The more you claim, the more you must enable

How to read the ledger below. Each numbered point carries the kind of evidence behind it. ESTABLISHED is stated in the authority's own text, with a locator. IMPLEMENTED is the rule as the authority actually applied it. OBSERVED is carried by a source outside this one. ASSUMED is the drafter's inference, flagged as such and not the authority speaking. Quotations are checked against a cached copy of the primary source by script.

In one paragraph

A unanimous Supreme Court affirmed the Federal Circuit in Amgen v. Sanofi on 18 May 2023 and held that a patent claiming an entire class of things must enable a skilled person to make and use the entire class, so the more one claims the more one must enable. Amgen disclosed 26 working antibodies and two methods for finding others, and the Court held those methods amounted to two research assignments that left scientists to random trial and error rather than enabling the claimed class. The Court left two qualifiers in place, that a specification may disclose a general quality running through the class instead of every embodiment, and that a reasonable amount of experimentation is permitted depending on the art, and it agreed there is one enablement standard rather than a special one for genus claims.

Reader summary

Amgen's claims covered every antibody that binds to a particular region of a protein and blocks it from binding to a receptor: a class defined entirely by what its members do. The specification identified 26 such antibodies by their amino acid sequences and gave two methods, which Amgen called a roadmap and conservative substitution, for finding others.

The Court affirmed invalidity, and built the holding out of four nineteenth and twentieth century cases rather than out of anything modern. The through line is one sentence: if a patent claims an entire class, the specification must enable a skilled person to make and use the entire class.

Two qualifiers keep that from being absolute, and both matter more to a drafter than the holding does. A specification need not describe every embodiment if it discloses a general quality running through the class that gives it a particular fitness for the purpose. And a specification may call for a reasonable amount of experimentation, with reasonableness depending on the invention and the art.

What Amgen offered fell outside both. The roadmap described its own trial-and-error screening procedure. Conservative substitution asked scientists to mutate known antibodies and test whether the results still worked. The Court's answer was that these amount to two research assignments, and its illustration is the one this case is remembered by: an inventor who discloses 26 working combinations of a 100-tumbler lock and then claims all working combinations has not enabled them.

The Court also rejected two framings expressly. Enablement is not measured by the cumulative time it takes to make every embodiment, and there is one statutory enablement standard rather than a special one for functional genus claims. Both concessions are useful, because they are what a practitioner can rely on when this case is cited against them too broadly.

Why this matters when you are drafting

Breadth is drawn from a specification you cannot amend. Claims 1 and 2. A continuation claim that widens scope, especially by functional language, is asking the parent to enable more than the parent's claims did. The disclosure is fixed; the demand on it is not.

The general-quality disclosure is the thing to look for and, in a parent, to write. Claims 3 and 9. Where the specification says what the members of a class have in common that makes them work, an example or two can carry the class. Where it only lists working examples, the class is the exposure. This is the single most actionable sentence in the opinion.

A protocol is not a disclosure of what the protocol would find. Claims 7 and 8. Screening instructions, substitution guidance and a description of the applicant's own search method are research assignments. They read like enablement in a specification and do not function as it.

Know the two concessions. Claims 12 and 13. When this case is cited against a broad claim, the response is that neither the cumulative effort to build out a genus nor genus claiming itself is the test. There is one standard, applied proportionately to what is claimed.

Reasonable experimentation is real and it is field-dependent. Claim 4. Some adaptation is permitted, and how much depends on the art. That is a factual question and it is where an unpredictable field costs a patentee the most.

This is the enablement half of what Tier B mostly treats as a written-description problem. seagen-v-daiichi-sankyo applies this case directly to invalidate a continuation on both grounds at once. The two requirements are distinct and a broad continuation claim can fail either.

What the authority establishes

[ESTABLISHED, the holding] "If a patent claims an entire class of processes, machines, manufactures, or compositions of matter, the patent's specification must enable a person skilled in the art to make and use the entire class." (P. 13.)

[ESTABLISHED, the same rule in the phrase everyone quotes] "the specification must enable the full scope of the invention as defined by its claims. The more one claims, the more one must enable." (P. 13.)

[ESTABLISHED, the first qualifier, and it is the drafting lever] "That is not to say a specification always must describe with particularity how to make and use every single embodiment within a claimed class." It may suffice to give an example or a few if the specification also discloses "some general quality" running through the class that gives it "a peculiar fitness for the particular purpose." (P. 13.)

[ESTABLISHED, the second qualifier] "a specification may call for a reasonable amount of experimentation to make and use a patented invention. What is reasonable in any case will depend on the nature of the invention and the underlying art." (P. 15.)

[ESTABLISHED, and this is the limit on the qualifier] "in allowing that much tolerance, courts cannot detract from the basic statutory requirement" that the specification describe the invention in terms enabling a skilled person to make and use it. (P. 15.)

[ESTABLISHED, what was conceded and what was not] "we do not doubt that Amgen's specification enables the 26 exemplary antibodies it identifies by their amino acid sequences." (P. 15.) The defect was scope, not disclosure quality.

[ESTABLISHED, the failure, stated in five words] Amgen's two methods "amount to little more than two research assignments." (P. 16.)

[ESTABLISHED, why each failed] The first "merely describes step-by-step Amgen's own trial-and-error method for finding functional antibodies" (P. 16.), and the second "requires scientists to make substitutions to the amino acid sequences of antibodies known to work and then test the resulting antibodies to see if they do too" (P. 17.).

[ESTABLISHED, the door the Court deliberately left open] Whether methods of this kind "might suffice to enable other claims in other patents" (P. 17.) is not decided, and the Court points to the possibility that "the inventor identifies a quality common to every functional embodiment" (P. 17.) Read with claim 3, that is the same lever stated twice.

[ESTABLISHED, what the methods left a scientist with] They "leave a scientist about where Sawyer and Man left Edison" (P. 17.), and that "is not enablement. More nearly, it is" (P. 17.) a hunting license, borrowing the phrase from Brenner v. Manson.

[ESTABLISHED, the analogy, and it is the most portable thing in the opinion] An inventor who discloses 26 successful combinations of a lock with 100 tumblers and then "tries to claim much more, namely all successful combinations" (P. 17.) while telling others "to randomly try a large set of combinations and then record the successful" (P. 17.) ones has produced something that "would not enable others to make and use functional combinations" (P. 17.) but instead leaves them to random trial-and-error discovery.

[ESTABLISHED, the first framing the Court rejected, and it helps patentees] "enablement is not measured against the cumulative time and effort it takes to make every embodiment within a claim". (P. 18.) Sheer volume of work is not the test.

[ESTABLISHED, the second framing the Court rejected, and it also helps patentees] "There is one statutory enablement standard." (P. 18.) There is no heightened bar for functional genus claims; there is only the consequence that "the more a party claims for itself the more it must enable." (P. 18.)

[ESTABLISHED, where the policy argument goes] Balancing incentives against public benefit "is a policy judgment that belongs to Congress." (P. 19.) Arguments about chilling innovation are not arguments this Court will weigh.

[ESTABLISHED, the bargain framing that closes it] Section 112 "reflects Congress's judgment that if an inventor claims a lot, but enables only a little, the public does not receive its benefit of the bargain." (P. 19.)

[IMPLEMENTED, and it is what makes this case usable rather than just quotable] Claims 3, 4 and 9 together define the safe harbour the holding leaves: an example or two plus a disclosed common quality, with a reasonable amount of adaptation permitted. The Court applies that frame to find Amgen outside it rather than announcing a new test, which is why claims 12 and 13 could be conceded without changing the result.

[ASSUMED, drafter's inference, flagged as such] For continuation practice the operative consequence is that a functionally defined claim added later inherits the parent's enablement, not the applicant's current knowledge. Nothing learned after filing helps, for the same reason nothing learned after filing helps written description. The opinion says nothing about continuations; this follows from claims 1 and 6 read against the fixed disclosure a continuation is stuck with.

Quotations, verified against the source

All from the opinion of the Court, never the syllabus. Page locators computed from the slip opinion's running header, which numbers the opinion separately from the syllabus.

"If a patent claims an entire class of processes, machines, manufactures, or compositions of matter, the patent's specification must enable a person skilled in the art to make and use the entire class." (P. 13.)
"the specification must enable the full scope of the invention as defined by its claims. The more one claims, the more one must enable." (P. 13.)
"That is not to say a specification always must describe with particularity how to make and use every single embodiment within a claimed class." (P. 13.)
"some general quality" (P. 13.)
"a peculiar fitness for the particular purpose." (P. 13.)
"a specification may call for a reasonable amount of experimentation to make and use a patented invention. What is reasonable in any case will depend on the nature of the invention and the underlying art." (P. 15.)
"in allowing that much tolerance, courts cannot detract from the basic statutory requirement" (P. 15.)
"we do not doubt that Amgen's specification enables the 26 exemplary antibodies it identifies by their amino acid sequences." (P. 15.)
"amount to little more than two research assignments." (P. 16.)
"merely describes step-by-step Amgen's own trial-and-error method for finding functional antibodies" (P. 16.)
"requires scientists to make substitutions to the amino acid sequences of antibodies known to work and then test the resulting antibodies to see if they do too" (P. 17.)
"might suffice to enable other claims in other patents" (P. 17.)
"the inventor identifies a quality common to every functional embodiment" (P. 17.)
"leave a scientist about where Sawyer and Man left Edison" (P. 17.)
"is not enablement. More nearly, it is" (P. 17.)
"tries to claim much more, namely all successful combinations" (P. 17.)
"to randomly try a large set of combinations and then record the successful" (P. 17.)
"would not enable others to make and use functional combinations" (P. 17.)
"enablement is not measured against the cumulative time and effort it takes to make every embodiment within a claim" (P. 18.)
"There is one statutory enablement standard." (P. 18.)
"the more a party claims for itself the more it must enable." (P. 18.)
"is a policy judgment that belongs to Congress." (P. 19.)
"reflects Congress's judgment that if an inventor claims a lot, but enables only a little, the public does not receive its benefit of the bargain." (P. 19.)

Tensions and open questions

How much common quality is enough is not addressed. Claims 3 and 9 both point at the same escape route without describing it. The Court says disclosing a general quality "may reliably enable" the class in some cases, which is a possibility rather than a standard.

The line between reasonable experimentation and a research assignment is undrawn. Claim 4 permits the first and claim 7 condemns the second, and Amgen's methods are described as falling on the wrong side without a criterion that would place a closer case.

Claim 9 is an express reservation and it is often read away. The Court says methods of this kind might suffice in other patents. Citing this case for the proposition that screening protocols never enable overstates it.

Nothing here quantifies scope. The opinion involves 26 disclosed antibodies against at least millions claimed, and 26 lock combinations against a space of 20 to the power of 100. Those figures describe these facts. The opinion states no ratio, and the CP1 section below forbids deriving one.

The interaction with written description is left alone. The requirements are distinct, and this case decides only enablement. A specification can fail one and satisfy the other, which is why seagen-v-daiichi-sankyo had to reach both separately.

What this record does and does not cover

The statutory analysis, the historical cases as the Court uses them, both qualifiers, the application to Amgen's two methods, the rejected framings and the closing were read end to end.

What was NOT read closely, and is therefore outside what this record can support: the science, the record below, and the amici briefing beyond the lock analogy the Court adopts. This record says nothing about written description, which the Court did not decide, nor about whether any narrower Amgen claim would have survived. The Federal Circuit's opinion is known here only through this Court's account of it.

Currency

This record was verified against its primary source on 2026-08-15. That is a fidelity check: the quotations, locators and attribution are accurate to that source. Whether the authority has since been narrowed, distinguished, abrogated or reversed has not been checked here. Confirm it still stands before relying on it.