Settings

Because a Word file does not contain pages. It contains text and formatting rules, and every machine that opens it lays the pages out again from scratch. Anything that changes how much room a line of text takes — a font that is missing and substituted, a different version of Word, a compatibility setting, an older version measuring with the printer's metrics — pushes one line onto the next page, and every page number after it shifts by one. A page number in Word is a fact about the machine that printed it. A citation that has to hold needs something the layout cannot move: a numbered paragraph, a PDF made once and shared, or a transcript with a fixed number of characters to a line and lines to a page.

What a Word file actually stores

A .docx is a zip of XML, and the main part of it is a run of paragraphs with their formatting. It records where you put a manual page break and nothing else about pages. Which words fall on page twelve is worked out when the file is opened, by measuring each character in the named font at the named size and stacking lines until the page is full. Word does write a note of where the breaks fell the last time it saved, and other programs read that as a hint; Word itself recomputes. Measure with a slightly different ruler and the answer changes, and the change is not local: a line that no longer fits on page three pushes a line onto page four, which pushes one onto page five, and so on to the end.

The four things that move the breaks

  • Fonts. The commonest by a distance. A font that is not installed on the second machine is substituted with one that is, and the substitute is a little wider or narrower. One word no longer fits on its line, the line no longer fits on its page, and everything after it moves. It happens most when a document crosses from Windows to a Mac or a phone, or is opened in Google Docs or LibreOffice, because each of those substitutes differently.
  • The printer. Older versions of Word measured text with the metrics of whichever printer driver was selected, so the same file paginated differently on two machines with two printers, and differently again after somebody changed the default printer. Current versions use their own measurements, and keep the old behaviour as a compatibility option (Use printer metrics to lay out document) that a document saved by an older version can carry with it.
  • The version of Word, and compatibility mode. The layout rules have changed between versions — line spacing, how tables and pictures anchor to text, kerning, hyphenation. A document opened in a newer version is either laid out with the new rules or held in compatibility mode with the old ones, depending on what saved it and whether anybody chose to convert it. Two people with two versions can each be right about their page count.
  • Anything anchored to the page rather than the text. A picture positioned relative to the page, a floating table, a text box, a heading set to start a new page. A small shift upstream changes what these collide with, and the result is sometimes a whole empty page that exists on one machine and not the other.

Why it matters more than it looks

For a document only you will print, it does not. It matters the moment somebody cites a page. “See page 42” in an email, a report, a witness statement, an examiner's notes on a thesis: whoever follows the reference opens the file on their own machine and turns to their own page 42, and the words are there or they are not. Nobody notices the shift, because there is always something on page 42.

Which is why the references that have to hold are never Word page numbers. Judgments are cited by numbered paragraph, statutes by section, transcripts by the page and line of a fixed layout. Each of those is a reference the layout cannot move, and each was chosen because a page number could.

What holds, and what does not

ReferenceHolds on another machine?Why
A Word page numberNoRecomputed by every machine that opens the file.
Word's own line numbers (Layout › Line Numbers)NoThey count the lines as laid out, so they move with the wrapping, for the same reasons the pages do.
A page of a PDFYes, for that PDFA PDF fixes the layout at the moment it is made. Two PDFs made from the same .docx on two machines can still differ, so cite the one file everybody has, and send it.
A numbered paragraphYesThe number travels with the text, not the layout. The one trap is automatic numbering that restarts when a list is split; check it, or type the numbers.
A headingMostlyText, not layout, but a long section under one heading is a vague reference.
Page and line of a fixed-column transcriptYesBoth are counted from the characters, so the same text gives the same numbers anywhere, provided the two settings are stated.

Making a citation that holds: a transcript

A transcript is where this matters most and where the fix is oldest. Court reporters produce transcripts with a fixed number of characters to a line and a fixed number of lines to a page, commonly twenty-five, and every reference to one is a page and a line: 42:17. The numbers come from the characters, not from a layout, so two printings agree. A transcript that arrives as a Word file, or as a block of text from a transcription app, has none of that: cite it and nobody can find the passage; print it on a different machine and the breaks move.

FOLIO does the one job. Save the Word file as plain text (File › Save As › Plain Text), paste it or open it, set the characters to a line and the lines to a page, and it wraps at word boundaries, numbers every line including the blank ones, and never reflows. Its own example shows what the layout does to a citation. The example transcript wrapped at 62 characters and 25 lines to a page is one page of eighteen lines, and clipboard is at 1:11 and 1:17. The same words wrapped at 40 characters and ten lines to a page are three pages, and clipboard is at 2:5 and 3:4. Same text, different citation, which is exactly what Word does to a document between two machines. The difference is that FOLIO's two settings are written on the front — wrapped at 62 characters, 25 lines to a page — so anybody with the text can reproduce the numbers, and the same text and settings give the same output on every load.

The other half is the index. List the terms you will be citing and every occurrence appears with its page and line, whole words only, so shutter does not also report shutters. That is the list an advocate builds with a highlighter the night before, and it comes out as a CSV.

What it is not: an official transcript, or a reader of Word files. It takes plain text, so runs of spaces collapse to one, which turns the double space after a speaker's name into a single one, and a word longer than the column overhangs rather than being split, because a citation that points at half a word is worse than a line that runs long. Nothing pasted into it is uploaded.

If it has to stay a Word document

  1. Embed the fonts (File › Options › Save › Embed fonts in the file, on Windows). That removes the commonest cause, at the price of a larger file, and it does not help when the file is opened by something other than Word.
  2. Turn off printer metrics if the document carries the option (File › Options › Advanced, under the compatibility options at the bottom).
  3. Leave compatibility mode (File › Info › Convert) so that everybody with a current version lays it out by the same rules — only worth doing if everybody has one.
  4. Make the PDF yourself and send that, and cite the PDF's pages. This is the honest answer for anything that will be read by people you did not choose.
  5. Number the paragraphs and cite those.

None of the first three makes a Word page number a stable reference. They make it shift less often, which is a different thing, and the day it shifts is the day somebody is looking for a passage in front of other people.

Questions people ask about Why a Word document's page numbers change on another computer

Why does the same Word document have a different number of pages on another computer?

Because the pages are not in the file. Word lays them out again every time the file is opened, measuring the text in whatever fonts that machine has, by the rules of whatever version of Word it is running. A substituted font, a different version, a compatibility setting or, in older versions, a different printer's metrics, each changes how much fits on a line, and one line moving moves every page break after it.

Does saving as PDF fix it?

For that PDF, yes: the layout is fixed at the moment it is made and everybody who opens the file sees the same pages. Two PDFs made from the same Word file on two machines can still differ, so make one, send it, and cite that one's pages.

Can I use Word's line numbers for citations?

Not across machines. Word's line numbering counts the lines as they are laid out, so it moves with the wrapping for the same reasons the page numbers do. A line number only holds when the wrapping is fixed by rule — so many characters to a line — rather than by layout.

What are the page and line numbers in a transcript?

A transcript is produced with a fixed number of characters to a line and lines to a page, commonly twenty-five, and every reference to it is a page and a line, 42:17. Because both are counted from the characters rather than from a layout, two printings of the same transcript agree, which is the whole reason the convention exists.

How do I get a Word transcript into FOLIO?

Save it as plain text from Word (File › Save As › Plain Text), then paste the text into FOLIO or open the file there. Set the characters to a line and the lines to a page, and write those two settings on the front of anything you cite from it, because anybody with the same text and settings gets the same numbers. Runs of spaces collapse to one on the way in.

Will FOLIO's numbers match the court reporter's?

Only if the text and the two settings are the same as theirs, and they usually are not, so treat it as a citable copy of your own rather than the official transcript. What it guarantees is narrower and more useful: the same text at the same settings gives the same page and line on every machine, every time, so a reference you give today still points at the same words next month.

Related tools