Why This Matters
A patent application can survive prosecution, issue as a granted patent, and still be worthless — struck down years later in litigation because the specification failed to meet the written description or enablement requirements of 35 U.S.C. § 112(a). Unlike a typo or a formatting error, a § 112 defect is not something an examiner or opposing counsel discovers and lets slide. It is a substantive, often fatal, flaw. Proofreading a patent application for § 112 compliance is therefore not a mechanical grammar check; it is a legal-technical audit of whether the words on the page actually support the claims being made.
This article lays out what written description and enablement require, how they differ, and a practical proofreading methodology for catching gaps before filing.
The Statutory Text
35 U.S.C. § 112(a) requires that the specification:
- Contain a written description of the invention
- Enable a person skilled in the art to make and use the invention
- Set forth the best mode contemplated by the inventor (largely unenforceable post-AIA, but still nominally required)
Though written description and enablement come from the same sentence, courts treat them as two separate, independently enforceable requirements. A specification can enable an invention without adequately describing it, and vice versa. A proofreader who checks only one has done half the job.
Written Description: Proving Possession
The written description requirement asks a narrow question: does the specification show that the inventor actually had possession of the claimed invention as of the filing date? This is not about whether a skilled artisan could go build the thing — that is enablement’s job. It is about whether the document, read as a whole, demonstrates that the applicant invented what is now being claimed.
This requirement bites hardest in three recurring scenarios:
Claim scope broader than disclosed embodiments. If the specification describes three specific examples but the claims recite a broad genus, the drafter must show the genus was actually possessed — not merely that the three species work. This is the core lesson of Ariad Pharmaceuticals v. Eli Lilly (Fed. Cir. 2010, en banc), which held that a functional claim to blocking NF-κB activity was not supported by a specification that disclosed no structures capable of doing so.
New matter added during prosecution. Amendments, especially claim amendments made in response to an office action, are a prime source of written description problems. If new claim language is not traceable to the original filing, it risks rejection under § 112(a) and cannot rely on the original filing date.
Priority claims to earlier applications. Continuation and continuation-in-part applications frequently claim priority to a parent’s filing date. If the claim language in the child application is not supported by the parent’s disclosure, the priority claim fails, and intervening prior art can invalidate the claim.
Written Description Proofreading Checklist
- For every claim term, locate the specific passage(s) in the specification that describe it. If you cannot point to a passage, flag it.
- For genus claims, confirm the specification discloses either a representative number of species across the genus or structural features common to the genus that correlate with the claimed function.
- For functional claims (“a means configured to achieve X”), confirm structure or algorithm is disclosed, not just the desired result.
- Cross-check every claim amendment against the as-filed specification. Mark any language that cannot be found verbatim or as a clear paraphrase of original disclosure.
- If priority is claimed to an earlier application, perform the same claim-to-specification mapping against the parent’s original disclosure, not just the current specification.
- Watch for claims that were broadened during prosecution (e.g., “a plurality of X” replacing “three X”) without corresponding support added to the specification.
Enablement: Proving Practicability
Enablement asks a different question: could a person of ordinary skill in the art (POSITA), reading the specification, make and use the full scope of the claimed invention without undue experimentation? Enablement is measured as of the filing date and is evaluated against the entire scope of the claim, not just a preferred embodiment.
The Supreme Court’s 2023 decision in Amgen v. Sanofi sharpened this requirement considerably, particularly for claims that cover a broad genus defined by function (there, antibodies that bind a particular epitope and block cholesterol from binding a receptor). The Court held that the more a claim demands of the reader — the more embodiments it sweeps in — the more the specification must enable, and that identifying a handful of working examples plus “trial and error” guidance is not enough if achieving the full scope requires substantial, undue experimentation.
Courts commonly apply the In re Wands factors to assess whether experimentation is “undue”:
- Quantity of experimentation necessary
- Amount of direction or guidance presented
- Presence or absence of working examples
- Nature of the invention
- State of the prior art
- Relative skill of those in the art
- Predictability or unpredictability of the art
- Breadth of the claims
No single factor is dispositive; they are weighed together.
Enablement Proofreading Checklist
- For each independent claim, ask: does the specification enable the full breadth of what is claimed, or only the specific embodiment(s) described?
- In unpredictable arts (biotech, chemistry, pharma), be especially skeptical of broad functional claims supported by only a few working examples. Post-Amgen, this is the single highest-risk pattern.
- Confirm at least one complete, operative example or working embodiment is described in enough detail that a POSITA could reproduce it without inventive effort.
- Check for claim limitations that require parameters, ranges, or conditions (temperatures, concentrations, tolerances) not actually disclosed anywhere in the specification.
- Look for unsupported superlatives or open-ended claim language (“any suitable material,” “substantially any configuration”) and verify the spec gives a POSITA a basis for selecting among the possibilities.
- If the invention depends on software, algorithms, or means-plus-function language, confirm sufficient structure/algorithm is disclosed under § 112(f) as well — enablement and definiteness overlap heavily here.
- Consider prosecution history: did the applicant argue broad claim scope to overcome prior art while simultaneously relying on narrow disclosure to argue enablement? That tension is a red flag for later invalidity arguments.
Distinguishing the Two During Proofreading
It helps to keep a simple mental split while reviewing a draft:
| Written Description | Enablement | |
| Core question | Did the inventor possess the invention? | Can a POSITA make and use it? |
| Focus | Disclosure vs. claim scope | Disclosure vs. practical operability |
| Typical failure | Claim broader than what was described | Claim broader than what was taught how to make/use |
| Common trigger | Claim amendments, new matter, priority claims | Broad functional claims, unpredictable arts, thin examples |
A claim term can be perfectly described (the spec clearly conveys the inventor had it in hand) yet still not be enabled (the spec doesn’t explain how to make it work across its full scope) — and the reverse is also possible, though less common in practice.
A Practical Line-by-Line Workflow
- Build a claim chart first. Before touching prose, list every claim limitation in a spreadsheet with a column for the supporting specification passage(s) and a column for notes on adequacy.
- Read claims before reading the specification. Reviewers who read the spec first tend to unconsciously “fill in” support that isn’t really there because they now know what the invention is supposed to do.
- Flag every claim term with no anchor. If a limitation cannot be tied to specific specification language, it needs either new disclosure (before filing) or a claim amendment (after filing, watching for new matter issues).
- Interrogate breadth versus examples. Count the working examples and compare that count, and their variety, against how broadly the claims sweep. A wide gap is the single most common source of both written description and enablement rejections/invalidity findings.
- Trace every amendment to its source. During prosecution, maintain a running log mapping each amended claim limitation to the exact original specification language it derives from.
- Check numerical ranges and parameters for basis. Every number in a claim (percentages, ranges, dimensions, temperatures) should appear, or be clearly derivable, from the specification.
- Review functional and means-plus-function claims for structural support. Confirm actual structure, material, or algorithm is disclosed, not just the outcome.
- Have a second reviewer unfamiliar with the invention do a fresh pass. Drafters and even attorneys who worked on prior drafts often read intent into the document that a skeptical examiner or opposing expert will not.
- Document the review. A record showing this analysis was performed is itself valuable if enablement or written description is later challenged, since it demonstrates a reasoned basis for the claim scope chosen.
Common Pitfalls Worth Flagging Every Time
- Aspirational claim language (“substantially,” “generally,” “optimized”) used without a definition or example anchoring it.
- Prophetic examples treated as if they were working examples — prophetic (untested, “would be expected to”) examples can support enablement in some circumstances, but they carry less weight and should be clearly distinguishable from actual reduction-to-practice examples in the text.
- Claims added late in prosecution that recombine elements from different, unconnected embodiments in the original spec — this can create both new-matter and enablement problems if the combination itself was never disclosed or shown to work together.
- Genus claims in unpredictable arts supported by a single species or a small, structurally similar cluster.
- Incorporation by reference used to paper over gaps. Material incorporated by reference can support disclosure, but only if the incorporation is proper and specific; vague or blanket incorporation is a weak substitute for actual written support.
Way Forward
Written description and enablement proofreading is best done as a distinct pass, separate from clarity, antecedent basis, and formal § 112(b) definiteness review, because it requires a different question at every line: not “is this clear?” but “is this supported, and does it work?” Building a claim-to-specification support chart early, and revisiting it at every amendment, is the single most effective habit for catching § 112(a) problems before they become the basis for a rejection, an interference, or an invalidity finding in litigation