Claim construction decides most patent cases, and it happens long before trial.
The court determines what the claim words mean, as a matter of law. Not the jury, and not the parties' experts.
Everything downstream follows from it. Whether the accused product falls inside the claim, and whether the prior art does, are both answered by the same construction.
Which is why cases settle in waves after the ruling. The uncertainty both sides were pricing disappears in a single order.
Why it is the pivot
| If a term is construed | Infringement | Validity |
|---|---|---|
| Broadly | More likely | Less likely — captures more prior art |
| Narrowly | Less likely | More likely — avoids prior art |
The two move in opposite directions, which is why parties do not simply argue for the broadest or narrowest reading available.
A patent holder pushing for a construction broad enough to capture the accused product may capture an anticipating reference at the same time. That tension disciplines the argument on both sides.
The Phillips hierarchy
*From Phillips v. AWH Corp., the framework courts apply.*
| Priority | Source | Weight |
|---|---|---|
| 1 | The claims themselves | Highest — including other claims |
| 2 | The specification | Very high — "the single best guide" |
| 3 | The prosecution history | High — what was said to get the patent |
| 4 | Extrinsic evidence | Lowest — dictionaries, treatises, experts |
Intrinsic evidence dominates. The claims, specification and file history are public, were created by the applicant, and reflect what was actually said — which is why courts prefer them to expert testimony created for litigation.
Claim differentiation is a useful tool within priority one. If a dependent claim adds a limitation, the independent claim presumptively does not already contain it.
Extrinsic evidence is admissible but rarely decisive. A dictionary definition that contradicts the specification loses.
The starting point: ordinary meaning
A claim term carries its ordinary and customary meaning to a person of ordinary skill in the art, at the time of the invention.
| Element | Why it matters |
|---|---|
| Ordinary meaning | The default; departures need justification |
| To a skilled person | Not to a layperson or a judge |
| In the art | Field-specific usage governs |
| At the time of the invention | Not today's usage |
"At the time of the invention" catches people out in fast-moving fields. A term that means something specific today may have meant something broader in 2011.
Two exceptions displace ordinary meaning: a clear definition given by the applicant acting as their own lexicographer, and a clear disavowal of scope.
The specification tension
| Position | Argument |
|---|---|
| Read the claim in light of the specification | Required by Phillips |
| Do not import limitations from embodiments | Also required |
These sit in permanent tension, and it is the most litigated aspect of construction. A specification describing only one embodiment informs the meaning of the claim terms without necessarily limiting the claim to that embodiment.
Which is a drafting lesson. A specification describing several alternatives supports a broader construction; one describing a single implementation invites an argument that the claim is limited to it. See patent application process.
Consistent terminology matters here too. Using two words for one component invites both indefiniteness and a narrowing construction.
Prosecution history and disclaimer
| What the applicant did | Effect on construction |
|---|---|
| Argued a term means X to overcome prior art | Generally bound to X |
| Amended to add a limitation | Cannot read the claim without it |
| Distinguished a reference on a basis | Bound by that basis |
| Accepted the examiner's characterisation | Bound |
| Said nothing | No disclaimer |
Claim scope disclaimer is construction; prosecution history estoppel is the related limit on equivalents. They are distinct doctrines that arise from the same record. See prosecution history estoppel.
The file wrapper is the first thing opposing counsel reads, precisely because this is where scope was given away.
Choosing which terms to dispute
Courts frequently limit the number of terms, commonly to around ten, so selection is strategic.
| Dispute a term when | Skip it when |
|---|---|
| Construction decides infringement | Both constructions produce the same result |
| Construction decides validity | It is peripheral |
| The other side needs a strained reading | You are comfortable either way |
| Indefiniteness is arguable | The term is plainly clear |
Disputing everything wastes the allocation and dilutes the terms that matter.
A term where both constructions lead to the same outcome is not worth briefing, however interesting the linguistic question.
Means-plus-function terms
| Feature | Detail |
|---|---|
| Statutory basis | §112(f) |
| Triggered by | "Means for" language, or functional language without structure |
| Construed as | The structure disclosed in the specification, plus equivalents |
| Risk | Indefinite if no structure is disclosed |
| Common in | Software claims using functional language |
The presumption is rebuttable in both directions. Using "means for" presumes §112(f) applies; avoiding it does not guarantee escape if the claim recites function without structure.
Software claims are the frequent casualty. A claim reciting a "module configured to" perform a function, with no corresponding algorithm in the specification, is a candidate for indefiniteness.
Indefiniteness
A claim is invalid if it fails to inform a skilled person about scope with reasonable certainty.
| Common indefiniteness triggers | Why |
|---|---|
| Terms of degree with no standard | "Substantially", "about", with nothing to measure against |
| Subjective terms | "Aesthetically pleasing", "user-friendly" |
| Means-plus-function with no structure | Nothing to construe |
| Inconsistent antecedent basis | Unclear which element is referenced |
| Contradictory specification | Reader cannot determine boundaries |
Terms of degree are not automatically indefinite. "Substantially flat" may be definite if the specification provides a way to measure it, and indefinite if it does not.
Worked example: one term, two outcomes
Claim 1 recites "a bypass channel positioned to equalise pressure across the poppet during transition."
| Disputed term | "during transition" |
|---|---|
| Patent holder's construction | "At any point while the poppet moves" |
| Accused party's construction | "Only while the poppet is between fully open and fully closed" |
What the evidence shows
| Source | Says |
|---|---|
| Other claims | Claim 4 recites "throughout the poppet's travel" — claim differentiation |
| Specification | Describes equalisation "as the poppet leaves its seat" |
| Prosecution history | Applicant distinguished prior art on the basis that it equalised continuously |
| Extrinsic | Expert testimony both ways |
The result
| Construction adopted | The narrower one |
| Reason | Prosecution history disclaimer — continuous equalisation was disavowed |
| Effect on infringement | Accused product equalises continuously — outside the claim |
| Effect on the case | Summary judgment of non-infringement |
The case was decided by something the applicant said in 2016 to get the patent allowed. Neither the product nor the claim language changed.
Claim differentiation supported the same reading, since claim 4 expressly covered the broader behaviour the holder was now arguing claim 1 already included.
Timing and procedure
| Step | Typical timing |
|---|---|
| Contentions exchanged | Months 4–8 |
| Terms identified and exchanged | Months 10–14 |
| Joint claim construction statement | Months 12–16 |
| Briefing | Months 12–18 |
| Markman hearing | Months 15–20 |
| Construction order | Shortly after |
| Settlement wave | Immediately after |
Local patent rules drive the schedule, and they vary considerably. Some districts front-load construction deliberately to resolve cases earlier.
Budget to construction, not to trial. That is where most cases end, and it is roughly forty per cent of the cost of a full trial. See patent litigation.
Appeal
| Basis | Standard of review |
|---|---|
| Intrinsic evidence | De novo |
| Subsidiary factual findings on extrinsic evidence | Clear error |
De novo review of intrinsic construction matters. The Federal Circuit reconsiders the pivotal ruling from scratch on the claims, specification and file history, so a case decided on construction is genuinely open on appeal.
Which affects settlement value. A district court win on construction is not final, and both sides price that.
The same standard at the PTAB
| Before 2018 | Since 2018 | |
|---|---|---|
| District court | Phillips | Phillips |
| PTAB post-grant trials | Broadest reasonable interpretation | Phillips |
| Examination | BRI | BRI |
Alignment removed a real inconsistency. The same claim could previously be read broadly at the Board and narrowly in court, which produced outcomes that were hard to reconcile.
Examination still uses BRI, deliberately, because an applicant can amend in response and a patent owner after grant cannot. See PTAB.
Claim construction: the checklist
- Read the file wrapper first. Disclaimer during prosecution is where most constructions are decided.
- Start from ordinary meaning to a skilled person, at the time of the invention.
- Check claim differentiation. A limitation in a dependent claim presumptively is not in the independent one.
- Look for lexicography and disavowal in the specification, the two things that displace ordinary meaning.
- Do not import embodiment limitations into claims that do not recite them — and expect the other side to try.
- Select disputed terms strategically. Courts limit the number, so brief only terms that change an outcome.
- Check for means-plus-function triggers, including functional language without "means for".
- Raise indefiniteness where terms of degree lack a measurable standard.
- Budget to the Markman hearing, since most cases end shortly after it.
- Remember intrinsic construction is reviewed de novo, so a construction loss is not necessarily final.