Bates numbering

A Bates number is not a page number

It is the permanent name of one page. Once a set is numbered, that reference is how the page is referred to from then on: in the index, in a witness statement, in a schedule of documents, and out loud across a room while everybody turns to it.

Which is why it works the way it does. It runs straight through the whole set and takes no notice of where one document ends and the next begins. Page numbering restarts with every document, and two documents both having a page 4 is exactly the problem this exists to solve.

It is standard in litigation and disclosure, and the same idea is used for handover packs, insurance claims, adjudication bundles and anything else where somebody has to be able to point at one page out of several hundred without ambiguity.

How this tool works

Each file is opened for its page count, and the references are worked out from that before anything is written, so the range each document will occupy appears against it in the list.

The merge and the stamp happen in one pass. The pages are copied into a new document in the order shown, and each one is stamped as it lands. Doing it the other way round, merging first and numbering afterwards, means the numbering depends on the merge having produced exactly the pages the plan assumed. It usually does. When it does not, every reference after that point is off by one and nothing in the file says so.

The page count is then checked against the number of references worked out, and a mismatch stops the job rather than stamping what it has. The map is generated from the same plan, so it cannot disagree with what was printed.

The failure that matters, and what is done about it

Two pages carrying the same reference. A reference in a statement then points at two different documents, and nobody finds out until it is read aloud.

The way that normally happens is padding. Ask for four digits, reach page 10000, and a tool that truncates gives you 0000, which is already the reference of an earlier page. Here the reference gets longer instead. It looks slightly wrong on the page and it is still unique, which is the right trade.

The other way is drift. If the documents are merged first and numbered afterwards, the numbering depends on the merge having produced exactly the pages, in exactly the order, that were counted. It usually does. When it does not, every reference after that point is off by one and nothing in the file says so. Here the merging and the stamping happen in one pass, the page count is checked against the number of references worked out, and a mismatch stops the job rather than stamping what it has.

The order of the files is the numbering

So the list is the main control on the page, and the references update underneath it as you move things around. Every document shows the range it will occupy before anything is stamped, because finding out the order was wrong after the set has been served is the whole problem in a different shape.

The map is the part nobody else gives you

Once the set is merged and stamped, the file records nothing about which original document each page came from. The reference is unique and it is also, on its own, one way.

So the CSV comes out alongside the PDF: every page, its reference, its position in the set, the file it came from and which page of that file it was. That is what an index or a schedule of documents is built from, and it is what lets you answer "where did ABC000417 come from" in a few seconds rather than by opening things until you find it.

Filenames with commas and quotes in them are escaped properly, which sounds like a small thing until a column shifts one to the right and the map maps to the wrong document.

Stamped on the page, not laid over it

The reference goes in a margin with a small white panel behind it, so it stays readable over a dark scan or a drawing border. Red by convention, so it is obviously a stamp and not part of the original. You choose the corner, the size and the margin.

Everything else about the pages is left exactly as it was. The documents are copied rather than redrawn, so text stays selectable, drawings stay sharp and the file stays roughly the size the originals were.

A disclosure set is the documents somebody is arguing about

The whole set is merged, numbered and mapped on this page using your own machine, which you can confirm in your browser network tab. That matters here more than on most tools: a disclosure set is by definition the documents somebody is arguing about, and the usual free option is to upload them to a company that keeps them for a few hours.

There is no page limit and no file limit, because there is no server being paid for by the page.