The bus Abena takes runs on the elevated track along the river corridor, retrofitted in 2039 when the lower grid flooded twice in one season and the city stopped pretending the subway would come back. She has a window seat facing east. The morning light is pale and specific. She keeps the Register in her lap with both hands flat on the cover.
The Register is a physical object, which is unusual. Most forensic trace analysts work in signed layers — their notes exist as authenticated overlays on the provenance chains themselves, timestamped and cryptographically bound to the documents they annotate. The notes travel with the evidence. That is what the courts recognize.
Abena started keeping the Register four years ago because the signed system was designed for conclusions. It had no field for I do not know what this means yet. No way to record the moment between seeing a pattern and understanding it. The Register was her answer: a place for the gap.
She opens it to an entry from 2039. Her handwriting is smaller there, more crowded, like she had not figured out how much space she needed. The entry reads: Cross-reference pattern in Meridian instruments — three issuers, different structures, same anomaly in the provenance tail. Could be a coincidence in vendor architecture. Could be something else. Cannot say which.
Below that, in later ink, she had added: Something else.
She reads it twice, then closes the Register and watches the river.
The office of Supervisor Chen is on the eleventh floor of a building that used to be a trading floor before automated market-making absorbed most of what human traders actually did. The floors below are now leased to compliance infrastructure — the server halls that run the chain verification services, the authentication nodes, the continuous audit processes mandated for any instrument of significant provenance complexity. You can feel the building humming when you press your hand flat against the wall. Abena has done this once before, just to confirm that the hum is real.
She arrives seven minutes early. The assistant at the reception node — a semi-autonomous process the office calls Yim, which Abena has never seen embodied in anything more physical than a voice and a small status light — tells her Chen is ready and to go in.
The office is a corner room. Two windows. The light is different here than it is on the river bus — more filtered, more controlled, the kind of light that does not have weather in it.
Chen is standing when Abena enters, which is either a professional courtesy or a signal. Abena has not worked out which.
Sit, Chen says. I read the preliminary.
I thought you might have questions.
I have a few. Chen settles into her chair. Her interface surface is active with what looks like the chain graph Abena submitted — three issuers, six instruments, the provenance anomaly rendered as a network map with highlighted edges. Walk me through the cross-firm pattern.
Abena puts the Register on the table between them. She does not explain it. She just puts it there.
Chen looks at it. She does not touch it.
Before I walk you through the pattern, Abena says, I want to show you what the Register is.
Abena opens to three pages without narrating them. The first: the 2039 entry with the addendum in later ink. The second: a page from 2041, dense with a notation system she developed after the Kessler case, margins annotated with small diagrams showing how she was thinking through a chain topology problem. The third: a recent entry, spare and precise, uncertainty flagged in color code — amber for possible, red for significant, the notation X-GR-01 in the upper right corner.
Chen looks at all three.
The absence of margin questions in the recent one is not confidence, Abena says. I built the questions into the entry structure. What looks simpler is the same amount of uncertainty, just more organized.
And X-GR-01.
My internal label for the cross-firm pattern. I named it before I could describe it. That is why I use working labels — they hold the shape of something I do not have language for yet.
Chen pulls her interface surface closer. The Meridian group. The Hauer-Reyes instruments. And the Sunstone secondary issuances. She looks up. You are saying these share a provenance anomaly.
The anomaly appears in the tail of each chain. The provenance tails on all six instruments terminate correctly — every link verifies, every signature authenticates. But they terminate at the same intermediate node. Three different issuers, different legal structures, different chain architectures. They should not share an intermediate node. The probability of that being coincidental vendor overlap is below two percent across four separate inference runs.
Chen is quiet. Outside, a delivery drone passes the building perimeter, its navigation lights tracking green. Every meter of that route is authenticated — signed by its onboard chip, timestamped, chain-verified, a continuous provenance record of exactly where it was and when. The Register in Abena's lap is none of those things. It is a physical notebook with no authentication chip, no cryptographic signature, no chain. It proves nothing happened except that a hand wrote in it.
That is the point.
The Register, Chen says. You brought the physical copy.
Yes.
Not the scan. Not a signed layer.
The scan is evidence. The original is presence. Abena keeps her voice level. I want the Register to be in the room. Not presented.
Chen reaches across the table for the first time and touches the cover — not opening it, just touching the edge with two fingers, the way you test the temperature of something. Then she moves her hand back.
What you are proposing, Chen says, is that we treat the Register as a source.
Yes. Not as an output. Not as a record of conclusions. As a source with a different kind of validity than the signed layers — one that documents how inference moves rather than where it lands. The signed layers prove the chain exists. The Register proves I was uncertain about what the chain meant, and for how long, and what I noticed before I understood it. Courts treat those as the same thing. They are not.
If the Register becomes a source, Chen says, it changes the evidentiary framework. Uncertainty becomes part of the record. Most investigators in this division would object.
I know.
They would say it weakens cases.
It would require courts to consider how investigators arrive at conclusions, not just whether the conclusions verify. That is not the same as weakening cases.
It is also not something I can approve on my own. But Chen is still looking at the Register. Who else has seen this?
No one. This is the first time I have put it on a table.
Leave me the scan, Chen says. And the pattern inference runs.
Already in your queue.
The Register goes back with you.
Of course.
Chen stands, which means the meeting is ending. Abena puts the Register back in her bag. The strap catches on the table corner and she has to adjust. A small thing. She notices it.
The X-GR-01 pattern, Chen says, when Abena is almost at the door. How long have you been watching it?
Abena turns. Since 2039. When I wrote that I did not know what it was.
Chen nods. Write it up. All of it — including the uncertainty.
That is the part most write-ups do not include.
That is why most write-ups are not useful.
On the bus back, Abena opens the Register to the 2039 entry again. Something else, in later ink. She thinks about what it would mean to write the full document Chen described — not just the six instruments and the shared intermediate node, but four years of not knowing, the working labels, the color codes, the amber flags. A document that showed how the inference moved.
She does not write anything yet. She is still inside the period before the words arrive.
She lets that stand.
The river below is the color of old authentication ink — a gray-green the water took on after the sediment treatments in 2041, a color that looks like something waiting to be signed. She watches it and keeps her hands flat on the cover of the Register.
The bus makes its stop. She does not get off. She is not ready to be somewhere else yet.
She rides one more stop past her station and walks back.