35 U.S.C. § 112: Written Description, Enablement, and Definiteness

The requirements of 35 U.S.C. § 112 govern what a patent application must disclose and how its claims must be written. The statute has six subsections. Section 112(a) demands three things at once: a written description of the invention, an enabling teaching of how to make and use it, and disclosure of the best mode the inventor knows. Section 112(b) requires claims that define the invention with reasonable certainty. Sections 112(c) through (e) set the rules for dependent and multiple dependent claims. Section 112(f) allows functional claiming through “means for” language, with a significant catch on scope. Failure on any of these fronts is one of the most common grounds for rejection during examination and invalidation in litigation.

Written Description

Section 112(a) first requires the specification to describe the invention in enough detail to show the inventor actually had it in hand on the filing date. This is often called the possession test. The specification, drawings, and original claims together must demonstrate that the inventor conceived the specific invention being claimed, not something adjacent to it.

The Federal Circuit confirmed in Ariad Pharmaceuticals, Inc. v. Eli Lilly and Co. that written description is a standalone requirement, separate from enablement.¹ You can teach a skilled reader how to build something and still fail written description if the filing does not show you possessed the specific invention you are now claiming.

The practical bite comes when applicants try to broaden claims after filing. If your original application describes a specific type of battery, you cannot later amend the claims to cover all energy storage devices unless the original disclosure supported that scope. Written description keeps patent rights tethered to what you actually showed on day one.

Enablement

The same subsection also requires the specification to teach a person skilled in the art how to make and use the invention without undue experimentation.² The skilled person is not a novice but someone with ordinary expertise in the field. The disclosure need not be a step-by-step recipe, but a reader with that ordinary expertise must be able to reproduce the invention without excessive trial and error.

The Wands Factors

Courts and examiners weigh eight factors from In re Wands when deciding whether experimentation crosses into “undue”:³

Breadth of the claims. Broader claims demand more thorough disclosure.
Nature of the invention. Some inventions are inherently harder to describe reproducibly.
State of the prior art. A well-explored field needs less hand-holding.
Level of ordinary skill. A field populated by PhDs tolerates more technical shorthand.
Predictability of the art. Mechanical inventions tend to be predictable; biotech and chemistry far less so.
Amount of direction provided in the specification.
Existence of working examples. Concrete demonstrations carry significant weight.
Quantity of experimentation actually required to fill in the gaps.

No single factor decides the question. Broad claims in an unpredictable field with no working examples are in trouble. A narrow claim in a mature mechanical art may survive on minimal description.

Amgen v. Sanofi

The Supreme Court’s 2023 decision in Amgen Inc. v. Sanofi reinforced that enablement must cover the full scope of what is claimed. Amgen had patented an entire class of antibodies defined by function: binding to a specific protein and blocking it from interacting with cholesterol receptors. The Court held that describing a handful of antibodies and then claiming every antibody performing that function was a “hunting license,” not enablement. Scientists would still have to run extensive trial-and-error work to identify which antibodies actually worked. The lesson generalizes: the more broadly you claim, the more you must enable. A few examples cannot carry a claim to an entire genus when the rest can only be found by painstaking experimentation.

Best Mode

The last piece of § 112(a) requires you to disclose the best way you personally know to carry out the invention at the time of filing.² The inquiry is subjective. The question is not what the objectively best method is; it is whether the inventor had a preferred approach and chose to conceal it. An inventor who knows a particular temperature produces the best results cannot claim the process at a generic temperature range while keeping the optimal setting as a trade secret.

The Leahy-Smith America Invents Act changed the consequences. Since 2011, failing to disclose the best mode cannot be used to invalidate an issued patent in litigation.⁴ The requirement still exists on paper, and examiners can enforce it during prosecution, but it has lost most of its practical teeth. Competitors cannot use a best-mode failure to challenge an issued patent in court.

Definite Claims

Section 112(b) requires the specification to end with claims that particularly point out and distinctly claim the invention.² Claims function as the property lines of the patent. If competitors cannot tell where your territory ends and their freedom to operate begins, the claims fail.

The Supreme Court set the standard in Nautilus, Inc. v. Biosig Instruments, Inc.: a claim is indefinite when it fails to inform skilled persons “with reasonable certainty” about the scope of the invention.⁵ Absolute precision is not required, but vagueness that leaves the public guessing is not tolerated.

Relative terms like “about,” “substantially,” and “similar” are not automatically fatal, but they need an anchor. If the specification supplies examples, figures, or context a skilled reader can use to fix the term’s boundaries, the claim can survive. A claim reciting a stretch rate “exceeding about 10% per second” was upheld because infringement could be measured with a stopwatch. A claim reciting a “similar apparatus” with no specification guidance was struck down as indefinite.⁶ Imprecise language works when the specification gives skilled readers enough context to draw a line. Naked approximations invite rejection.

Dependent and Multiple Dependent Claims

Sections 112(c) through (e) govern how claims relate to each other. A dependent claim refers back to an earlier claim and adds a further limitation. If independent claim 1 describes a chair, dependent claim 2 might specify that the chair has an adjustable backrest. Layering claims this way is protective: if a broad independent claim is invalidated, narrower dependent claims may still survive.²

A multiple dependent claim references more than one earlier claim in the alternative, for example, “The chair of claim 1 or claim 2, further comprising armrests.” Useful for efficient drafting, but restricted: a multiple dependent claim cannot serve as the basis for another multiple dependent claim, which prevents endlessly branching claim trees.²

Fees reflect the complexity. Any application containing a multiple dependent claim triggers a $925 surcharge at the standard rate ($370 small entity, $185 micro entity). Each claim beyond 20 total adds $200 ($80 small, $40 micro), and each independent claim beyond the first three adds $600 ($240 small, $120 micro).⁷ The bill grows quickly on applications with extensive claim sets, and patent attorneys often trade comprehensive coverage against filing cost.

Means-Plus-Function Claims

Section 112(f) lets you describe a claim element by what it does rather than what it is. Instead of listing every circuit configuration that could process data, you can claim “a means for processing data.”² The shortcut carries a real tradeoff. The claim’s legal scope is limited to the specific structures described in the specification and their equivalents.⁸ The claim reads broadly on its face and is construed narrowly against what the patent actually discloses.

The phrases “means for” and “step for” create a presumption that § 112(f) applies. When those words appear, courts and examiners look to the specification for corresponding structure. If the specification identifies no structure that performs the claimed function, the claim is invalid for indefiniteness. When a claim avoids “means for” and “step for,” a strong presumption runs the other way, even if the claim uses functional language, and that presumption is hard to overcome.

The strategic tension is real. Means-plus-function format spares you from listing every embodiment, but it ties your protection to what you actually disclosed. Competitors who reach the same result through a fundamentally different structure may fall outside the claim entirely. Many practitioners avoid the format when they can, preferring structural language that can be interpreted more broadly.

Interaction With the New-Matter Bar

Section 112 works alongside 35 U.S.C. § 132, which forbids introducing “new matter” into an application through amendment.⁹ The written description requirement supplies the test. Any new claim language or specification edit must find support in the application as originally filed. Support can come from the specification text, the original claims, or the drawings, and does not need to be word-for-word identical. But the original filing must reasonably convey to a skilled reader that the inventor possessed the subject matter now being claimed.

The rule catches applicants trying to expand reach during prosecution. Adding a limitation that was never described, deleting a key limitation to broaden coverage, or redefining a specification term can each draw a new-matter rejection. The standard is stricter than it looks: disclosure that merely makes the amended claim obvious is not enough. The original filing must show possession of the specific subject matter.

Responding to a § 112 Rejection

When an examiner issues an office action with a § 112 rejection, you have three months from the mailing date to respond. That deadline can be extended up to six months total by paying escalating extension fees, but missing the final deadline abandons the application.

The right response depends on which requirement triggered the rejection. For written description or enablement under § 112(a), you can narrow the claims so they are fully supported by the existing disclosure, or argue that the specification already contains support the examiner overlooked. Adding fresh description to the specification is usually not available because of the new-matter bar. For definiteness under § 112(b), the fix often means replacing vague terms with more precise language or pointing to specification passages that supply context for relative terms. For a § 112(f) rejection, you may need to amend the specification to identify the structure performing the function, or rewrite the claim to avoid means-plus-function format altogether.

The worst outcome is a final rejection you cannot overcome. Remaining options narrow to a continuation application (which restarts prosecution on the same disclosure), an appeal to the Patent Trial and Appeal Board, or abandonment. Most § 112 issues are fixable with careful amendment, but the fix almost always leaves you with narrower claims than you originally wanted.

1
United States Patent and Trademark Office. MPEP 2163 – Guidelines for the Examination of Patent Applications Under the 35 USC 112(a) Written Description Requirement
2
Office of the Law Revision Counsel. 35 USC 112 Specification
3
United States Patent and Trademark Office. MPEP 2164 – The Enablement Requirement
4
United States Patent and Trademark Office. America Invents Act (AIA) Frequently Asked Questions
5
Justia Law. Nautilus Inc v Biosig Instruments Inc, 572 US 898 (2014)
6
United States Patent and Trademark Office. MPEP 2173 – Claims Must Particularly Point Out and Distinctly Claim the Subject Matter
7
United States Patent and Trademark Office. USPTO Fee Schedule
8
United States Patent and Trademark Office. MPEP 2181 – Identifying and Interpreting a 35 USC 112(f) Limitation
9
Office of the Law Revision Counsel. 35 USC 132