Put events in order with the source for each one next to it, and say plainly whether a line rests on a document, on what somebody said, or on your own inference. Attach the file an entry relies on to record its SHA-256, so the chronology and the exhibit list refer to the same thing. Export a clean chronology to hand over. Kept in this browser, nothing uploaded.
One event per entry. If only the date is known, leave the time off; it will sort ahead of that day's timed entries.
Always shown in date order. Gaps of a month or more are marked, because a quiet stretch in a timeline is often where the questions are.
The printable chronology is what you hand over. The JSON is what you load back here to carry on, or merge with a colleague's entries.
Order is not cause. Two events next to each other are just two events next to each other. The tool sorts and labels; it does not conclude, and a chronology that reads as though it does should be rewritten.
The basis column is the honest part. "Documented" means a record made at the time. "Reported" means somebody said so later. "Inferred" means the author worked it out. Keeping those apart is most of what makes a chronology worth reading.
Dates come from you. Attaching a file suggests its modified time as a starting point, and that is all it is. A file's clock says when it was saved, not when the thing in it happened.
Digests tie entries to files. Where an entry carries a SHA-256, the same digest will appear in a DOCKET schedule or CHAIN log built from the same file, so a reader can match the two without trusting the filename.
Nothing leaves this tab. The chronology lives in this browser's local storage until you clear it. The export is the record.