Bates numbering for legal discovery — what the rules actually require

Bates numbering exists because Federal Rule of Civil Procedure 34 — and its state-court analogs — require parties to produce documents in a form where every individual page can be identified and cited. Without stable page identifiers, depositions degrade into argument over which page the witness is being shown; motion practice becomes unworkable when one side cites "the third email in the second binder" and the other side has reorganized the binder. The Bates Manufacturing Company's hand-numbering machines were the original solution — a self-incrementing stamp pressed onto every page of a production set. The name stuck even after the hardware did not.

Modern productions follow conventions that have hardened over three decades of e-discovery practice: a short alphanumeric prefix tied to the producing party (ABC, JOHNSON-, SMITH-001), followed by a zero-padded sequential number wide enough to cover the expected production size. ABC000001 means width six; SMITH-00001 means width five with a hyphenated prefix. The stamp typically sits in the bottom-right corner in monospaced font (Courier is the de facto standard) so that every number occupies the same visual width, which keeps load-file coordinate references consistent in review platforms like Relativity, Concordance, and Everlaw.

Why a browser-local tool matters for privilege

Every web-based bates tool that asks you to upload your production set creates a privilege problem. FRCP 26(b)(5)(B) and the parallel state rules permit clawback of inadvertently produced privileged material — but only if the producing party took reasonable steps to prevent disclosure. Sending a 50,000-page production through a third-party SaaS vendor's bates utility puts the documents on someone else's infrastructure for the duration of processing; if a privileged document slips through your privilege review and gets bates-stamped on a vendor server, you have created a disclosure event before the production even leaves your office.

The same concern applies to attorney work product, trade secrets covered by protective orders, and HIPAA-regulated medical records. Courts have been increasingly willing to scrutinize the "reasonable steps" inquiry under Rule 502(b), and a vendor-uploaded processing step is a weaker defense than a fully client-side workflow. Browser-local stamping eliminates the upload entirely — pdf-lib runs the page modifications in JavaScript, the file never crosses the network boundary of your machine, and there is no third-party processor to disclose to opposing counsel.

Multi-file continuous numbering — the production-volume workflow

Real productions arrive in volumes. Volume 1 might contain the custodial documents for the first witness; volume 2 contains the centralized accounting records; volume 3 contains a later supplemental production ordered to cure deficiencies. Each volume needs a distinct, non-overlapping bates range. The standard practice is to track the last number used in volume 1, then start volume 2 at that number plus one. The tool enforces this directly — after stamping, the last number actually written is displayed, and the starting-number field for the next run lets you set the resume point.

For a single-volume production with multiple PDFs (one per custodian, one per document family), drag every PDF into the upload zone in production order, set the starting number once, and the tool numbers continuously across files. The output is a single ZIP containing every stamped PDF ready to ship to opposing counsel or upload to your review platform's production module.

Beyond bates — privilege logs, exhibit numbering, deposition exhibits

The same stamping infrastructure powers a few adjacent legal workflows. Privilege log entries often carry their own identifier convention (PRIV, LOG-, WP- for work product); set the prefix accordingly and stamp your redacted privilege production with a non-overlapping range. Deposition exhibits require sequential numbering tied to a deponent (SMITH-EX-001, JOHNSON-EX-001); the prefix field accepts up to ten characters, including hyphens. Trial exhibits follow yet another convention (PX1 for plaintiff, DX1 for defendant); use the tool with prefix PX or DX and width 1 to produce trial-style identifiers.

Position, margin, and the rotated-page problem

Bottom-right is the discovery default because it does not interfere with document-native footers (page numbers, document IDs from the originating application, signatures). Top-right is sometimes preferred when the document already has a footer block, or when the receiving party's review platform crops the bottom margin. Top-left and bottom-left are uncommon in US practice but appear in some international productions. Margin is measured in PDF points (72 per inch); the 36-point default places the stamp half an inch from the page edge, which clears most printer-safe margins.

Scanned discovery sets often arrive with rotated pages — the scanner misread the orientation, or the source document was landscape but the scanner produced portrait pages with a 90-degree rotate flag. The tool reads the page's /Rotate value before stamping and maps the visual corner (what the user sees) to the underlying coordinate system, so the bates number appears upright in the bottom-right of the page as it displays — not in the unrotated coordinate frame, which would put the stamp sideways or upside-down. This matters because review platforms render with rotation applied; an unrotated stamp would land at the wrong corner of the visible page.

Quality assurance — what to check before producing

Before sending a stamped production set to opposing counsel, verify three things. First, open the first and last page of the first PDF and confirm the bates number range matches what you expected. Second, open the first page of the second PDF and confirm the number is exactly one greater than the last number of the first PDF — gaps or overlaps will raise objections. Third, search the stamped PDF for the bates number; if it appears as searchable text, the stamp was written into the content stream correctly (the tool always does this — image-based stamps would defeat the purpose of a citable identifier).

For matter-critical productions, run a few sample pages through your review platform's load-file generator before stamping the full set. This catches cases where the platform's coordinate parser disagrees with the stamp position — for example, if your platform expects the bates number in a specific OCR text region and the chosen margin places it just outside that region. Adjusting the margin or position before stamping the full production is far cheaper than re-stamping.

Companion tools for legal-discovery PDF workflows

Bates numbering is one step in a broader discovery workflow. Other PDF Mavericks tools that pair with this one: Redact PDF for permanent redaction of privileged or PII content before stamping; Delete PDF Pages for trimming wholly-privileged pages out of a document before bates stamping the remainder; Merge PDF for combining individual document families into a single production volume before bates; Split PDF if you need to break a stamped production into smaller deliverable units while preserving bates ranges; and Unlock PDF for decrypting password-protected source documents before they enter the stamping pipeline. All of these run client-side in the browser — no upload, no third-party processor, no privilege exposure.

What this tool does not do

A few honest limits. The tool does not generate a load file (DAT, OPT, or CSV) tying bates ranges to document-level metadata — that is a review-platform export, not a stamping operation. The tool does not OCR image-based PDFs before stamping; if you need searchable text underneath the stamp, run the document through OCR PDF first. The tool does not strip metadata; bates stamps go on the page content, but the document's underlying author, creation date, and revision history remain unchanged. Use the blog for guides on a complete production-prep workflow.

Frequently asked questions