Almost every application of any commercial interest is rejected at least once. The exchange that follows is the ordinary mechanism by which scope gets settled, and it is conducted almost entirely in writing, on a public file, in a register that will be read by opponents for the following two decades.
01 / The Trigger
What an Examination Report Actually Demands
A substantive report arrives typically somewhere between eighteen and thirty-six months after filing, later in congested technical fields. It cites documents, usually between one and five, and works through the claims in order: this claim is anticipated by document D1; that claim would be obvious from D1 combined with D2; these claims are not objected to on novelty but are unclear or unsupported. Formal objections about antecedents, reference numerals and claim dependency usually sit at the end and are the cheapest part to fix.
The response period is commonly two to four months from the report date, extendable in most systems on payment of a fee, sometimes twice. That is the entire window in which a strategy has to be chosen. Two facts shape it. First, the twenty-year term runs from the filing date, so a leisurely three-round exchange costs enforceable years at the far end of the patent's life. Second, and more importantly, the room available for manoeuvre was fixed at filing and cannot be enlarged now.
The single most useful habit at this stage is to read the report as an offer rather than a verdict. An examiner who lists which claims are unobjectionable has told the applicant exactly where the allowable boundary lies. Claim amendments built from that list are far more likely to be accepted first time than amendments invented independently of it.
02 / The Rule
Nothing New May Be Added
One constraint governs every response: no matter may be introduced that was not disclosed in the application as filed. Claims can be rewritten, split, combined, renumbered and narrowed freely, but every word of the new version must find support in the document already on file. Material added afterwards is new matter and is refused, and if it slips through it can leave the granted claim vulnerable for the rest of its life.
In practice this makes amendment an exercise in mining rather than writing. The safest change lifts wording verbatim from an existing dependent claim. The next safest quotes a sentence from the description. Less safe, and treated with real severity in some jurisdictions, is the intermediate generalisation: taking one feature out of a paragraph that described it only in combination with two others, and claiming it alone. The description said A with B and C; the amendment claims A; the objection is that A on its own was never disclosed.
This is why a specification that records tolerance bands, alternative materials, rejected variants and the reasons for each is worth so much more than a clean account of one working prototype. Every one of those disclosures is a lawful amendment waiting to be used. A description of a single embodiment in a single material offers precisely one response, and it is usually a bad one. The habit of writing down the alternatives while the project is still messy is visible in development histories such as the account of how a kitchen mixing device moved from sketch to product, where the variants that were abandoned early are the ones that later had somewhere to be used.
An amendment cannot invent anything. It can only spend what the specification already contains.
The constraint behind every response
03 / Instruments
Six Moves, in Rough Order of Cost
There is more than one way to answer a report, and the cheapest options are routinely skipped in favour of the most expensive. Claim amendments are only one instrument among six.
The first is to argue without amending at all. Where the examiner has plainly misread a cited document, or has combined two documents a skilled reader would have no reason to combine, a reasoned reply that changes no wording preserves the full scope. It costs a round of correspondence and nothing else, and it is the only move that gives up nothing — though every sentence of it stays on the file.
The second is to fold a dependent limitation upward into the main claim, then cancel the now-redundant dependent. This is the workhorse. It is safe on new matter, it is usually persuasive because the examiner has already conceded that claim, and the price is known in advance.
The third is to draw a feature from the description into the claim, which reaches further than the dependents allow but exposes the applicant to a support objection. The fourth is to divide: keep the narrower position in the current application to secure a grant, and pursue the wider subject matter in a divisional filed before the parent grants — a deadline that is easy to miss and impossible to reopen. The fifth is a request for an interview or telephone consultation, which in many offices resolves in twenty minutes an argument that would otherwise consume two written rounds and four months. The sixth, and the most permanent, is to accept a narrowing an examiner has proposed simply to close the file.
- CheapReasoned argument, formal corrections, an interview to clarify a misread citation.
- PricedImporting a dependent limitation, or a feature from the description with support in hand.
- CostlyDivisional filings, unexplained concessions, and any narrowing accepted only to end the correspondence.
04 / The Bill
What the File Remembers
The hidden cost of prosecution is not the fees. It is that the written record becomes part of how the granted claim is read. In many territories, a limitation added to overcome prior art restricts the applicant's ability later to argue that a competitor's near-equivalent variation should be treated as covered. Having narrowed to a rigid membrane to escape a citation, the owner cannot easily complain about a semi-rigid one.
Arguments carry the same freight even when no wording changes. A submission asserting that the invention works because the sensor samples at high frequency will be quoted back by anyone whose product samples slowly, whether or not sampling rate ever entered the claim. The discipline that follows is narrow writing: address the cited document specifically, say what distinguishes the claim from it, and stop. Volunteering a general theory of why the invention is clever is how applicants surrender scope they were never asked to give up.
Claim amendments should therefore be planned as a sequence rather than improvised one report at a time. If three narrowings are conceivably available, take the one that costs least commercially first, and keep the others in reserve for the second report. Applicants who concede everything in round one have nothing left to trade in round two, which is precisely when it is needed. Long technical development cycles make this obvious in retrospect — fields where testing and approval run for years, of the kind described in coverage of inventor-led work on environmental technology, routinely see three or four exchanges before allowance.
05 / Sequence
How the Exchange Usually Ends
The common pattern is short. A first report objects to the main claim and concedes three dependents. The response imports the least commercially damaging of the three, corrects the formal objections, and argues briefly against a second citation. A further report either allows the case or raises one residual clarity objection, which is resolved in a page. Grant follows some months later, with publication of the granted text and the start of periodic renewal fees that escalate over the remaining term.
What separates that from a case that stalls is rarely the strength of the invention. It is whether the original specification left anywhere to retreat to, and whether the applicant kept a ledger of what had already been given away. Each concession should be recorded with the reason for it, because in the third round nobody remembers why the word "annular" was inserted in the first, and the answer determines whether it can be argued around later.
Two structural points are worth holding on to at the end. Narrowing is not defeat: a narrower granted claim that reads cleanly onto a competitor's product is worth more than a wide claim that never grants. And the layered set drafted at the start is what makes any of this navigable, which is why the architecture of a claim set built with fallback positions and a clear-eyed view of how much reach a disclosure can actually support matter more than anything an amendment can achieve. Both begin with the disciplined drafting that precedes filing. Independent inventors working through their first examination tend to underestimate how much of it is procedure rather than argument — a point made repeatedly in this profile of decades spent alongside first-time applicants, and echoed in an account of inventor-led design practice in the built environment, where the same slow, documented iteration decides what can eventually be protected.
Give up the smallest thing that will get the case allowed, say only what the objection requires, and write down what it cost.
Three rules for every response
Prosecution is a negotiation conducted in permanent ink. The applications that come out of it with something worth owning are the ones that entered with fallbacks already written and spent them one at a time.
End of report