typesetting.ai

Widows and orphans, and where a page is allowed to break

Published 4 min read

By the typesetting.ai editorial team

An orphan is the first line of a paragraph left alone at the bottom of a page, with the rest of the paragraph overleaf. A widow is the last line of a paragraph left alone at the top of the next page.

The mnemonic that sticks: an orphan is alone at the start with no family behind it; a widow is alone at the end, left behind. Nobody agrees on which is worse, and it does not matter — the fix is the same for both.

Why they read as a mistake

Neither costs the reader any comprehension. A stranded line is perfectly legible. What it does is break the page's shape.

A printed page is a rectangle of text, and the eye reads its edges before it reads a word. A single line marooned under a block of white space, or floating above one, registers as damage — the same way a hairline crack in a wall does. Readers who could not name the problem still feel that something is wrong with the book.

A widow at the top of a page is the worse of the two in practice, because a page turn is already a small interruption and landing on one orphaned line makes the reader feel they have missed something.

The rule is two

The convention is that at least two lines of any paragraph must stay together on each side of a page break. In CSS, which is what most modern book renderers actually use, it is exactly that:

p {
  orphans: 2;
  widows: 2;
}

Which reads as: never leave fewer than two lines at the bottom of a page, and never carry fewer than two lines to the top of the next one.

Some houses use three for a more generous setting. Two is the default almost everywhere and is what this site's print formatter applies.

What the renderer does with a paragraph it cannot break

It pushes. If a paragraph would leave one line stranded, the whole paragraph moves to the next page, and the current page ends a line short.

That is the trade, and it is worth being explicit about: you cannot have perfectly even page depths and no widows at the same time. One of them has to give. Book typesetting gives on page depth, because a page that is one line short is nearly invisible while a stranded line is not.

In a professionally set book you can find pages that differ from their neighbours by a line. That is not sloppiness — it is the widow control working.

Things that must not break at all

Some elements are indivisible, and the rule for them is stronger than "keep two lines together":

In CSS these get page-break-inside: avoid rather than a widow count.

None of this applies to your ebook

Widows and orphans are a property of fixed pages, and a reflowable ebook has no fixed pages. The reader picks the type size, the device picks the page height, and the text re-breaks every time either changes. There is no page for a line to be stranded at the bottom of, because that page did not exist until a moment ago and will not survive a font-size change.

Reading systems do apply widow and orphan control themselves as they paginate, and the properties are part of the ebook stylesheet, but the outcome is the device's to decide. You cannot control it and you should not try.

This is the general shape of the difference between the two formats, and it is worth internalising: in print you are deciding, in an ebook you are suggesting. Effort spent controlling ebook page breaks is effort wasted on a decision that was never yours.

What actually goes wrong

Almost every widow problem in a self-published book comes from the same source: the author hit Enter a few times to push text onto the next page.

Manual page breaks are absolute. They survive every later change, so the moment you add a paragraph in chapter three, every hand-placed break after it is in the wrong place — and now you have blank half-pages as well as widows.

Let the renderer break the pages. It knows the page height and can re-decide every break the moment anything changes, which is precisely what a hand-placed break cannot do.