The Signed Declaration That Contradicted Its Own Free Text

2026-09-16 · Philip Choo · AI9OS

A client returns a provenance declaration. At the top, a tick-box certifies that everything listed was obtained lawfully. Below it, a free-text table describes each document and where it came from, and several of those descriptions record material obtained without the subject's knowledge.

One sheet. One signature. One date. The two halves of the page say opposite things, and nobody noticed for weeks.

Why a signed form is the least-read document in a file

Everything else that arrives gets interrogated. A bank statement is checked against the account it claims to describe. A photograph gets its metadata looked at. A witness account is tested against the timeline.

A signed declaration gets filed.

That is not laziness. It is what a signature is for. The document exists to convert a set of assertions into something settled, and once it is executed it reads as the answer rather than as more evidence to weigh. The tick is treated as the meaning of the page and the handwriting underneath it is treated as detail.

There is a second reason, less flattering. The declaration is usually the last administrative step before the interesting work starts. It is processed by someone who wants to get to the interesting work.

The contradiction is not the problem. The discovery route is

A client who fills in a form inconsistently has done something ordinary. They may not understand what "lawfully obtained" means as a term of art. They may have read it as a question about whether they were entitled to the information rather than about how they came to hold it. They may have ticked the box that made the form finishable.

The damage is not in the contradiction. It is in who finds it.

If you find it at intake, you have an awkward conversation and a narrower matter. If opposing counsel finds it at the worst possible moment, you have something else entirely. A document you tendered now contains its own impeachment, in your client's handwriting, on a page you accepted and relied on.

And it does not stay contained. A tribunal that has watched one exhibit collapse for a reason that was visible on its face applies that discovery backwards across everything else you filed. Findings that were properly obtained, properly graded and properly reasoned now sit next to a form you did not read. The cost lands on the sound work, which is the part that makes this worth a process rather than a resolution.

Read the ticks against the handwriting before you accept it

Not after. Acceptance is the moment the document becomes yours.

The check is mechanical and it takes minutes. For every item in the free text, ask whether the narrative description is consistent with the certification at the top of the page. Not whether it is lawful, which is not your question to answer. Whether the two halves of the form agree.

Where they disagree, three things happen and none of them is a judgment call:

1. Record both. The tick and the text, quoted, as they were received. Not a summary of what the client meant.
2. Quarantine the affected material. It stays in custody and out of the working set. A quarantined document is still worth reading, because its contents tell you what to ask for from a source you can properly use.
3. Refer the question to the instructing solicitor. Provenance and admissibility are theirs. Handing it over is not passing the buck; it is putting the question in front of the person whose question it is.

The two things not to do, and the second is the trap

Do not resolve it yourself. Deciding that the client "obviously meant" the lawful reading is a finding you have made about a document, without evidence, in your own favour. It is also the version that will be read out to you later.

And do not invite the client to complete a cleaner one.

This is the trap, because it feels like good administration. The form is confusing, the client misunderstood it, so you send a fresh copy with an explanation and file the tidy version.

You have now produced two signed provenance declarations covering the same material, differing on the question that matters, with the neater one produced after somebody noticed the problem. That is materially worse than one honest messy form. The first version does not stop existing, its existence is discoverable, and the sequence of events is exactly what it looks like.

One truthful document that creates a problem is worth more than two that create a problem and a question about your conduct.

The same failure as a blank field, one step later

This is the intake family again. A blank field is not a no because absence and silence arrive looking identical. A signed declaration is the same instrument one step further on: it takes a set of ambiguous human answers and stamps them as settled, which removes the last prompt anyone had to look closely.

The form is the point at which your file stops being a set of claims and starts being a record. What gets read at that moment is what you are stuck with.

If you are the one paying for the work

"Who read my declaration against my own answers, and when?"

Before acceptance, or when the bundle was being assembled? The gap between those two answers is where this defect lives.

And if a form ever comes back to you for a second signature, ask what changed and why, and make sure the answer is written down somewhere other than the new form.

(General information for practitioners, not legal advice.)