Non-breaking hyphen, U+2011

A hyphen that forbids a line break at its position. It is visually identical to the ASCII hyphen you type, which is what makes it interesting: you cannot see it, and most cleaning tools do not know about it.

How it works

e2 80 91. It draws exactly like the hyphen-minus on your keyboard, and differs only in its line-break class, which forbids a break at that position. Nothing about the rendered glyph tells you which one you are looking at, which is what makes it more troublesome than the characters people already know about. Every layer beneath rendering treats it as a different character: a search for a hyphenated term misses it, a part number or product code fails to match a catalogue, a slug generator produces a different slug, and a diff reports a change on a line that reads identically. It also survives cleaning tools that handle the em and en dashes, because those tools were written against the characters people had heard of. We found five of them in a single Microsoft Copilot reply, in ordinary compounds, by running a code-point census rather than reading the text.

What it is legitimately for

Keeping a hyphenated term or a part number together across a line break.

What it breaks

• A search for a hyphenated term misses every instance written with this character.
• A part number or product code fails to match your catalogue.
• A URL slug generated from a heading containing one produces a 404.
• A diff shows a change on a line that reads identically.

How to find it yourself

Look for e2 80 91 in a hex dump. You will not spot it by eye: it looks exactly like a hyphen.

What Clipboard Sanitizer does

Replaced with an ASCII hyphen.