scattered, and too old to search exhaustively.

One boundary matters. Prior art is what was available to the public. Information held in secret is generally not prior art, which is the seam where patents and trade secrets meet, and the subject of a later article. A secret process used behind closed doors by its own owner does not necessarily bar that owner from patenting, though the rules here are technical and differ by jurisdiction.

The Priority Date Draws the Line

Prior art is everything that qualifies before a particular date, so the date is half the question. That date is the effective filing date of the application, which a provisional application can set early. Anything that became public before that date is prior art. Anything that became public after it is not.

Because the United States awards a patent to the first inventor to file, the filing date is also the finish line in a race. An inventor who delays risks two things at once, that someone else files first, and that the growing body of publications and products becomes prior art against the later filing. Filing early is, in part, a way of drawing the prior-art line before more of the world can cross it.

The Novelty Bar

The first requirement an invention must clear is novelty, the bar that Section 102 of the patent statute sets. An invention is not novel if a single piece of prior art discloses every element of the claim. When one reference does this, the claim is said to be anticipated, and it cannot be patented as written. The test is strict in form. A single reference must contain all of the claimed elements, arranged as the claim arranges them. Five references that together disclose every element do not anticipate, because novelty is judged against one reference at a time.

Novelty is the easier bar to clear and the easier to understand. Genuinely new combinations are common. The harder bar is the next one.

The Non-Obviousness Bar

An invention can be novel and still fail to deserve a patent because it is obvious. Non-obviousness, the bar of Section 103, asks whether the differences between the invention and the prior art would have been obvious, at the time of filing, to a person of ordinary skill in the art, a hypothetical practitioner who knows all of the relevant prior art and has ordinary competence in the field. Unlike novelty, obviousness may be shown by combining several references, when a skilled person would have had reason to combine them with a reasonable expectation of success.

The framework comes from the Supreme Court. Graham v. John Deere Co. set out the inquiry, namely the scope and content of the prior art, the differences between the prior art and the claim, the level of ordinary skill in the field, and secondary considerations such as commercial success, long-felt but unmet need, and the failure of others. KSR v. Teleflex later loosened a rigid version of the test, holding that a combination of familiar elements by known methods is likely obvious when it yields no more than predictable results. The practical effect of KSR was to make obviousness easier to assert and harder to escape, and it is the bar on which most patentability fights are won and lost.

The Grace Period and Its Limits

The United States offers a narrow safe harbor. An inventor’s own disclosure, made one year or less before filing, does not count as prior art against that inventor. This grace period allows a limited window in which an inventor can publish, present, or sell and still file afterward.

The safe harbor is a trap if it is misread. Most other countries require absolute novelty, with no grace period at all, so a public disclosure before filing destroys the right to a patent in Europe and much of the world, even though the same disclosure would be forgiven in the United States. An inventor who presents a paper or demonstrates a product before filing may keep a United States option while forfeiting every foreign one. This is the practical reason that filing precedes disclosure wherever foreign rights might matter, and it is one more place where the discipline of confidentiality before filing protects an option that loose talk would destroy.

Searching the Prior Art

Two different searches use the same body of prior art to answer two different questions, and they are routinely confused.

A patentability search, sometimes called a novelty search, looks for prior art that would bar the inventor’s own application. It asks whether the invention is likely new and non-obvious before the cost of filing is incurred. A freedom-to-operate search asks the opposite-facing question, whether making or selling a product would infringe the in-force patents of others. The first protects the application. The second protects the product. An invention can be perfectly patentable and still infringe someone else’s patent, and a product can be free to operate while the underlying invention is unpatentable, so both searches are run when both questions matter.

No search is exhaustive. The prior art is too large to search completely, and the most dangerous prior art is often the hardest to find, namely a foreign document, a non-patent publication, or a prior public use that no database indexes well. A clean search lowers the risk. It does not remove it.

Defensive Publication

Prior art can be created on purpose. Defensive publication is the deliberate act of publishing a description of an invention so that it becomes prior art and prevents anyone, including competitors, from later patenting it. It is the mirror image of a patent. A patent discloses in exchange for exclusion. A defensive publication discloses in exchange for nothing but freedom, namely the assurance that the published idea stays open and cannot be fenced off by another.

A company chooses defensive publication when it wants to use an invention but does not want to pay to patent and defend it, and when its main fear is not that a competitor copies the idea but that a competitor patents it and then excludes everyone, including the originator. It is a cheap and durable way to keep a piece of the commons open, and it is a tool the founder should know exists.

Prior Art as the Invalidation Lever

A patent is examined against the prior art the examiner finds, and an examiner has limited time and an imperfect view of a boundless field. A great deal of prior art is never seen during examination. This is why a granted patent is not a final verdict of validity. It is a considered opinion that can be overturned when better prior art surfaces.

Prior art is the principal weapon for overturning it. An accused infringer searches for the prior art the examiner missed, and a single anticipating reference, or a strong obviousness combination, can invalidate a claim. Since 2012, inter partes review has given challengers a dedicated administrative path to attack a patent’s validity on prior-art grounds, using earlier patents and printed publications, and it has invalidated a large share of the claims it has reviewed. The lesson for anyone relying on a patent is that the patent is only as strong as the prior art it has actually survived. Undiscovered prior art is a latent flaw, and it is the reason a patent that looks formidable on its face can prove brittle under challenge. The moat article and the enforcement article return to this point, because it is decisive for whether a patent is worth what its owner hopes.

Epistemic State

The settled matters here are the existence and structure of the two bars, namely novelty judged against a single reference and non-obviousness judged against the prior art as a whole, the Graham and KSR framework for obviousness, the existence of the United States grace period and its absence in most other countries, the distinction between a patentability search and a freedom-to-operate search, and the role of prior art in invalidation and inter partes review. These should hold against independent verification.

The application of these rules is less settled than their structure. Obviousness in particular is fact-intensive and contested, and reasonable examiners and courts reach different conclusions on similar facts. The description of the grace period is United States specific, and the foreign-novelty point is stated at the level of principle rather than country by country.

The strategic observations, namely that no search is exhaustive and that undiscovered prior art is a latent flaw in any patent, are well supported by the invalidation record but remain general claims rather than predictions about a particular patent. Throughout, this is general information, United States centric, and not legal advice.

Out of Scope

The detailed procedure of an inter partes review, the standards of proof, and the tactics of a validity challenge belong to the enforcement article and to counsel. The full case law of obviousness beyond the Graham and KSR framework is a specialist subject. The mechanics of prior-art searching, namely the databases, the classification systems, and the search strategies, are a craft of their own and are not treated here. The country-by-country detail of foreign novelty rules is left to local counsel in each jurisdiction where protection is sought.

Anticipation as Set Containment, an Optional Formalization

This section restates the novelty rule in the language of sets, for readers who find a formal statement clearer than a verbal one. It introduces no new doctrine, and a reader who prefers the prose can skip to the conclusion without loss.

Model a claim as a set of limitations, the discrete elements it requires, written

\[C = \{\, e_1, e_2, \ldots, e_n \,\}.\]

Model a prior-art reference $R$ as the set of features it discloses, written $D_R$. On this model a reference anticipates the claim exactly when every claimed element appears in that one reference, that is, when the claim is a subset of the reference,

\[C \subseteq D_R.\]

The single-reference rule becomes precise. Anticipation requires one reference $R$ for which $C \subseteq D_R$. It is not enough that the claim is covered by the union of several references,

\[C \subseteq D_{R_1} \cup D_{R_2} \cup \cdots \cup D_{R_k},\]

because novelty is judged one reference at a time. A claim can lie inside the union of the prior art as a whole while lying inside no single member of it, and that claim is novel.

Obviousness is the doctrine that reaches the union. Let $M$ denote the predicate that a person of ordinary skill would have had reason to combine the references with a reasonable expectation of success. A first approximation of the obviousness bar is that the claim fails when

\[C \subseteq \bigcup_{i} D_{R_i} \quad \text{and} \quad M.\]

Novelty asks only about containment in a single set. Obviousness asks about containment in a union and then adds the predicate $M$. This is the formal reason novelty is the clean bar and obviousness the contested one, since the predicate $M$ is the entire Graham and KSR inquiry compressed into a single symbol, and all of its difficulty lives there.

Two caveats keep the model honest. A real claim is more than an unordered set, since it also requires its elements to be arranged and related as the claim specifies, so strict anticipation demands not merely $C \subseteq D_R$ but that the reference arrange the elements as the claim does. And the predicate $M$ compresses a fact-intensive body of case law into one symbol, which serves exposition rather than rendering faithfully how a court reasons. With those caveats, the set view captures the essential shapes, that novelty is containment in one set and obviousness is containment in a union together with a reason to form it.

Conclusion

Prior art is the foundation of patentability because it sets both gates. It sets the gate an invention must pass through to earn a patent, through the bars of novelty and non-obviousness, and it sets the gate through which a challenger later attacks, by finding what the examiner did not. A patent is best understood not as a thing an inventor owns outright but as a position held relative to the prior art, strong where the art is sparse and weak where the art is dense. The date draws the line, the two bars measure the distance from the line, and the search, however careful, never sees the whole field. An inventor who understands this files early, discloses late, searches before spending, and treats a granted patent as a strong opinion rather than a certainty. The articles that follow ask what such a patent is worth, namely whether it is a moat, and what it takes to enforce it.

References