Why page 1 of your PDF is not page 1, and how to fix it

Someone tells you the figure is on page eight. You type eight into the reader and land on the middle of a different section. You scroll, find the page that says eight at the bottom, and your reader calls it page twelve.

Nobody has made a mistake. There are genuinely two numbering systems at work, and they are not required to agree.

Two numberings, one document

The first is the sheet position: this is the fifth sheet of paper in the file. Your reader uses it, it always starts at one, and it always increases by one. It is a property of the file rather than the content.

The second is whatever is printed on the page. That was decided by whoever wrote the document, and it follows the conventions of publishing: front matter numbered in roman numerals, the main body restarting at one, appendices numbered separately, and unnumbered cover pages at the front.

A report with a cover, a contents page and two pages of front matter has its printed page one on the fifth sheet. Every cross-reference in the document is then four out of step with the reader, and stays that way to the end.

• Cover pages and blank separators that carry no printed number
• Front matter numbered i, ii, iii before the body restarts at 1
• Appendices numbered A-1, A-2 rather than continuing the sequence
• Documents assembled from several sources, each with its own numbering
• Pages inserted or removed after the numbers were printed

Why it matters more than it sounds

In anything where people cite pages to each other, a four-page offset is a persistent tax. Legal bundles, tender submissions, academic work and long reports all suffer from it, and the confusion is worst when some people quote printed numbers and others quote reader numbers without saying which.

This is exactly why legal work uses Bates numbering: a single unambiguous number stamped on every page of a bundle, independent of whatever each original document called itself. When there is one number and everyone uses it, the ambiguity disappears.

Adding numbering that actually helps

1. Decide which pages should carry a number. Covers and blank separators usually should not.
2. Decide where the sequence starts. If the document already has printed numbers, match them rather than adding a competing set.
3. Choose a position that will not collide with existing content, and check the pages that already have footers.
4. Use a format that removes doubt. Page 4 of 32 tells a reader far more than a bare 4.
5. Page through the result rather than checking only the first page.

That fourth point is the most valuable and the most often skipped. A page reading four of thirty-two immediately tells the reader whether anything is missing from the copy in front of them, which a bare number cannot.

When there are already numbers on the page

Adding a second set on top of an existing one makes the confusion worse rather than better. If a document is already numbered and the numbers merely start at an offset, the honest fix is to leave the page alone and tell people the offset.

Where you are assembling a bundle from several numbered documents, the right move is to add one continuous sequence in a distinct position — bottom right, say, if the originals number at the bottom centre — and then refer to that sequence exclusively. Say clearly which numbering you are using when you cite a page, and most of this problem simply stops happening.