How to Bates stamp emails
Stamping is mechanical. Deciding what to stamp, in what order, with what format, and how attachments relate to their parent messages is not, and those decisions are effectively irreversible once a production has gone out under numbers the other side is now citing.
This covers the conventions rather than the button-clicking, because the button-clicking is different in every numbering tool and the conventions are the same everywhere.
What the number is for
A Bates number gives every page in a production a unique name that both sides can use without ambiguity. Once assigned, it is that page’s identity for the rest of the case: it appears in the privilege log, on deposition exhibits, in briefs, and in any dispute about what was produced.
Two rules follow, and they are absolute:
A number identifies exactly one page, forever. Never reused, never reassigned, not even after a production is withdrawn and reissued.
Gaps are meaningful. A production running JONES000001 to JONES004500
with 200–240 missing is telling the other side that forty pages were withheld,
and they are entitled to an explanation. That explanation is the
privilege log.
Why email is the hard case
Scanned paper arrives pre-paginated. A page exists, you image it, you stamp it. Email has no inherent page count — the pagination is manufactured by whatever rendered the message into pages.
That has three consequences people meet in the wrong order:
Rendering decides pagination. The same message rendered by Outlook’s print dialog, a browser’s print-to-PDF, and a conversion tool produces three different page counts. If your set was assembled from all three, the page counts in your log will not match anything.
Quoted threads inflate everything. A twelve-message thread quoted at each reply contains its own earlier messages repeatedly. Stamped in full, you pay for those pages, the other side reviews them, and the same content carries a dozen different numbers.
Attachments are separate documents. One email with three attachments is four documents with a relationship, not one document with extra pages.
The first is the one that ruins productions, and it has a simple fix: render everything once, through one tool, before anything is stamped.
The order of operations
This sequence is not stylistic. Each step depends on the one before it.
Convert first. Every responsive message through the same pipeline, so page counts are a function of content rather than of who printed what.
Sort deliberately. Filename order is arbitrary — it reflects export order or subject-line alphabetisation, neither of which means anything. Sorting by the date on the message produces a sequence that reads chronologically, which is what makes a numbered thread comprehensible to a judge reading it once.
Stamp the complete set. Including documents you will withhold. This is the step most often done backwards, and doing it backwards leaves the withheld documents with no identifier to log.
Then pull and log. Remove the privileged documents, record their ranges and the basis for withholding.
Combine before you stamp
The most useful practical trick in email numbering: produce one continuous PDF of the whole batch in the sorted order, and stamp that single file in one pass.
- The sequence cannot break, because there is one file.
- The order is the order you chose, not one a tool inferred from a folder listing.
- Reconciling the produced range against your log becomes reading two numbers rather than auditing four hundred filenames.
Keep the individual message PDFs too. When one email becomes Exhibit C you want that message standing alone, not a range carved out of a combined document.
Attachments and ranges
An attachment gets its own numbers because it is its own document and will be cited on its own. What has to survive is the relationship — that this spreadsheet arrived on that email.
Keep the ranges adjacent: parent message first, then its attachments, then the next message. In a straight PDF production that adjacency plus the folder structure is usually enough. In a platform production the relationship is carried explicitly in the load file’s parent-child fields, and adjacency alone is not sufficient.
Where a protocol requires attachments produced natively rather than imaged, the common practice is to assign the native a number and produce a placeholder page bearing that number in the image set.
Renumbering, and why you should not
Occasionally a production has to be reissued — a document was missed, a privilege call was reversed, a format was wrong. The instinct is to renumber the corrected set from the beginning so the sequence is tidy.
Resist it. The other side has already cited the original numbers in
correspondence, in deposition notices, possibly in filings, and those citations
do not update. A page that was JONES000412 and is now JONES000398 breaks
every reference to it, and the resulting confusion lands on you.
The convention is to leave issued numbers alone and add. A supplemental production starts where the last one ended. A document withdrawn stays withdrawn at its original number, accounted for in the log. A document added late takes the next available number regardless of where it sits chronologically. The sequence is a register of what was produced and when, not a tidy index of the final document set, and treating it as the latter is what creates the mess.
Confidentiality designations
Where a protective order provides for designations, they are usually applied at
the same time as the numbers, as a suffix or a separate legend —
JONES000001 — CONFIDENTIAL.
Two things to get right: designate at the document level rather than page-by-page unless the order requires otherwise, and keep a record of what was designated, because designations get challenged and you will need to justify them individually rather than as a batch.
Keep a register
Somewhere outside the production itself, record which ranges have been issued in this matter, to whom, and when. Productions come in waves, and the second wave has to start where the first ended.
This is unglamorous and it is the thing that prevents the one unrecoverable error in the whole process. A misprinted stamp can be re-run. A number issued twice, to two different pages, in a case where both have already been cited, cannot be quietly fixed.
Doing it in EML Exhibit
-
Choose the prefix
A party or matter abbreviation, in caps, without spaces or punctuation that a filesystem will rewrite. SMITH, JONES_DEF, ACME. It identifies who produced the document for the life of the case.
-
Fix the padding
Decide the digit count up front and never change it. Six digits handles a million pages and is the common default. Padding is what makes the set sort correctly in every tool that will ever touch it.
-
Render the whole set through one pipeline
Convert every message to PDF in one pass so pagination is consistent. The stamp is applied to pages, so inconsistent pagination produces a set that cannot be reconciled against a log.
-
Decide the order before stamping
Sort by the date on the message, not by filename. The number sequence encodes an order permanently, and chronological is the order that reads as a narrative.
-
Stamp the complete set, including withheld documents
Number everything that came out of review, then pull the privileged documents. Withheld documents need stable identifiers for the log.
-
Stamp one combined file where you can
A single continuous PDF stamped in one pass cannot lose its sequence. Stamping four hundred separate files depends on the tool preserving your order, which is a dependency worth removing.
-
Record the range you used
Keep a register of which ranges have been issued in the matter. Reusing a number is the one error that cannot be quietly fixed.
What prefix format should we use?
Whatever the protocol specifies, or your firm’s existing convention if it
does not. The common shape is a party abbreviation plus a zero-padded
sequence — JONES000001. Some protocols require a separator
(JONES-000001) or a confidentiality suffix on designated documents. The
letters matter less than consistency; the padding matters a lot.
Why does zero padding matter so much?
Because without it the set sorts as 1, 10, 100, 2 in essentially every file browser, spreadsheet and database. That is discovered at the worst possible moment, usually while trying to locate a document during a deposition. Pick a width that comfortably exceeds the expected page count and stay with it.
Where on the page should the stamp go?
Bottom right is the overwhelming convention, because it is where a reader flipping pages looks and because it is least likely to collide with document content. Check that it does not overprint the rendered message — a stamp sitting on top of the last line of a body is a re-run.
Do attachments get their own numbers?
Yes. An attachment is a separate document, and it needs its own range so it can be cited independently. The relationship to the parent message is expressed by keeping the ranges adjacent and, in a platform production, by the parent-child fields in the load file. See parent-child email productions.
Does EML Exhibit apply the stamp?
No. It produces the input a numbering tool needs — consistent pagination, an optional single combined PDF, a matter prefix on every filename — and the stamp itself comes from Acrobat or your firm’s numbering software. Native stamping is on the roadmap. See preparing email PDFs for Bates numbering.