What is Tofu?

several blocks of real tofu readily prepared to be cooked
Photo by Taylor Kiser on Unsplash

You might know those prevalent white blocks of coagulated soy milk called tofu from cooking. But did you know that this is also a typographical technical term?

Tofu on the screen

When a computer decodes text for display it searches for a glyph from an installed font for each character. It is possible, though, that no supporting font is installed on the system for a given character. In this case the software reaches for a special glyph in the default font named .notdef. Sometimes this fallback glyph is just a simple white square.

Now, if you want to read text in a language, where the glyphs are not present on your computer, the display looks something like this:

□□□□□ □□□□□ □□□□□ □□□ □□□□, □□□□□□□□□□□□ □□□□□□□□□□ □□□□.

This has the appearance of blocks of literal tofu lying next to each other. In Japan, where the term “tofu” for this phenomenon was initially coined, the problem was especially rampant. Software being written with only one character set in mind could easily become unusable when fed with text in a different encoding, even if it knew how to switch between them. This might not be too large an issue for a European, when their umlauts are suddenly missing. But if two completely different writing systems have to battle for the font stack you easily end up with a whole screen filled only with white boxes.

The proof of the tofu is in the eating

Today operating systems come with better internationalized base fonts pre-installed. And to end the phenomenon once and for all an initiative started by Google and aided by Adobe wants to provide a font for each character in the Unicode standard, the Noto font family.

The name “Noto” is actually an abbreviation of the font’s aim: “No Tofu”. Together with GNU Unifont this font family also powers the character display on codepoints.net almost exclusively.

In day-to-day use tofu is luckily becoming less of an issue than it was in the 90’s. If you stumble upon it nowadays, it is mostly due to new emojis being codified by Unicode, but not being implemented in your platform’s emoji font.

Another situation where you might encounter a variant of tofu today is, when a web font hasn’t got all glyphs to render a specific word. The browser will then substitute the missing characters with glyphs from other fonts, which gives the text sometimes the impression of a ransom letter:

naïve résumé


The earliest occurrences of the term “tofu” with the meaning of garbled text are lost in the mist of time. The Noto Font project started in 2014, and most search results (including for the Japanese term “豆腐”) date from the following years. In 2014 it was added to the Wiktionary and started to gain widespread attention.

On the Unicode mailing list it appeared for the first time in 2009, while Chris Lilley mentioned it on a W3C mailing list as early as May 2000. This suggests that at this time the term must have been in regular use in Japan for some time.

The phenomenon itself seems to be known and named at least in 1990, where according to this answer Japanese users coined the term “hexagana” for substitution glyphs that showed the hexadecimal code for an unknown character instead of a simple empty box.

How to Cook with Tofu?

Having text unreadable because of missing glyphs is still the lesser of two evils that regularly happen to electronic text. At least it can be made readable again by finding and installing a suitable font.

If the content is mangled due to wrongly specified encodings, though, it might be irreversibly damaged. This phenomenon is called Mojibake. It appears, when the changing and/or interpretation of font encoding went wrong. You might have encountered a strange “ä” instead of an “ä”. Fixing this problem is often extremely tedious.

But even tofu itself is not all made equal. The Fake Unicode twitter account detailed different ways fonts can handle missing glyphs, from terrible (“show nothing at all”) to actually somewhat useful (“show the hex code of the unknown character”). If you want to know, whether a brand-new emoji will show as tofu on Android or iOS, you can consult Is It Tofu? for an answer (including a rough percentage of users that will see the emoji).

So, all is not lost, if you encounter these blocks again in the future. And to make use of them even if they are illegible on your system, try pasting them in the codepoints.net search field to find out, what your tofu is actually made of.


With special thanks to Andrew Cunningham, Chris Lilley, and Richard Ishida for their input to this post!