En dash, em dash, hyphen: which one, and when

No spell-checker has ever flagged a hyphen standing in for an en dash. No reviewer has ever failed to notice. Dashes are the clearest example of a class of error that is invisible to software, invisible to the author who has read the sentence forty times, and immediately visible to a copyeditor — which is why we wrote a macro for it, and why it is worth ten minutes of your attention now.

There are four marks. Three are punctuation and one is mathematics.

The four marks

The hyphen (-) joins. It builds compound words and compound modifiers, and it divides words at a line break. It is the only one of the four on a standard keyboard, which is the root of the entire problem.

The en dash (–), the width of a lowercase n, spans. It sits between the two ends of a range, and it joins elements that are parallel rather than subordinate.

The em dash (—), the width of a capital M, interrupts. It sets off a parenthetical break in a sentence — like this one — or introduces an abrupt turn at the end of a clause.

The minus sign (−) is not a dash at all. It is a mathematical operator with its own Unicode point (U+2212), designed to sit at the same height and width as the horizontal stroke of a plus sign. A hyphen used as a minus sign is a typographic error that becomes conspicuous the moment the manuscript is typeset.

Where the en dash actually belongs

Ranges are the obvious case and the one most often got wrong: 2014–2019, pp. 231–248, Figures 1–6, 25–30 mg. The en dash here means to or through.

That meaning creates the most common range error in submitted manuscripts. If you have already written the word from or between, the dash is redundant and wrong. Write from 2014 to 2019, not from 2014–2019. Write between 25 and 30 mg, not between 25–30 mg. Pick the words or pick the dash; never both.

The less obvious case is the compound adjective whose parts are not single words. A hyphen joins single words; when one side of the compound is itself an open or hyphenated term, an en dash carries the join across the whole unit:

• post–World War II reconstruction — the prefix attaches to all three words, not just to World
• Nobel Prize–winning chemist — the prize is not Prize-winning
• New York–based laboratory

Several style guides also use the en dash between elements of equal weight, where a hyphen would imply that one modifies the other: dose–response curve, mother–daughter pairs, cost–benefit analysis, the Michaelis–Menten equation. That last pattern matters in scientific writing more than anywhere else, because it distinguishes two named people from one hyphenated surname. Michaelis–Menten with an en dash is two researchers. With a hyphen it reads as one person with a double-barrelled name.

Where the em dash belongs, and how to space it

The em dash marks a break in thought sharper than a comma and less formal than parentheses. Use it in pairs for an interruption in the middle of a sentence, or singly to introduce a turn at the end. Two per sentence is a maximum; more than two in a paragraph and the prose starts to feel breathless, which is a judgment call an editor will raise as a query rather than a correction.

Spacing is where the style guides genuinely diverge, and where a manuscript most often becomes internally inconsistent:

• Chicago and APA set the em dash closed up — no spaces on either side.
• AP sets it with a space on each side.
• Many British and Commonwealth house styles avoid the em dash altogether and use a spaced en dash for the same job — like this.

None of these is more correct than the others. What is incorrect is doing two of them in one manuscript, which is the state most drafts arrive in, because different sections were written at different times or by different co-authors.

The errors that survive every check

These are, in order, what we find most often when a manuscript reaches the desk:

• Hyphens standing in for en dashes in ranges. Overwhelmingly the most common. Word's autocorrect converts some of them and not others, so the same paper contains both.
• Double hyphens (--) left as an em dash placeholder. A typewriter convention that survives in plain-text drafts, LaTeX source, and anything pasted out of a code editor.
• Hyphens used as minus signs in inline maths and in negative values in tables — -3.4 rather than −3.4.
• Inconsistent em-dash spacing between sections, as above.
• A hyphen in a compound that should be open, or an open compound that should be hyphenated. The general rule: hyphenate a compound modifier before the noun (a well-known result), leave it open after (the result is well known). Adverbs ending in -ly never take the hyphen: a highly cited paper, not a highly-cited paper.
• Line-break hyphens baked into the text after a copy-paste out of a PDF, leaving meth- odology mid-sentence.

Getting them into the file

Nobody should be hand-typing Unicode. The practical routes:

• Word: en dash is Ctrl+Minus on the numeric keypad, em dash is Ctrl+Alt+Minus. Autoformat will convert some hyphens as you type, which is precisely why the results are inconsistent — it fires on some patterns and not others.
• macOS, anywhere: Option+Hyphen for an en dash, Option+Shift+Hyphen for an em dash.
• LaTeX: -- produces an en dash, --- an em dash, and a single - in maths mode is already a proper minus. This is one of the few places the plain-text convention is correct rather than a placeholder.
• HTML: – and —, or the literal characters in a UTF-8 file.

A find-and-replace pass you can run yourself

If you want to clean a draft before it goes anywhere, this order catches most of it in about five minutes. Work on a copy, and review each change rather than replacing all — every one of these has legitimate exceptions.

• Search for a digit, hyphen, digit (([0-9])-([0-9]) with wildcards on) and convert the hyphen to an en dash. Check for phone numbers, ISBNs, and gene names before accepting.
• Search for -- and decide, case by case, whether each is an em dash or a range.
• Search for from followed by an en dash within a few words, and for between the same way. Fix the construction, not the dash.
• Search for a space, hyphen, space ( - ). This is almost always an em dash that never happened.
• In tables and equations, search for a hyphen preceded by a space or a cell boundary and followed by a digit. Those are minus signs.

Then read the compound modifiers, which no search will find for you.

Why we automated it

Everything above is mechanical. It is a set of rules with clear inputs, a handful of documented exceptions, and no judgment worth a human's attention — which makes it exactly the kind of work that should be done by a program before an editor opens the file. Our sweep for this runs in a couple of seconds on a full manuscript, normalises ranges and spacing to the target journal's house style, flags the compounds it is not sure about rather than changing them, and returns every intervention as a tracked change for the author to accept or reject.

What it does not do is decide whether your sentence needed the interruption in the first place. That is still an editor's call, and it always will be.