Innovate Articles

Selection Inventions in the U.S.: Novelty, Obviousness, and Written Description in a Tightening Landscape

Samantha Page

Introduction

Selection inventions occupy a familiar space in U.S. patent practice: the claim targets a specific species, subgenus, formulation, or numerical sub-range that lies within a broader disclosure. In the chemical and materials arts in particular, business-relevant innovation is often found not in an entirely new class of compounds or compositions, but in identifying which members of a known field deliver a materially improved result.

U.S. law does not apply a stand-alone “selection invention” test. Instead, these claims rise or fall under familiar requirements: novelty (§102), nonobviousness (§103), and written description (§112). What has changed in recent years is not the statutory framework, but the intensity of the fact-finding, particularly on written description for genus and subgenus claiming. This article highlights the practical issues selection inventions raise at the USPTO and in litigation, with an emphasis on lessons from recent Federal Circuit decisions.

What is a Selection Invention?

A selection invention claims a narrower subset of subject matter that is encompassed by a broader disclosure—often the applicant’s own earlier filing, a competitor’s publication, or a crowded body of art. For example, in the chemical, pharmaceutical, and materials arts, small structural or compositional changes can produce big differences in properties (stability, manufacturability, safety, conductivity, bioavailability, and the like). But the concept applies just as well to other technical areas when claims are drafted as a subset of disclosed alternatives.

Common selection patterns include:

• A narrow numerical sub-range selected from a broader disclosed range;
• A specific species (or small subset of species) selected from a disclosed genus; or
• A particular combination chosen from many possible component lists (e.g., which salt, dopant, additive, or excipient to use).

Because selection claims carve out a subset from broader disclosures, disputes tend to follow familiar lines: genus-species anticipation, motivation-to-select obviousness, and whether the original filing supports the subset.

The Statutory Framework: §102, §103, and §112

At the USPTO (and later, in court), selection inventions are examined under the same provisions as any other claim: novelty under 35 U.S.C. §102, nonobviousness under §103, and disclosure (including written description and enablement) under §112. The fact that a claim is a “selection” typically changes the facts and the arguments rather than the legal test.

Novelty (§102) and the “At Once Envisage” Question

A common novelty dispute for selection inventions is whether a broad disclosure (a genus or a broad range) anticipates a narrower claim (a species or sub-range). The U.S. anticipation standard remains strict: a single reference must disclose every limitation, expressly or inherently. A claim is not automatically anticipated just because it falls within a disclosed range or genus.

In In re Petering [1], the court articulated what has become the classic framework for genus–species anticipation in selection cases. There, the prior art disclosed a narrowly defined class of approximately twenty compounds. Because the reference effectively described a limited and predictable set, the court held that a person of ordinary skill would “at once envisage” each member of the class, rendering an individual species anticipated. Importantly, Petering did not establish a numerical threshold or a general rule applicable to all disclosed genera. Rather, it stands for the fact‑intensive proposition that anticipation may lie where the prior art defines a genuinely small and structurally circumscribed class such that each member is immediately apparent to the skilled artisan.

The Federal Circuit’s decision in Mylan v. Merck [2] reinforces the narrow, fact‑dependent reach of Petering. The prior art in Mylan disclosed sitagliptin as one of thirty‑three compounds in a genus of DP‑IV inhibitors and identified phosphoric acid among eight “particularly preferred” acids for salt formation. Accounting for multiple stoichiometries, the disclosure was treated as encompassing 957 predicted salts, some of which might not even form under experimental conditions. The court affirmed the PTAB’s finding that such a disclosure did not cause a skilled artisan to “at once envisage” the claimed 1:1 sitagliptin dihydrogenphosphate salt. In doing so, the court emphasized that selection from a large combinatorial space does not amount to anticipation merely because the claimed species falls within the literal scope of the disclosure.

Obviousness (§103): Why Pick That One?

Selection inventions often face their hardest test under §103. The question is not simply whether the prior art “covers” the claim; it is whether the art would have given a skilled artisan a reason to make the selection and a reasonable expectation of success. Evidence of unexpected results or “criticality” of a claimed range can be powerful and, in many selection cases, practically necessary to rebut an asserted prima facie case of obviousness, even though it is not a formal requirement.

In re Baird [3] is frequently cited for its treatment of obviousness in the context of selection inventions. There, the Federal Circuit rejected an obviousness theory premised solely on the fact that the claimed compound fell within a broadly disclosed genus. The court explained that obviousness requires more than generic coverage; the prior art must supply a reasoned basis for selecting the particular species claimed, with a reasonable expectation of success. Where the record offers no teaching or motivation explaining why a skilled artisan would have singled out the claimed compound from among many alternatives, obviousness cannot be sustained.

A recent nonprecedential decision, Maxell, Ltd. v. Amperex Technology Ltd. [4], illustrates how selection arguments can fail under §103 when the record supplies a concrete, property-driven rationale for choosing the claimed subset. There, the dispute centered on selecting Mg and Al (from among disclosed options) for a lithium-ion cathode material and selecting amounts within an overlapping prior-art range; the Board credited unrebutted expert testimony corroborated by technical references, and the Federal Circuit affirmed while rejecting the argument that the combination was merely a “needle in a haystack.”

Practically, obviousness of selection inventions often turns on what the record actually teaches about why a skilled artisan would pick that member. If the Office can identify a concrete reason to make the selection, the applicant typically needs objective evidence to respond.

Written Description (§112): The “Selection” Must Be in the Filing

While §§102 and 103 remain central, recent case law has shifted attention for selection claims to §112. Written description asks whether the application as filed reasonably conveys to a person of ordinary skill that the inventor was in possession of the claimed subject matter as of the filing date. In the context of chemical and biotechnology genus—and subgenus—claims, the Federal Circuit has consistently required, as articulated in Ariad Pharms. v. Eli Lilly & Co. [5], either (i) disclosure of a representative number of species within the claimed scope, or (ii) disclosure of common structural features sufficient to allow skilled artisans to “visualize or recognize” the members of the claimed genus.

In Seagen v. Daiichi Sankyo [6], the Federal Circuit held that written description for selection-style genus/subgenus claims requires more than literal encompassment within a broad disclosure. The asserted claims depended on priority to an earlier application that encompassed over 47 million possible tetrapeptide units while the claimed Glycine/Phenylalanine-only tetrapeptide subgenus encompassed 81 species, yet the earlier filing did not identify the later-claimed subset with reasonably specific disclosure. The Seagen decision reinforces that for selection-style claiming under §112, the selection itself must be disclosed—not reconstructed after the fact from a vast genus.

The Federal Circuit’s decision in Duke University v. Sandoz [7] further emphasizes the narrow margin for sustaining chemical subgenus claims without disclosure that identifies the claimed subset with sufficient clarity. The asserted claim covered a subgenus of prostaglandin analogs defined by structural features not expressly exemplified in the specification. The court reversed the validity verdict, explaining that the specification disclosed neither a representative number of species within the claimed subgenus nor meaningful “blaze marks” directing the skilled person from a disclosure the court characterized as describing billions of compounds to the smaller set actually claimed. The court also rejected reliance on a generic “hairpin” backbone as a supposed common structural feature because it was shared by the entire disclosed universe, and therefore did not guide the skilled artisan to the claimed subset. Duke highlights that where commercial value is expected to reside in a narrowed selection, reliance on broad genus disclosure alone is often insufficient to satisfy §112 absent exemplification or other disclosure that genuinely points to the selection (rather than to features shared across the entire disclosed universe).

Prosecuting Selection Claims

Taken together, these cases confirm that U.S. law does not treat “selection inventions” as a distinct doctrinal category; instead, recent decisions increasingly demand that the selection be justified twice: once by what the prior art would or would not lead a skilled artisan to choose under §§102 and 103, and again by what the application as filed actually demonstrates the inventors possessed under §112.

Selection claims tend to prosecute best when the specification makes the subset unmistakable and explains why it matters. If you expect value to reside in a particular species, subgenus, or sub-range, identify it explicitly in the application—not just in the claims—and provide clear guideposts supporting it as of the filing date. Where feasible, include examples that fall within the likely “commercial” subset, along with at least one comparison (e.g., a nearby species or an out-of-range example) to support later arguments on unexpected results or criticality. Finally, draft with the obviousness record in mind: anticipate motivation-to-select rationales and document relevant constraints and tradeoffs. Post-filing data can help rebut §103, but it cannot cure §112 support gaps.

Conclusion

Selection inventions remain a practical way to extend or deepen patent coverage in crowded fields. The strongest selection invention filings are not just narrower; they are more explicit about what was selected, why it was selected, and what the inventors knew about the selection as of the filing date.

References

[1] In re Petering, 301 F.2d 676 (C.C.P.A. 1962).

[2] Mylan Pharms. Inc. v. Merck Sharp & Dohme Corp., 50 F.4th 147 (Fed. Cir. 2022).

[3] In re Baird, 16 F.3d 380 (Fed. Cir. 1994).

[4] Maxell, Ltd. v. Amperex Technology Ltd., Nos. 2023-2256, 2023-2257, 2023-2258 (Fed. Cir. Jan. 23, 2025) (nonprecedential).

[5] Ariad Pharms., Inc. v. Eli Lilly & Co., 598 F.3d 1336 (Fed. Cir. 2010) (en banc).

[6] Seagen Inc. v. Daiichi Sankyo Co., Ltd., 160 F.4th 1322 (Fed. Cir. 2025).

[7] Duke Univ. v. Sandoz Inc., 160 F.4th 1305 (Fed. Cir. 2025).