If patent law had a favorite phrase, it might be: “Show your work.” Not in a high-school-math way, but in a very expensive, very federal, very don’t-lose-your-billion-dollar-claim kind of way. And when it comes to the written description requirement under 35 U.S.C. § 112(a), the Federal Circuit has been sending a message that could not be more plain: if you want broad patent protection, your specification has to clearly show that you actually invented what you later claim.
That is the real heartbeat of the doctrine. A patent is not a permission slip to claim whatever science may eventually wander into your neighborhood. It is a bargain. You tell the public what you invented in full, clear, concise, and exact terms, and in return, the law gives you a limited right to exclude others. When the written description is fuzzy, overly functional, or built like a maze with no exit signs, the Federal Circuit has shown a growing willingness to say, “Nice try, but no.”
This trend matters far beyond patent litigators and in-house counsel. It affects biotech startups racing to file early, pharmaceutical companies building life-cycle strategies, university tech-transfer offices polishing research portfolios, and patent prosecutors trying to turn scientific complexity into claim language that will survive contact with reality. The lesson is not that patents must be long for the sake of being long. The lesson is that they must be clear enough to prove possession of the claimed invention.
Why the Written Description Requirement Has Teeth
Section 112(a) says the specification must contain “a written description of the invention” and explain how to make and use it in “full, clear, concise, and exact terms.” That text does not exist for decoration. It is the statutory backbone of the written description requirement, and the Federal Circuit has repeatedly treated it as a gatekeeper rather than a polite suggestion.
The foundational modern case is Ariad Pharmaceuticals, Inc. v. Eli Lilly & Co., the Federal Circuit’s 2010 en banc decision. That case settled a major fight by confirming that written description is a requirement separate from enablement. In plain English, it is not enough for a patent to teach skilled people how to make and use something. The patent must also show that the inventor was actually in possession of the claimed subject matter as of the filing date.
That distinction matters because enablement and written description often travel together, but they are not twins. They are more like siblings who borrow each other’s clothes and then argue about who owns the jacket. A patent may teach a lot and still fail to clearly describe the specific thing later claimed. That is why the Federal Circuit has kept asking a basic but brutal question: does the specification reasonably convey to a skilled artisan that the inventor truly had the invention in hand?
What “Clear” Really Means in Federal Circuit Case Law
Clear written description does not always mean listing every possible embodiment under the sun. The Federal Circuit has never imposed a single magic formula. Instead, the court looks at the nature of the technology, the scope of the claims, and whether the field is predictable or highly unpredictable. A mechanical invention may survive with one level of detail; a biotech or chemical genus claim often needs much more.
For genus claims, the court usually wants one of two things: a representative number of species or common structural features that let a skilled artisan recognize which members belong in the claimed genus. That sounds manageable until litigation begins and somebody asks whether your “representative” examples are actually representative, or whether your “common features” are just a pile of hopeful adjectives wearing a lab coat.
The court also talks about “blaze marks,” one of patent law’s more memorable metaphors. The idea is simple: if your disclosure is a forest of possibilities, the specification needs marks on the trail leading a skilled artisan to the specific tree, subgenus, or path you later claim. Without those markers, the patent starts to look less like disclosure and more like a scavenger hunt with missing instructions.
Recent Federal Circuit Cases Show a Pattern
Ariad: Possession, Not Wishful Thinking
In Ariad, the court made clear that broad functional claiming without adequate support is dangerous territory. The patent at issue tried to claim methods of reducing NF-κB activity, but the specification did not sufficiently describe how to achieve the claimed result across the full scope. The decision is now the doctrinal anchor for the idea that a patent cannot simply announce a scientific goal and call it an invention. Patent law rewards disclosure, not ambition dressed up as certainty.
Juno: Two Examples Do Not Equal the Universe
The 2021 decision in Juno Therapeutics v. Kite Pharma pushed that logic hard in the antibody space. The claims covered CAR-T technology using any single-chain antibody fragment that bound any selected target of clinical interest. The problem was scope. The patent disclosed only a couple of examples, while the claims reached an enormous functional genus. The Federal Circuit held that the written description failed because the patent did not provide a representative sample of species or defining characteristics that would identify which fragments across that broad genus would actually work.
That case became a warning siren for functional genus claims in biotech. Saying “anything that binds the target” may sound elegant in a claim chart, but elegance does not equal possession.
Biogen: One Number in a Range Is Not Always Enough
Also in 2021, Biogen v. Mylan showed that written description problems are not limited to broad antibody claims. There, the dispute focused on a 480 mg/day dosage of dimethyl fumarate for multiple sclerosis. The specification mentioned ranges and scattered dosage information, but the Federal Circuit concluded that the patent did not clearly show possession of that specific claimed regimen as of the filing date. The court was not persuaded by post hoc arguments that a skilled artisan would have been drawn to the claimed dose.
That decision offered a painful reminder: a number appearing somewhere in a disclosure is not automatically the same thing as a clear description of a claimed invention. Patent drafting is not a game of accidental bingo.
Regents, PureCircle, Duke, and Seagen: Follow the Trail or Lose the Claim
In the years that followed, the Federal Circuit kept repeating the same core message with different fact patterns. In Regents of the University of Minnesota v. Gilead Sciences, the court emphasized that optional choices without meaningful direction do not establish possession. In a memorable Yogi Berra-inspired line, the opinion mocked the indeterminacy of a disclosure that effectively told readers to take every fork in the road at once.
In PureCircle v. SweeGen, the court again rejected a genus claim that described an enzyme largely by what it did, rather than by a representative number of species or clear structural features. In Duke University v. Sandoz, the court found that a patent describing billions of compound possibilities did not provide adequate blaze marks to the claimed subgenus. And in Seagen v. Daiichi Sankyo, decided in December 2025, the court reinforced that a broad disclosure covering millions of possibilities does not automatically support a later claim to a narrow subgenus or species. A giant menu is not the same thing as ordering dinner.
Taken together, these decisions reveal a consistent approach. The court is suspicious of patents that begin with a broad universe, point vaguely toward a useful corner of it, and then later claim that corner as if it had been clearly identified all along.
The 2026 Twist: Teva v. Lilly Shows Clarity Is Contextual
Now for the important nuance. The Federal Circuit is not saying that every genus claim is doomed unless the patent reads like a phone book for molecules. In April 2026, in Teva Pharmaceuticals International GmbH v. Eli Lilly and Co., the court reversed a judgment as a matter of law that had invalidated Teva’s claims for lack of written description.
That case involved method claims using humanized anti-CGRP antagonist antibodies to treat headache. Lilly argued that the specification disclosed only one humanized antibody and therefore lacked a representative number of species. The Federal Circuit disagreed that the story was so simple. It emphasized context: murine anti-CGRP antibodies were disclosed, humanization was routine by the priority date, anti-CGRP antagonist antibodies were well known in the art, and a skilled artisan could understand from the specification that all humanized anti-CGRP antagonist antibodies would treat headache.
In other words, the court did not abandon the written description requirement. It applied it with attention to what the invention actually was and how much background knowledge skilled artisans already had. That matters. Teva suggests that when a claimed genus is well understood in the art and not itself the true point of novelty, background knowledge can help fill in the picture. Clear written description, then, is not just about quantity. It is about whether the disclosure, read in context, really communicates possession.
That is why the article’s title works on two levels. Yes, written descriptions must be clear. But “clear” in Federal Circuit jurisprudence does not always mean “endlessly detailed.” It means coherent, directed, and credible to a skilled artisan.
What Patent Drafters Should Learn From This Trend
1. Draft for the Full Scope You Actually Want
If the business goal is a broad genus claim, the specification should be built for that goal from day one. That means more examples, better structural explanations, and fallback positions that can survive if the broadest claim later collapses.
2. Functional Language Needs Real Support
Functional claiming is not forbidden. But if a claim defines something mainly by what it does, the specification should show which things do that job and why. Otherwise, the court may treat the claim as an attempt to monopolize results rather than describe an invention.
3. Do Not Rely on a “Kitchen Sink” Disclosure
Throwing hundreds of options into a specification can feel safe, but without prioritization, examples, or signposts, it may create the opposite effect. The Federal Circuit often treats sprawling disclosures as evidence that the inventors had not yet singled out the later-claimed invention.
4. Use the Science of the Field Wisely
Teva and other cases show that background knowledge can matter. But it is risky to assume the prior art will rescue a thin disclosure. If the field is unpredictable, or if the claimed subgenus is the real invention, the patent should say more, not less.
5. Litigation Will Read the Spec Like a Crime Scene
Years after filing, every sentence may be examined to see whether it points toward the claim or away from it. Preferred embodiments, examples, ranges, and even passing statements can become decisive. Patent prosecutors draft for today; litigators live with those words for a decade.
Why This Doctrine Is Good for the Patent System
There is a policy reason behind all of this. Written description keeps the patent system honest. It stops applicants from claiming more than they actually invented and prevents hindsight from turning a rough research idea into a broad legal monopoly. Especially in biotechnology and chemistry, where fields can be unpredictable and commercially enormous, the doctrine protects the public from patents that promise the moon after discovering a flashlight.
At the same time, the Federal Circuit’s more nuanced cases show that the rule is not anti-inventor. It is anti-vagueness. When a patent clearly communicates possession, even without cataloging every species in existence, the court can and does uphold the claims. The key is whether the disclosure points with confidence to the claimed invention instead of waving generally at an entire scientific continent.
Experiences From the Real World of Patent Drafting and Litigation
Anyone who has worked near patent drafting in life sciences knows the emotional arc of this issue. It usually starts with excitement. A research team has a promising result, a drug candidate looks good, an antibody binds beautifully, or a platform finally behaves in a way that makes everyone in the room sit up straighter. The company wants to file quickly, because timing matters. Investors are watching. Competitors exist. Someone says, “Let’s claim this broadly.” That sentence is both understandable and dangerous.
In practice, the hardest conversations often happen between scientists and patent counsel during those early drafting sessions. Scientists naturally think in research possibilities: if this molecule works, maybe related ones will too; if one antibody hits the target, perhaps humanized versions should follow; if one structure performs well, maybe an entire family deserves coverage. Patent counsel, meanwhile, has to ask the annoying but vital questions: Which variants have actually been made? Which sequences are known? Which structures are supported? What data exists? What can a skilled artisan reasonably infer, and what would still require guesswork?
That gap between scientific optimism and legal proof is where written description fights are born. Many inventors experience real frustration when they learn that a broad concept is not the same thing as a clearly supported claim. They know the science is heading in a certain direction, and often they are right. But patent law is not grading the lab’s future intuition. It is grading what the application actually tells the public on the filing date.
Litigation makes this even more vivid. A patent that looked broad and elegant during prosecution can look fragile under cross-examination. Opposing counsel will put the claim on one screen and the specification on another and ask, with surgical calm, “Where is this exact thing described?” If the answer requires six inferential leaps, three expert declarations, and a small prayer circle, the written description problem becomes painfully visible. By contrast, when the specification includes concrete examples, explains why they are representative, identifies common structural features, and provides sensible fallback positions, the patent feels sturdier from the inside out.
There is also a portfolio-management lesson here. The best patent teams do not treat the specification as a rushed container for claims. They treat it as the long-term evidence file for possession. They preserve sequences, structures, alternatives, reasons for selection, preferred embodiments, and technical explanations that can matter years later. That work is rarely glamorous. It does not make headlines. But when a case reaches the Federal Circuit, those choices often determine whether the patent reads like a genuine invention or an ambitious draft trying to outrun the science.
Final Takeaway
The Federal Circuit’s message is now impossible to miss: written descriptions must be clear because patents are supposed to reward what inventors actually disclosed, not what they later wished they had claimed more carefully. The strongest patent specifications do not merely recite options. They guide, identify, and prove possession. And in today’s case law, that difference can be worth everything.
Note: This article is for general informational purposes only and is not legal advice.
