Patent Claim Drafting

“comprising” (open)

Open-ended: the claim covers a device/method that has the listed elements AND possibly more. An accused product with the listed elements plus extra features still infringes.

Broadest, most common transitional phrase. 'A widget comprising A, B, and C' is infringed by a widget with A, B, C, and D.

“consisting of” (closed)

Closed: the claim covers ONLY the listed elements and excludes anything additional. Adding an unlisted element avoids infringement.

Narrowest. Used when you must exclude additional components — common in chemistry to avoid prior art. 'Consisting of A, B, C' is NOT infringed by A, B, C, D.

“consisting essentially of” (partly closed)

Middle ground: covers the listed elements plus any additional elements that do NOT materially affect the basic and novel characteristics of the invention.

Mostly used in chemical/pharma claims. Litigation often turns on whether an added ingredient 'materially affects' the invention.

Antecedent basis — 'a' then 'the'

The first time an element appears, introduce it with 'a' or 'an' ('a processor'). Every later reference uses 'the' or 'said' ('the processor') to point back to that same element. Referring to 'the processor' before any 'a processor' was introduced creates a lack-of-antecedent-basis problem that can render a claim indefinite under § 112(b). Precise antecedent basis is the grammar of valid claims.

The claim-breadth ladder

Good claim sets cascade from broad to narrow. The broadest independent claim captures the core inventive concept with the fewest limitations (maximizing coverage). Dependent claims add limitations, creating progressively narrower fallback positions that survive even if the broad claim is invalidated over prior art. You want the broadest claim that is still valid — and a ladder of narrower claims beneath it as insurance.

Every limitation is a way out for an infringer

Under the all-elements rule, an accused product infringes only if it has EVERY element of a claim. Each word you add to a claim is one more requirement the infringer's product must meet — and one more way they can design around it. Drafters fight to keep independent claims lean; unnecessary limitations narrow scope for no benefit.

The means-plus-function trap — § 112(f)

Writing a claim element as 'means for [performing a function]' (or sometimes 'module/unit/mechanism for...') invokes 35 U.S.C. § 112(f): the element is construed to cover ONLY the specific structure disclosed in the specification for that function, plus equivalents — far narrower than the functional language suggests. If the spec discloses no corresponding structure (a common problem for software 'means'), the claim can be invalid as indefinite. Modern drafters usually avoid 'means' language unless they specifically want § 112(f) treatment and have disclosed the structure.

Prosecution narrows what you wrote

The claims that issue are rarely the claims you filed. Amendments and arguments made to overcome prior art during prosecution narrow the claims and create prosecution history estoppel — surrendered scope cannot be recaptured through the doctrine of equivalents (Festo). Draft and argue with the future infringement case in mind: every characterization of the claims becomes part of the permanent record courts will read.

Support everything in the specification

A claim is only as good as the specification behind it. Under § 112(a), claims must be supported by an adequate written description showing the inventor possessed the full claimed scope, and the spec must enable a skilled person to make and use that scope. Claims broader than the disclosure are vulnerable to written-description and enablement attacks (Amgen v. Sanofi). Draft the spec to support claims broader than you initially file, so you have room to broaden and to add claims in continuations.

Why does 'comprising' matter so much in a patent claim?

Because the transitional phrase that connects a claim's preamble to its elements determines whether the claim is 'open' or 'closed' — and that controls what infringes. 'Comprising' (and synonyms like 'including' or 'containing') is OPEN: a claim reciting 'a device comprising A, B, and C' is infringed by any device that contains A, B, and C, regardless of whether it also contains additional elements D, E, or F. This makes 'comprising' the broadest and most common transitional phrase, because an infringer cannot escape simply by adding features. By contrast, 'consisting of' is CLOSED: a claim reciting 'a composition consisting of A, B, and C' covers ONLY a composition of exactly those elements — adding an unlisted element D avoids infringement. And 'consisting essentially of' is a middle ground covering the listed elements plus anything that does not materially affect the basic and novel characteristics of the invention. The practical consequence is enormous: choosing 'consisting of' when you meant 'comprising' can hand infringers an easy design-around (just add a component), while choosing 'comprising' when prior art requires a closed claim can render the claim invalid. The choice is usually deliberate — 'comprising' for breadth in most mechanical, electrical, and software claims; 'consisting of' or 'consisting essentially of' in chemistry and pharma where you need to exclude additional ingredients to distinguish prior art. A single transitional word can decide an infringement case, which is why it is one of the first things a patent litigator checks.

What is antecedent basis in patent claims?

Antecedent basis is the requirement that every claim term referred to with a definite article ('the' or 'said') must have been previously introduced in the claim with an indefinite article ('a' or 'an') or otherwise clearly established. It is the grammar that makes claims definite and traceable. The rule in practice: the FIRST time you introduce an element, use 'a' or 'an' — 'a housing,' 'a first sensor,' 'a controller.' Every SUBSEQUENT reference to that same element uses 'the' or 'said' — 'the housing,' 'the first sensor,' 'said controller' — to point unambiguously back to the element already introduced. Problems arise when: (1) A claim refers to 'the housing' without ever having introduced 'a housing' — a lack of antecedent basis, which can render the claim indefinite under 35 U.S.C. § 112(b) because it is unclear what 'the housing' refers to. (2) A claim introduces 'a sensor' twice, creating ambiguity about whether there are one or two sensors. (3) A dependent claim refers to an element with 'the' that exists only in a different, non-referenced claim. Why it matters beyond grammar: indefiniteness under the Nautilus standard (claims must inform a skilled person of the scope with reasonable certainty) can invalidate a claim, and antecedent-basis errors are a classic, avoidable source of indefiniteness. Examiners frequently issue § 112(b) rejections for antecedent-basis problems, and careful drafters track every element's introduction and reference. Note that perfect antecedent basis is not always strictly required if the meaning is reasonably clear, but disciplined antecedent-basis practice is the safest way to avoid indefiniteness disputes in both prosecution and litigation.

How broad should a patent claim be?

As broad as possible while remaining valid — which is a genuine tension, not a platitude. The competing forces: (1) Broader is better for coverage. A claim with fewer limitations covers more products and is harder to design around, because under the all-elements rule an accused product must contain EVERY limitation to infringe. Fewer limitations means fewer escape routes for infringers. So drafters strip independent claims down to the essential inventive concept and resist adding unnecessary limitations. (2) Broader is more vulnerable to invalidity. A broad claim is more likely to read on prior art (anticipation under § 102 or obviousness under § 103) and more likely to exceed what the specification supports (written description and enablement under § 112(a), as in Amgen v. Sanofi where broad functional genus claims failed for lack of enablement). A claim broad enough to cover everything is often broad enough to be invalid. The resolution is the claim-breadth ladder: file the broadest independent claim you believe is valid over the known prior art, then a series of progressively narrower dependent claims that add limitations one at a time. The broad claim maximizes coverage if it survives; the narrower dependent claims are fallback positions that remain valid (and enforceable) even if the broad claim is invalidated. You also draft multiple independent claims of different scope and in different statutory categories (e.g., a method claim and a system claim) to capture infringers who might practice only part of the invention. The art of claim drafting is finding the broadest defensible scope and building insurance beneath it — and leaving room in the specification to pursue even broader or differently-focused claims in continuations as the commercial picture and prior art landscape clarify.

What is the means-plus-function trap in claim drafting?

The means-plus-function 'trap' is the way 35 U.S.C. § 112(f) can silently narrow — or invalidate — a claim element written in functional terms. Section 112(f) permits an element to be expressed as 'a means [or step] for performing a specified function' without reciting the structure that performs it. But the trade-off is severe: such an element is construed to cover ONLY the specific structure (or material or acts) disclosed in the specification for performing that function, plus equivalents thereof. So instead of covering any means of achieving the function (as the words might suggest), the claim is limited to what you actually described in the spec. Two traps follow: (1) Unexpected narrowing. Drafters who use 'means for [function]' thinking it broadly captures all ways of performing the function get the opposite — narrow coverage tied to their disclosed embodiment. (2) Invalidity for lack of structure. If the specification does not disclose ADEQUATE corresponding structure for the claimed function, a § 112(f) element is indefinite and the claim is invalid. This is an acute problem for software/computer-implemented 'means,' where courts have required the spec to disclose an algorithm (not just a general-purpose computer) as the corresponding structure; a 'means for calculating X' with no disclosed algorithm fails. The modern complication: § 112(f) is triggered by functional claiming even WITHOUT the word 'means' — courts (Williamson v. Citrix, Fed. Cir. 2015 en banc) held that nonce words like 'module,' 'mechanism,' 'unit,' or 'device for [function]' can invoke § 112(f) if they fail to connote sufficient structure. Practical drafting response: most practitioners AVOID means-plus-function and functional 'nonce word' claiming unless they specifically want § 112(f) treatment, instead reciting actual structure; and where functional claiming is used, they ensure the specification discloses concrete corresponding structure (including algorithms for software functions). The trap catches drafters who treat functional language as a shortcut to breadth when it is actually a path to narrowness or indefiniteness.

Should I draft my own patent claims?

Understanding claim drafting is valuable for any inventor or founder, but drafting the claims of a patent you intend to rely on is one of the highest-stakes, most specialized tasks in patent practice — and it is where professional drafting earns its cost. Why claims are uniquely unforgiving: (1) Claims define the entire legal scope of the patent; everything else (the specification, drawings) exists to support them. A poorly drafted claim can be too narrow to stop competitors, too broad to be valid, or indefinite enough to be unenforceable — and these defects are often not fixable after issuance. (2) The drafting choices interact in non-obvious ways: transitional phrases set openness; antecedent basis governs definiteness; functional language can trigger § 112(f); claim breadth trades coverage against validity; and each word becomes a limitation an infringer can design around or a hook for an invalidity attack. (3) Claims must anticipate future events the drafter cannot see — how competitors will try to design around them, what prior art will surface, how a court years later will construe each term under the Phillips framework. Experienced patent practitioners draft with the eventual litigation in mind. What inventors can productively do themselves: understand the concepts (so you can work effectively with counsel and evaluate the claims you're paying for); draft a thorough invention disclosure capturing every variation, alternative, and use of the invention (which gives the attorney the raw material to draft broad, well-supported claims); and review draft claims critically, asking whether they cover the commercially important embodiments and the obvious design-arounds. For a provisional application used purely to secure a priority date, some inventors draft their own (a provisional needs no formal claims), but the non-provisional claims that determine enforceable scope are where most inventors are well served by a registered patent attorney or agent. The cost of professional claim drafting is small relative to the cost of discovering, during litigation or a financing, that the claims don't actually protect the business.