How to Build a Case Chronology From Scanned Documents

To build a case chronology from a file that is mostly scanned paper, convert the scans to searchable text, then enter one row for every dated event: the date, what happened, who was involved, and the page that proves it. Sort by date, flag what is disputed, and check every cite against the page image.

What belongs in a litigation case chronology?

A litigation case chronology holds events, one per row, each with a date, a plain statement of what happened, the people involved and a pinpoint cite to its source. Argument, characterization and anything without a source stay out. If a row cannot be cited, it is a question for someone, not an entry.

The columns that earn their place:

Two disciplines keep the table useful as it grows.

One event per row. A letter that says "we delivered the steel on 3 May and you rejected it on 6 May" is two events on two dates. Enter two rows that cite the same page.

The event, not the document. "Letter from Voss to Tran" is an index entry. A chronology entry says what the letter shows happened: "Voss notified Tran that the crane permit had lapsed." The sending of the letter can be an event of its own when notice matters. The things the letter reports are separate events on their own dates.

Write each event so that both sides would accept the sentence as a fair reading of the source, and put disagreement in the status column. "Greywater withheld the April payment" can be cited. "Greywater wrongfully withheld the April payment" is a conclusion.

Which scanned documents feed a chronology first?

Start with the scanned documents that fix the outline of the story with the least reading: the pleadings, the contract and its amendments, and the correspondence between the parties. Add the high-volume records and the testimony once that spine exists, so each new entry has neighbours to be checked against.

A working order for a commercial dispute:

Each tier fills gaps left by the one before. The answer tells you which complaint dates are admitted, so those rows can be marked undisputed on day one. The contract tells you what the deadline was before the letters start arguing about it.

Start from the complaint when you build the first rows. Each allegation gives you a date to test against the letters. Where the answer admits a fact, mark that row as admitted on the page you cite. A cause of action that depends on notice or on damages needs those dates in order. If you are collecting pages for a prima facie showing, cite the scan that carries each fact.

Two kinds of source have their own methods and their own posts. Testimony goes in from a page-line summary, and the guide on how to summarize a deposition transcript covers producing one. Treatment records follow encounter dates, as set out in the section on building a medical chronology from converted records. For which papers need converting at each point in a lawsuit, see the stage-by-stage guide to litigation OCR.

If the file has been indexed, sort the index by date and work down it. If it has not, sort the scanned files into these tiers by eye before converting anything, and convert tier by tier.

Pulling dated events out of converted letters, emails and records

Pull events out by converting each scanned document to text, searching the text for every expression of a date, reading the sentence around each hit on the page image, and writing one row per event. The search makes sure no date is passed over. The reading decides what happened on it.

Convert

Each scan needs a text layer or a text export before you can search it. In Schema Law you upload the PDF, PNG or JPEG in the browser, view the parsed pages, and export. A searchable PDF keeps the page image and adds hidden text, which suits this job because the hit and the picture of the page are in one file. For a document with no tables, the CSV export lists the text line by line with a page number beside each line, which helps when you want to paste a sentence and its page into the quote and source columns. The export formats section of the site lists all eight.

Search for dates in every form they take

A date on paper is not always digits. Search each document for:

• Month names and their abbreviations: January through December, Jan., Sept., and so on.
• Numeric dates: search for the year, such as 2022 and /22, to catch 5/3/2022 and 5/3/22.
• Weekdays: Monday through Sunday, which lead you to "last Friday" and "this coming Tuesday".
• Relative words: yesterday, today, tomorrow, last week, next month, "the following day".
• Anchors: "on or about", "as of", "effective", "no later than", "within ten days".

Recognition can misread a date stamp or a faint dateline, and nothing in the converted text marks a doubtful character. Page through each document once after searching, looking only at the places dates live: the top of each letter, fax header lines, received stamps, and signature blocks.

Read, then write the row

A worked example from the fictional case Penhallow Builders v. Greywater Hotel Group. A scanned letter, stamped PEN000231 to PEN000232, is dated 9 May 2022, from Lena Voss of Penhallow to Marcus Tran of Greywater. Its second paragraph reads: "The structural steel arrived on site 3 May. Your engineer rejected it on 6 May without a written reason. Unless we hear otherwise by 16 May we will treat the rejection as withdrawn."

That one paragraph yields four rows:

Notice what the status column is doing. That Voss wrote the letter and what it says can be read off the page. Whether the steel arrived on 3 May is Penhallow's account until a delivery log or the other side confirms it, and the event sentence says whose account it is.

Keep the pace

Work one document at a time, enter rows in any order, and sort at the end of each session. Resist reading for interest. If a passage raises a question, put it in the notes column and move on.

How do you cite the source page for every chronology entry?

Cite the source page by its permanent label: the Bates number for a produced document, page and line for a transcript, and the filed document's own page for a pleading. Enter the cite when you enter the row, never afterwards, and never cite the page counter your PDF viewer shows.

The reason for pinpoint cites is where the chronology ends up. On summary judgment in federal court, a party asserting that a fact cannot be or is genuinely disputed must support the assertion by, in the words of Rule 56(c)(1)(A), "citing to particular parts of materials in the record". State rules and local rules set their own requirements, so check the ones for your court. A chronology that already has a particular page on each row is most of the way there.

Use one cite format per kind of source, and write the formats at the top of the sheet:

The last row is a stopgap. A cite to "page 41 of a PDF" breaks the day someone removes a blank page or merges two files. Flag every such row, and replace the cite once the document is numbered. The post on keeping each page tied to its Bates number explains how to get from a position in a file to the stamp on that page.

Three habits keep cites from rotting:

1. Cite the page the fact is on, not the first page of the document. A reader checking row 200 should land on the sentence.
2. Copy the supporting words into the quote column. If the cite is later found to be off by a page, the quote lets you search for the right one.
3. Verify in batches. At the end of each session, open ten rows at random, go to each cited page, and confirm the fact is there. Fix what you find and note that you checked.

When two sources support the same event, list both in the source column, separated by a semicolon. A delivery on 3 May shown by a letter and by a signed delivery ticket is a stronger row than either alone, and the lawyer will want to choose which to cite.

Undated documents, date ranges and conflicting dates in a chronology

Give every row a sortable date even when the source is vague, and record how firm that date is in a column of its own. Enter ranges with a start and an end. When two sources give different dates for one event, keep both rows and mark the conflict. Choosing between them is not the chronology's job.

A certainty column

For derived dates, write the arithmetic in the notes column so the next reader can redo it. For inferred dates, write the basis. An inferred date with no stated basis is a guess that will be mistaken for a fact once the table is sorted.

Sorting what is not exact

A spreadsheet needs a real date to sort on. Use a convention and state it at the top of the sheet:

• Month known, day unknown: enter the first of the month, mark Approximate, and begin the event text with "In May 2022".
• Only the year known: enter 1 January of that year, mark Approximate, and begin with "In 2022".
• Undated document, placed by context: enter the inferred date and mark Inferred.
• Undated, and nothing places it: keep the row on a separate tab named Unplaced. It stays visible, and it does not pollute the sorted view.

Ranges

For something that lasted, such as a shutdown, a notice period or a course of negotiation, add an End date column. Enter the start in Date and the finish in End date, and write the event as a span: "Site closed by order of the fire marshal." One row with two dates is clearer than a start row and a stop row a hundred lines apart. Add a second row at the end date only if the ending was an event in its own right.

Conflicts

Suppose Voss's letter puts the rejection of the steel on 6 May, and Greywater's site diary, at GHG001477, records it on 5 May. Enter both:

Before treating it as a real conflict, look at both page images. A 5 and a 6 are close enough in a faint scan or in handwriting that one of the two dates may be a misreading, by the software or by you. If the images plainly differ, the conflict is evidence, and the lawyer will want to see it exactly as the sources state it.

A date wildly out of sequence deserves the same check. A reply dated before the letter it answers may be a recognition error, a typing error in the original, or a fact worth a deposition question. Only the image tells you which.

Turning a case chronology into a statement of facts or witness outline

Turn the chronology into a statement of facts by filtering for the rows that matter to the motion and are marked undisputed, then writing each as a numbered sentence with its cite. Turn it into a witness outline by filtering on one person's name and adding the document you will show for each row.

A statement of facts

1. Filter by issue. Keep the rows tagged to the claims or defenses the motion addresses.
2. Filter by status. Undisputed rows are candidates. Disputed rows show the lawyer where the other side will push back.
3. Rewrite each event as a full sentence. The event column was written without adjectives for this moment. Drop phrases such as "according to" only where the lawyer is satisfied the fact itself is established.
4. Attach the cite in the form the court requires. A Bates cite in the chronology becomes a cite to an exhibit, a declaration paragraph or a record page in the filing. Add a column that maps each source to its exhibit number.
5. Check every cite again. Open the page for each numbered fact in the draft. This is the pass that matters most, because the reader is now a judge.

Courts differ on the form of a statement of facts: separate document or part of the brief, numbered paragraphs or narrative, what a response must look like. Read the local rules and the judge's standing order before formatting anything. The lawyer decides which facts are material and how each is put. The chronology supplies the candidates and the proof.

A witness outline

Filter the People column for the witness, keep the date order, and add two columns: the document to show, and the question area. For Marcus Tran, the filter returns every row where he wrote, received, attended or was named. Read down it and three things appear that are hard to see any other way:

• Gaps. Weeks in which the witness appears in no document are worth asking about.
• Sequences. A notice on the 9th, silence, and a payment withheld on the 20th read as a story when stacked.
• Conflicts. Disputed rows involving the witness are the questions that have to be asked.

Export the filtered rows to a document with the source column intact, so whoever takes the deposition can call for a page by number.

Keep one master

Every statement of facts and outline is a filtered copy. Corrections go into the master chronology first and flow outward, never the other way. Note on each derived document the date of the master it was cut from. When new pages arrive, convert them, add their rows, and re-cut. The Schema Law pricing page shows the page allowance on each plan if later productions are large.