What is the difference between a hyphen, an en dash and an em dash?

A hyphen joins: compound words, and words broken at the end of a line. An en dash, the width of a lowercase n, spans: ranges of numbers, dates and pages, and pairs of names. An em dash, the width of a capital M, interrupts: it breaks a sentence where a comma is too weak and parentheses too strong. Only the last two are dashes; the hyphen is a joiner that happens to be short and straight.

Should there be spaces around an em dash?

In American practice, no. Chicago sets the em dash closed up against the words on either side. British practice commonly uses a spaced en dash instead, which does the same job with a shorter mark and air around it. Both are correct within their own convention; using an unspaced em dash in one paragraph and a spaced en dash in the next is not.

How do I type an em dash?

On a Mac, Shift-Option-hyphen for an em dash and Option-hyphen for an en dash. On Windows, Alt-0151 and Alt-0150, or Ctrl-Alt-minus and Ctrl-minus on the numeric keypad in Word. What you should not do is type two hyphens and hope: that is a typewriter workaround, and if autocorrect does not convert it, it prints as two hyphens.

When do I use an en dash?

For spans and connections between equal things. Number ranges: pages 45–60, 1939–45. Pairs of names where both stand separately: the Bose–Einstein condensate, the London–Edinburgh train. The test is whether you could say 'to' or 'and' in its place. If you could say 'and' between two words that form a single idea, you want a hyphen instead.

Is a hyphen acceptable instead of an en dash in a range?

It is common and it is wrong in book typesetting. A hyphen in a range is visibly too short and reads as a compound rather than a span, so 1939-45 looks like a single hyphenated term. In printed books the en dash is the expected mark, and it is one of the details a copy editor changes globally.

What about the long dash used for a repeated author name or a redaction?

That is a 2-em or 3-em dash, a separate mark. Chicago uses a 3-em dash in bibliographies for a repeated author's name, and a 2-em dash to stand for omitted letters or a missing word. Neither belongs in ordinary prose, and neither is produced by typing several em dashes in a row.

Hyphens, en dashes and em dashes in a book

The three marks are different lengths, which is the only thing most people know about them, and it is the least useful thing. Length is how you recognise them. What separates them is that they do three jobs which have nothing to do with each other.

One joins. One spans. One interrupts.

The hyphen joins

- — the key on the keyboard.

It builds compound words: well-known, mother-in-law, twenty-three. It also breaks words at the end of a line, which is the other half of its job and the reason hyphenation settings exist at all.

The hyphen is not a dash. It is a connector that is short and straight, and it is the only one of the three that a keyboard gives you directly. That accident is why manuscripts are full of hyphens standing in for the other two.

The en dash spans

– — the width of a lowercase n.

It sits between two things and means to or and:

Ranges: pages 45–60, 1939–45, Monday–Friday
Connections between things that remain separate: the Bose–Einstein condensate, the London–Edinburgh train, a Labour–Liberal coalition

The test is whether you can substitute to or and without the sense changing. If you can, it is an en dash. If the two words form one idea rather than a pair, it is a hyphen: a well-known author is not well and known.

A hyphen used in a range is the most common dash error in manuscripts, and in a printed book it is visible. 1939-45 reads as a hyphenated compound rather than a span, because the mark is too short to look like a bridge between two numbers.

The em dash interrupts

— — the width of a capital M.

It breaks a sentence. Butterick’s description of when to reach for it is the useful one: where a comma is too weak, but a colon, semicolon or parentheses are too strong.

Two of them work as a stronger pair of parentheses, setting off an aside without closing it away. One on its own throws the sentence in a new direction.

In dialogue it does a specific job that no other mark does, which is cutting a line off mid-utterance, covered in formatting dialogue.

The em dash is also the mark most easily overused. A page with four of them stops reading as interruption and starts reading as a habit, because the effect depends on being unexpected.

Spaced or unspaced

This is a genuine fork between conventions rather than a right answer.

American practice, following Chicago, closes the em dash up against the words on either side, with no spaces—like this.

British practice more commonly uses a spaced en dash – like this – for the same job. Shorter mark, air around it.

Bringhurst prefers the spaced en dash generally. Butterick notes that this is not common in American typography, and allows spaces around an em dash when it looks crushed, particularly on screen.

Pick one by where your readers are, then apply it to the whole book. The failure is not choosing wrongly; it is choosing twice.

Ranges keep the en dash closed up in both conventions: 45–60, never 45 – 60.

The two-hyphen habit

Typing -- for a dash is a typewriter workaround that survived into word processors because autocorrect usually converts it. Usually is the problem. Paste from a plain-text file, an email, or a writing app with the substitution turned off, and the two hyphens travel into the typeset book and print as two hyphens.

Nueel converts -- to an em dash and - between numbers to an en dash as you type, which handles the common cases. The one to check by eye is imported text, where the substitution never ran: a search for a double hyphen across a manuscript takes a second and usually finds something.

If you keep only one thing: a hyphen joins words, an en dash joins numbers, an em dash breaks sentences. The lengths follow from the jobs, not the other way round.

Sources

Matthew Butterick, Hyphens and dashes in Practical Typography — the em dash as the mark between a comma and parentheses, and the note on Bringhurst’s spaced en dash
The Chicago Manual of Style, Q&A on hyphens, en dashes and em dashes and Hyphens and dashes: a refresher — closed-up dashes, en dashes in ranges, and the 2-em and 3-em dashes

Common questions

What is the difference between a hyphen, an en dash and an em dash?: A hyphen joins: compound words, and words broken at the end of a line. An en dash, the width of a lowercase n, spans: ranges of numbers, dates and pages, and pairs of names. An em dash, the width of a capital M, interrupts: it breaks a sentence where a comma is too weak and parentheses too strong. Only the last two are dashes; the hyphen is a joiner that happens to be short and straight.
Should there be spaces around an em dash?: In American practice, no. Chicago sets the em dash closed up against the words on either side. British practice commonly uses a spaced en dash instead, which does the same job with a shorter mark and air around it. Both are correct within their own convention; using an unspaced em dash in one paragraph and a spaced en dash in the next is not.
How do I type an em dash?: On a Mac, Shift-Option-hyphen for an em dash and Option-hyphen for an en dash. On Windows, Alt-0151 and Alt-0150, or Ctrl-Alt-minus and Ctrl-minus on the numeric keypad in Word. What you should not do is type two hyphens and hope: that is a typewriter workaround, and if autocorrect does not convert it, it prints as two hyphens.
When do I use an en dash?: For spans and connections between equal things. Number ranges: pages 45–60, 1939–45. Pairs of names where both stand separately: the Bose–Einstein condensate, the London–Edinburgh train. The test is whether you could say 'to' or 'and' in its place. If you could say 'and' between two words that form a single idea, you want a hyphen instead.
Is a hyphen acceptable instead of an en dash in a range?: It is common and it is wrong in book typesetting. A hyphen in a range is visibly too short and reads as a compound rather than a span, so 1939-45 looks like a single hyphenated term. In printed books the en dash is the expected mark, and it is one of the details a copy editor changes globally.
What about the long dash used for a repeated author name or a redaction?: That is a 2-em or 3-em dash, a separate mark. Chicago uses a 3-em dash in bibliographies for a repeated author's name, and a 2-em dash to stand for omitted letters or a missing word. Neither belongs in ordinary prose, and neither is produced by typing several em dashes in a row.