published 2022 July | last updated 2022 September

Advanced Dashes

A photo of a restaurant storefront reading Supreme: New York Style Pizza-Liquor, but the dash between pizza and liquor could be mistaken for a hyphen
I love pizza-liquor, but I've never had it New York style.

There are three common dashes in written English:

The hyphen is used to join words, prefixes, and/or suffixes together, like father-in-law.

The en dash is used to create ranges, like 20202030.

The em dash can be used as a more powerful parenthetical, a replacement for a colon, or to represent an interruption of a speaker.

One Size Doesn't Fit All

A myth states the em dash is so named because it is "one M wide", and the en dash "one N wide". This is... almost true. The confusing correction is that an em dash is so named because it is 1em wide. Incidentally, the en is defined as half of one em; much of my focus will be on the em.

The Em

To talk about the em dash necessitates talk about the em. The em (pronounced /ɛm/) is a unit of length defined as the height of the character bounding box.1 This height changes with font family and size, so the length of the em changes with it: 1em is always one font-heighth long.

Specifying a font size in ems, then, is an exercise in ratios. The CSS declaration font-size: 1em is equivalent to font-size: 100%, while 1.5em is equivalent to 150% and so on; the current font size is used as a base to specify the desired font size.

The em's eccentric offering is that it can be used outside the context of font sizes. It can set margins, widths, even blur radii based on the font size of the containing context. It can scale an image up and down with the text around it, whether the text's changing size is a product of assistive technologies or simply of a component being reused in multiple contexts.

The em came about in the times of the printing press, when every M type was large and perfectly square, thus convenient to use as a proportional measurement for a given font and size. When the M type eventually rebelled against being square, the modern em shifted its definition to the remaining constant, the height. At the time this was effectively a non-change, but the updated definition allowed the M and the em to be decoupled.

The Em Dash

With the tailwind given by the definition of the em, the em dash is easily defined: a dash that is 1em long. In a way, the em dash took up the mantle of the square M type. Because the em dash is 1em long, and the em is defined as the height of the font, the em dash's bounding box is an exact squarein any font, at any size.

The Root Em

The rem or root em is a twist on the em. Where the em is defined using the current context's font size, the root em is defined using the root context's font size (in a web page, <html>). This allows escape from relativity hell without reverting to absolutes:

 1html: 16pt              (default     = 16pt)
 2  → body: 1.5em         (16pt * 1.5  = 24pt)
 3    → .container: .75em (24pt *  .75 = 18pt)
 4      → h1: 4em         (18pt * 4    = 72pt)
 5      → p: 1em          (18pt * 1    = 18pt)
 6    → footer: 0.5rem    (16pt * 0.5  =  8pt)
 7      
 8key:
 9context: requested size (calculation = final size)
10  → child element

Aside: The Point

It’s hard to define the em in terms of points without defining the point. The point has had a tumultuous and unstandardized history both physically and digitally, but today the world has predominantly settled on 1/72in per point. This varies greatly among displays, devices, and assistive technologies; for example, a 21” 1440p display has more pixels per inch than a 27” 1440p display. To wrangle the variability, there exist new unstandardized concepts like density-independent pixels on Android and effective pixels on Windows. The tumultuous history, as it happens, hasn't ended.

Based on the displays I develop on, I’ve settled on a rough mental framework of “1px of em length per pt” because it’s close enough to true and easy to remember: a 16pt font has a 16px em and is therefore 16px high.

Advanced Usage

Advanced usage of dashes unfortunately revolves around avoiding common issues.

Em Dash Overuse

The em dash tends to be a trap to a new writer. It is sympathetic to stream-of-consciousness writing, wherein the writer is in "append-only" modethinking not of the structure of the sentence but of what new words may clarify it when tacked on. For people whose writings spring from internal monologues this seems a neat way to avoid adding structure to organic thoughts.

Organic thoughts have their place in writing, but it turns out spending time perfecting sentence structure pays dividends to the reader. Long flowing thoughts connected by em dashes take mental effort to keep up with. Because the pre-dash thought may continue post-dash, the reader must keep its grammar and intention in an internal buffer, whichfor example, if the author decides to describe something tangential or dive into an example to assist with understanding, hoping their explanation can let the rest of the sentence off the hook but ultimately making the reader do more work to understand itcan be hard to do.

Separately, the em dash is a large glyph and hard to go unnoticed. Its use is obvious even in peripheral vision, so its overuse becomes a stain on the page before reading it.

Belittled Hyphens

When one side of a hyphen is a compound word like "frozen yogurt" its spaces can visually overpower the hyphen: pre-frozen yogurt looks like [pre-frozen] yogurt, not pre-[frozen yogurt].

One solution replaces spaces with hyphens (pre-frozen-yogurt), which may or may not suit the hyphenation. The opposite (pre frozen yogurt) is another tool in the bag, but the best it can do is shift ambiguity, not remove it.

\(\LaTeX\) users can replace each space with a \thinspace to de-emphasize them relative to hyphens, though the difference is subtle enough that you may also consider elongating nearby spaces:

1a pre-frozen\thinspace{}yogurt time
2a\enspace{}pre-frozen\thinspace{}yogurt\enspace{}time

\[\text{a pre-frozen\thinspace{}yogurt time}\]

\[\text{a\enspace{}pre-frozen\thinspace{}yogurt\enspace{}time}\]

Personally, the most common tool I reach for is to restructure the sentence to avoid the problem:

1a time before frozen yogurt

It's crude, but dependable.

Broken Dashes

If a line soft-wraps immediately before or after a dash, what should its behavior be? It's accepted that a hyphen can appear as the last character on a line, whether it was already there or is being introduced to join a word split across lines. En and em dashes are a little harder to parse.

1v left page edge             v right page edge
2|                            |
3I am looking at dates about 20
430 days out.
1|                            |
2I'm looking at dates about 20
330 days out.
1|                            |
2I am pleased they thought that
3however odd it may be.
1|                            |
2I'm pleased they thought that
3however odd it may be.

Strangely, the hyphen is the only dash that has an option here: the non-breaking hyphen or "hard hyphen" is a hyphen glyph that sends a signal to the word wrapper that a wrap must not happen after it.

Drowned Dashes

When using justified text elongated spaces can overpower dashes where non-elongated spaces would not have.

For example, see this simulation of left-aligned text:

1v left page edge             v right page edge
2|                            |
3I said they werewolf
4playdates don't last forever!

And this simulation of justified text:

1|                            |
2I    said    they   werewolf
3playdates don't last forever!

In the justified simulation, "werewolf" may read like the mythical creature "were-wolf" because the em dash looks short relative to the longer spaces.

A solution surrounds each parenthetical em dash with a space on either side, causing its total width to scale up at twice the rate of spaces:

1|                            |
2I   said  they  were    wolf
3playdates don't last forever!

Input Difficulties

Standard US keyboards support only the hyphen, so many en and em dash users simply compose them from hyphens and spaces.

No one should ever fault that, but for those who are willing to commit something new to memory:

For all others, read on.

The Hyphen-Powered En Dash

The en dash looks similar enough to the hyphen in most fonts that people often settle for a single hyphen: 2020-2030. \(\LaTeX\) requires two (--) for the en dash.

The Hyphen-Powered Em Dash

A common replacement for the em dash is a space on either end of two hyphens (--). Because \(\LaTeX\) reserves two hyphens for the en dash, it requires three (---) for the em.

Output Difficulties

At the time of writing, twos.dev renders in a monospaced font for aesthetic reasons. Unfortunately, the glyphs for many dashes are virtually indistinguishable in monospaced fonts. To make the differences visible, I wrote a bespoke preprocessing step to render twos.dev's dashes in a variable width font.

I'd love for there to be a "monospace with exceptions" font that takes this chore out of my hands.

Bonus Round: Esoteric Dashes

The hyphen, en dash, and em dash get all the love, but behind the scenes are a silent majority of dashes that don't often get to see the light of day.

The swung dash () is an elongated tilde used to stand in for a word being defined in a dictionary.2

boot (n)

ex: Let me put on my other .

The horizontal bar is a way to introduce quotations. Confusingly, its length is almost always identical to the em dash.

O Miss Douce! Miss Kennedy protested. You horrid thing!

James Joyce's Ulysses p. 335

The hyphen bullet is a hyphen to be used in place of a bullet point.

1- This is a hyphen
2 This is a hyphen bullet

The figure dash is a variant of the en dash having the same width as digits (which are uniformly wide in most fonts). It is meant for phone numbers and other numeric contexts where columnar alignment is required or pleasing.

1Figure dash: 867‒5309
2Number:      86705309
3En dash:     8675309
4Hyphen:      867-5309

Lastly, my favorite: the soft hyphen is a zero-width, invisible character that (opposite to the hard hyphen) denotes a place the word wrapper is welcome to wrap. This can be used in the middle of a compound word or long line of inert code to provide a cleaner wrap.

1v left page edge            v right page edge
2|                           |
3No soft hyphen:
4Supercalifragilisticexpialid-
5ocious
6
7Soft hyphen after 'istic':
8Supercalifragilistic-
9expialidocious

How dashing!


  1. "Bounding box" is strong; font designers are free to design outside it.
  2. http://wordnetweb.princeton.edu/perl/webwn?s=swung+dash