Smart quotes and dashes: the punctuation of a real book

A typeset book uses curly quotation marks, a real apostrophe, en dashes in number ranges and em dashes for parenthetical breaks — none of which are the characters on your keyboard. Write with straight quotes and hyphens if you like, but the export has to convert them, because straight quotes in a printed book are the single fastest way to look self-published.

There is a set of marks that separates a page that looks typeset from one that looks typed, and almost none of them are on a keyboard. A reader will not be able to name what is wrong with a page of straight quotes and hyphens-used-as-dashes. They will just find it slightly cheap, in a way they attribute to the writing.

The marks that are not on your keyboard

Six characters do most of the work. The left column is what a keyboard gives you; the right is what a book actually uses.

You typeA book printsNameUsed for
"“ ”Curly double quotesSpeech and quotation
'‘ ’Curly single quotesQuotes inside quotes
'Right single quoteEvery apostrophe, without exception
-En dashNumber and date ranges
--Em dashA break in a sentence
...EllipsisAn omission or a trailing off

Quotes: straight, curly and the wrong way round

Straight quotes are a typewriter artefact. A typewriter had one key and no room for four glyphs, so it used a vertical mark that stood in for opening quote, closing quote, apostrophe, prime and inch mark. Every one of those is a different character in real typesetting, and books have used the distinction for four hundred years.

Converting them is nearly always automatic — the renderer looks at what is on either side of the mark and picks an opening or a closing form. The place it gets this wrong is predictable, and it is the reason to check rather than trust:

  • Words that start with an apostrophe. 'Twas, 'tis, 'em, '90s and rock 'n' roll all begin with an omission, so they need a closing quote — ’ — where an automatic converter will happily give you an opening one.
  • Feet and inches. 6′ 2″ uses prime and double prime, which are upright and slanted, not curly quotes. A house described as 30' wide with a curly quote is wrong twice.
  • Quotes inside quotes. English nests double outside, single inside. Two adjacent closing marks — ’” — get a hair of space between them in careful typesetting, and nothing at all in most.
  • Dialogue running over paragraphs. A speech continuing into a new paragraph opens with a quote mark and does not close the previous one. Converters handle this correctly; writers often do not.

If your converter does the substitution, the way to check is not to read the manuscript — it is to export a page with dialogue on it and look at the PDF at 200%. Wrong-facing apostrophes are invisible at reading size and obvious at that magnification.

Three dashes, three jobs

The hyphen, the en dash and the em dash are different characters with different widths and different meanings, and using one for another is the second most common giveaway after straight quotes.

MarkWidthJobExample
Hyphen -NarrowJoins wordsself-published, twenty-one
En dash –Width of an nRanges, and "to"pages 12–18, the London–Paris train
Em dash —Width of an mA break in a sentenceIt worked — eventually

The rule for ranges is worth learning because it is unambiguous: anything you could read as "to" or "through" takes an en dash, not a hyphen. Pages 12–18. Monday–Friday. 1939–1945. A hyphen there is not a stylistic choice, it is a typo that survived.

The em dash has one genuine style decision attached to it. American practice sets it tight — like this — with no spaces. British practice more often uses a spaced en dash – like this – instead. Both are correct. Mixing them in one book is not, and mixing them is what happens when a manuscript is assembled from several drafts.

Ellipses, and the space around them

Three full stops in a row is not an ellipsis; it is three full stops, and a justified line will happily stretch the gaps between them until they read as four. The single ellipsis character keeps its spacing under justification, which is the whole reason it exists.

  • Inside a sentence, an ellipsis usually takes a space on both sides: she paused … and went on.
  • At the end of a sentence that trails off, it takes the ellipsis and then the full stop, if the sentence was complete.
  • Never four dots in a row typed by hand. If you need an ellipsis and a full stop, that is one of each, in that order.
  • Never let one break across a line. A converter that treats the character as one glyph handles this for free; three periods can and will split.

Where straight marks must stay straight

There is one category of text where every substitution above is a bug, and it is the one that most often gets ruined: anything a reader is expected to type back in.

const greeting = "hello"   // straight quotes — this runs
const broken   = “hello”   // curly quotes — this is a syntax error

Code, file paths, terminal commands, URLs, regular expressions and configuration files must come out of the export with exactly the characters that went in. A renderer that applies smart punctuation inside a fenced block has produced a technical book whose examples do not run — which is why code blocks are handled as a separate kind of content rather than as text that happens to be monospaced.

The same goes for inline code spans. --flag in a sentence about a command line must stay two hyphens; converted to an em dash it becomes a flag nobody can type. If you are writing anything technical, this is the first thing to test in an export, before typefaces and margins.

Fixing a manuscript you have already written

  1. Decide the dash style first. Tight em dash or spaced en dash. Everything else follows from it.
  2. Search for ` - ` — a hyphen with a space on each side. In prose it is almost always a dash that was never converted.
  3. Search for `--`. Two hyphens is a typewriter em dash and should be a real one, except inside code.
  4. Search for `...` and replace with the ellipsis character, outside code blocks.
  5. Search for a straight apostrophe followed by a letter at the start of a word: 't, 's, 'n, '9. These are the ones an automatic converter turns the wrong way.
  6. Export and read the first page at 200%. Quotes, apostrophes, dashes. It takes a minute and it is the only check that sees the real output.

None of this is worth doing three times, which is the argument for doing it once at the end rather than as you write. Punctuation is the last pass, after the words have stopped moving — the same place in the sequence as checking for widows and orphans, and for the same reason: both are properties of the finished page, not of the draft.

It also puts a floor under everything else you have decided. A book with the right typeface, sensible margins and straight quotes still reads as amateur, and the reader will not know why. The rest of the typographic basics are worth having next to this one.

Read next

Design · 5 min read

Book typography for people who are not typographers

Most of the gap between an amateur page and a professional one comes down to six things: line length, type size, line spacing, margins, how many typefaces you use, and whether the heading sizes form a clear hierarchy. None of them require taste to get right — they have known good ranges.

Read it
Design · 5 min read

Widows, orphans and the page breaks that spoil a book

A widow is the last line of a paragraph stranded alone at the top of a page. An orphan is the first line of a paragraph left alone at the bottom. Both are bad breaks, and so are a heading at the foot of a page, a split table and a broken code block. Every one of them is fixed the same way: give the typesetter permission to move the break, or edit the text until the break lands somewhere harmless.

Read it
How-to · 6 min read

Code blocks in a PDF: monospace, long lines and the page edge

Use fenced blocks, keep every line under about 80 characters, and let the typesetter hold each block on one page. Monospace type is wider than body type at the same size, so a listing fits fewer characters on a line than prose does — the fix is shorter lines in the source, not smaller type in the PDF.

Read it

Write it in Markdown. Ship a book.

bukpress turns the file you are already writing into a typeset PDF and EPUB — your type, your covers, no watermark on any plan.

Start free — no card required