There is no such thing as a strong claim in the abstract. There is only a claim that is wide enough to be worth enforcing and narrow enough to withstand being attacked — and the window between those two conditions is often only a few words across.
01 / The Trade-Off
Two Failures, Pointing in Opposite Directions
A claim can fail in exactly two ways, and they are mirror images. Too wide, and it reads onto something that already existed: the granting office finds an earlier publication containing every element, the claim is refused for want of novelty, and years of filing costs buy an argument instead of a right. Too narrow, and it grants smoothly and protects almost nothing, because a competitor reads it, identifies the one element that was never essential, leaves that element out, and sells the product anyway.
The second failure is the more common and much the more expensive, because it is invisible. Nobody sends a letter explaining that your patent has been designed around. The application simply grants, the certificate arrives, the annual renewal fees are paid for a decade, and the day it is finally needed the wording turns out to describe a bracket nobody else uses. A refusal, by contrast, announces itself in writing and can still be argued about.
Deciding where to sit between those failures is a commercial judgement dressed as a drafting one. It depends on how crowded the technical field already is, how much money is available to fight an examination, and how many years of enforceable term the applicant can afford to spend arguing — the twenty-year clock runs from the filing date, so every year of dispute is a year subtracted from the end. It also depends on understanding what the resulting right actually consists of, which is a narrower thing than most first-time applicants assume; a plain description of what a patent does and does not grant its owner is worth reading before the drafting choices are made rather than after.
02 / Mechanism
Where Reach Actually Comes From
Scope is not a quality of a claim; it is an arithmetic consequence of four things, all of them controllable at the sentence level.
The first is element count. Every additional element narrows the claim, because an accused product must now contain one more thing before it falls inside. A four-element claim is broader than a seven-element claim describing the same invention, always, without exception. This is why the subtraction exercise done before drafting matters so much, and why the honest question about each element is not "is this in my product?" but "does the effect disappear without it?"
The second is the level of abstraction of each term. "A fastener" is broader than "a threaded fastener", which is broader than "an M4 stainless steel machine screw". Each step up the ladder of generality widens the claim, and each step is only available if the description supports it — a claim to fasteners generally, in a specification that discusses nothing but screws, invites an objection that the applicant has claimed more than they taught.
The third is functional versus structural language. Defining an element by what it does rather than what it is can be enormously broad, and is treated with corresponding suspicion. In several jurisdictions, wording of the form "means for retaining" is read as limited to the retaining structures actually described in the specification plus their close equivalents, so the apparent width evaporates on construction. Pure result language — "whereby the strap does not tangle" — adds no reach at all, because the claim is measured by its structural recitations, not by the outcome they are said to produce.
The fourth is the treatment of numbers. A claim reciting a single value covers that value. A claim reciting a range covers the range, and a well-built description states the widest workable band, a preferred narrower band, and at least one worked example inside it. That layering later provides room to retreat from 20–95 degrees to 40–60 degrees without adding new matter — a manoeuvre that is unavailable to anyone who wrote down only the temperature their prototype happened to run at.
- WiderFewer elements, more general terms, stated ranges, open transitional wording.
- NarrowerExtra elements, named materials, single values, closed transitional wording.
- NeutralStatements of purpose, advantages and intended results. They persuade, but they do not define.
Claim breadth is not a matter of confident writing. It is a count of how many things a competitor has to include before they are inside your fence.
The only definition that survives contact with an examiner
03 / Attrition
Four Ways Reach Gets Taken Away
Breadth is lost in four identifiable places, and only the first of them is obvious.
Prior art is the visible mechanism. An earlier publication is cited, the claim is narrowed to avoid it, and the boundary moves inward by the width of whatever limitation was added. Something is exchanged for a grant, and at least the price is stated on the face of the file.
Insufficient disclosure is the quiet one. A claim may not extend beyond what the specification actually enables a skilled reader to reproduce across the full width claimed. Claiming a family of compounds while describing one, or a range of operating pressures while testing at a single pressure, invites a refusal that no amount of argument can cure, because the missing material cannot be added after the filing date. The upper bound on any claim is therefore set by the description on day one, not by the ambition of the drafter in month eighteen.
Argument is the third and the most frequently overlooked. Statements made in writing to overcome an objection sit permanently on the public file. Explaining that the invention works because the membrane is rigid, in order to distinguish a flexible-membrane document, is a concession that will be quoted back by any competitor whose product uses something semi-rigid. In many territories such statements limit how the granted wording may later be read, whether or not the claim itself was ever amended.
The fourth is procedural drift: divided applications, unity objections that split one set into several, and features surrendered to expedite grant late in a long examination. None of these announces itself as a loss of scope, and cumulatively they account for a great deal of it. The choice of route contributes too, since the type of application filed governs which of these pressures apply and when they arrive — the range of application types and the proceedings attached to each is a scope decision made before any examiner is involved.
04 / Measurement
Testing a Claim Before Anyone Else Does
Two exercises, both cheap, expose most scope problems before filing.
The first is the design-around. Hand the claim to somebody competent and unsentimental and ask them to build a product that solves the same problem while falling outside the wording. If they succeed in an afternoon by substituting one element, the claim is too narrow in exactly that place, and the fix is to raise the level of abstraction of that element rather than to add adjectives. If they cannot avoid it without abandoning the effect altogether, the claim is doing its job.
The second is the adversarial read. Ignore what the claim was meant to say and take it entirely literally, then go looking for anything old that fits the description. Not something similar — something that satisfies every recited element. If a product sold thirty years ago answers the wording, the claim is too wide, and the correct response is a further element rather than a further qualifier. Adjectives narrow the claim without adding a limitation an examiner can be pointed at; they cost reach and buy nothing.
Both exercises expose the same recurring bias. Inventors instinctively pull claim breadth toward the version of the invention they find most elegant, which is rarely the version a competitor would copy. Expectation is part of the problem: audiences arrive with an idea of what a device ought to be able to do, and that ambient sense of the possible — traced in this look at how speculative fiction shapes what people expect a new device to do — is not the same as the structure a claim is permitted to cover.
05 / Judgement
Choosing a Width You Can Afford to Defend
The final decision is not about maximising anything. It is about direction. Scope should be widest along the axis a competitor would most plausibly move — the substitutable material, the alternative geometry, the equivalent step — and can be allowed to sit narrow along axes nobody would bother changing. A claim that is broad in an irrelevant direction is simply fragile for no return.
Two practical constraints keep the decision honest. Enforcement costs money the applicant may never have, so a right that only pays off after a multi-year dispute is worth less than one that is clear enough to make a competitor's own advisers recommend a licence. And breadth has to be provable across its whole width: a boundary that cannot be defended with the description already on file is not scope but exposure.
The workable answer is usually a set of graded positions rather than one heroic sentence — the widest claim the disclosure will carry at the top, with narrower rungs beneath it, each one a place to stand if the rung above gives way. That is the point of building a layered claim set rather than a single boundary line, and it starts from the same disciplined description that turns a rough concept into a defensible first claim.
Every word after the last necessary one is scope given away for free. Every word missing before it is a patent that never had any.
The whole of scope, in one line
Claim breadth is decided by counting, not by conviction — how many elements, how general each term, how wide each range, and how much of it the description on file can actually support.
End of report