Pet peeve: empty .notdef character.

André G. Isaak

I just spent some time browsing fonts on typekit and was surprised by the number of fonts I came across which had a blank .notdef character. What's worse, a few had blank, zero-width .notdef characters.

I just thought I’d use this forum to urge people to please include some sort of glyph in this slot. Be creative or just use a simple square. Just please don't leave it blank!

AbrahamLee

This makes me wonder, what’s the most creative .notdef glyph you’ve seen?

Florian Hardwig

Be creative or just use a simple square

In any case, make sure that the glyph stands out, in terms of typographic color. A .notdef glyph that looks too much like a regular character is almost as bad as an empty one.

Theunis de Jong

AbrahamLee: well, that'd be one of my own of course  

This is the default (and not changeable) .notdef for the quick-font InDesign utility IndyFont:


It always gets inserted in any font you create with it, and you cannot change it to something else (at least, not from within IndyFont). Thus, it doubles as a unique signature.

It's intentionally based on the logogram of IF itself, and with a background added to make it not look like any "standard" character.

The danger of having a sufficiently creative glyph is indeed that people will use it. That even goes for the blank nondescript square; I've seen those used as checkboxes way too often.

André G. Isaak

I don’t know if this is still the case, but the old type 1 fonts from Font Bureau all had a unique dingbat as a .notdef character (e.g. Aardvark had an aardvark). While this was certainly creative, I'd much rather the fonts have used a standard square and put the dingbat in the apple logo position where it could actually be used (assuming this wasn’t used for a FB logo of some sort).

Bhikkhu Pesala

In my opinion, a simple rectangle is unsuitable for a .notdef character. There is a glyph in Geometric Shapes that is a simple rectangle: U+25AF WHITE VERTICAL RECTANGLE and no doubt several other glyphs too. 

A rectangle with a cross could be mistaken for U+2612 BALLOT BOX WITH X

I suggest always using a rectangle with a question mark inside it. 

André G. Isaak

I suggest always using a rectangle with a question mark inside it. 

Even this, though, could be mistaken for some sort of ‘help’ character. I don’t know that it’s possible to prevent all users from mistaking the .notdef glyph for an actual character. (I mean unless you do something like the following)

James Puckett

I always use the box-with-an-x-inside design. It sees to be the most common design, so I assume that it is what people expect to see when proofing.

Arthur Reinders Folmer

This makes me wonder, what’s the most creative .notdef glyph you’ve seen?

I think there is a lot of fun to be had in the .notdef's. As long as it is (slightly clear) that something is missing, or that they actually give you another usable character instead. 

Not really to toot my own horn, but here are some of my solutions, looking for ways to merge the .notdef with the overall typeface concept.

For Fabel, a typeface based on animals, the .notdef is the Conservation status extinct symbol (since hey, the animal for your character isn't there!) This character was made to make it quite clear something is missing.



For my top-secret tinfoil hat dingbats the /notdef was a top secret stamp, since even if the typeface reveals crackpot theories, there will always be secrets! (this time, that's unsupported glyphs sadly) For this I chose to visualise the .notdef close to the rest of the icons. (since even if there is something wrong, it still is useful!)



More examples are welcome!

Chris Lozos

I always use a tombstone with a question mark reversed out of it as if to say "the glyph who
lives here has died and I don't know who it is"
  

LeMo aka PatternMan aka Frank E Blokland

My .notdef is named ‘Venus de Typo’. How helpless she is: she can’t even raise her arms to express that she does not have clue about the required character, nor can she point in any direction.

Kent Lew

I don’t know if this is still the case, but the old type 1 fonts from Font Bureau all had a unique dingbat as a .notdef character (e.g. Aardvark had an aardvark).

Hmm. As I recall, the old Font Bureau PS Type 1 fonts were generated with empty .notdefs. The unique dingbat glyph (known in-house as the “party” character) was drawn for the apple character slot (uniF8FF).

André G. Isaak

@Kent Lew

Turns out you are correct. I'd mistakenly thought this was the .notdef character because, while it may have been intended for the apple slot, they also copied it into every empty glyph slot in the codepage.

Chris Lozos

Typically, the foundry logo appears in what you call the Apple slot.  Apple put their own logo there in all the fonts they bundled with their system software.

André G. Isaak

Apple put their logo there, but they also assigned it a position on their standard  keyboard layout. I think that's what lead to its use as a logo-position catching on. (this, of course, was pre-unicode so I'm not calling it uniF8FF — that came later).

Khaled Hosny

Good thing is that many applications ignore the .notdef glyph in the font and just draw a hex box showing the hex number of the character.

Roel Nieskens

And when used as a web font, it will never show up.

Kent Lew

while it may have been intended for the apple slot, they also copied it into every empty glyph slot in the codepage.

André — True. I’d forgotten that when a designer hadn’t filled all slots, the party character was dumped in.

Johannes Neumeier

Roel Nieskens said:

And when used as a web font, it will never show up.

Interestingly enough, I recently had a situation where I explicitly wanted to show a webfont's .notdef character... is that a typoxymoron?

Kent Lew

Can you do it by using html entity to enter a non-existent code? Or maybe by entering the <NUL> code? For instance: �

(I went into HTML mode and typed "&#0000" to get that display, which I assume is a .notdef. But I’m not sure if it’s the .notdef specific to the Typedrawers webfont or from a fallback.)

André G. Isaak

That character appears to be uniFFFD (replacement character to replace unrepresentable character) rather than .notdef

John Hudson

I went into HTML mode and typed "&#0000" to get that display, which I assume is a .notdef.

.notdef is a glyph level convention, not a character encoding. Specifically, .notdef is whatever is GID 0 in a TTF or OTF.

U+0000 is the ASCII control character NULL, which was recommended to be GID 1 in older TTFs and normally represented by a zero-width, empty glyph. [Out of habit and having made fonts for Microsoft for so many years, I still build all my sources with the first three glyphs as .notdef, NULL (U+0000), and CR (U+000D), even though the latter two are not required anymore).

Peter Constable

I would recommend not using a glyph that is interesting enough that a user potentially might actually want to put into a document unless the glyph is also mapped by a PUA code point. But then, the reader is being set up for a conundrum: _Am I seeing a glyph an author wanted me to see, or something to indicate a character isn't being displayed?_ For that reason, I'd be inclined to stick with something that is commonly used in fonts, such as a simple thin-stroked box (distinct from the Unicode box symbol characters), or a box with a question mark or diagonals from opposite corners.

Claudio Piccinini

James Puckett said:

I always use the box-with-an-x-inside design. It sees to be the most common design, so I assume that it is what people expect to see when proofing.

I agree as it’s what I would expect most. I find a bit ambiguous the question mark solution, while doing a vertically proportioned rectangle with the cross inside makes it clear it’s not a checkbox mark or the similar and I find it’s the best solution.

Georg Seifert

This is what Glyphs puts in the font if you don't have a .notdef glyph:

Claudio Piccinini

Georg Seifert said:

This is what Glyphs puts in the font if you don't have a .notdef glyph:

That is also nice, but it’s in english alone. The "crossed out" rectangle is more universally recognizable, IMO, despite the language and it has immediacy as well.

Mark Simonson

It just needs to have <locl> support.

Claudio Piccinini

Mark Simonson said:

It just needs to have <locl> support.

But again, does the average user know what "glyph" means in typography?
In italian, “glifo" has a quite different main meaning, you should write "manca il carattere", which is too long and causes confusion, as "carattere [tipografico]" also means "typeface".

Mark Simonson

I agree that Glyphs' "NO GLYPH" solution is not great because it's in English (I always roll my own anyway), but it does have the virtue of being a character that no one would find useful.

(The <locl> idea was meant as a joke.)

Claudio Piccinini

Mark Simonson said:

I agree that Glyphs' "NO GLYPH" solution is not great because it's in English (I always roll my own anyway), but it does have the virtue of being a character that no one would find useful.

(The <locl> idea was meant as a joke.)

Ah, I did not get it was meant as a joke…
Well, the "no glyph" icon is nice, but looks almost like a logo.

André G. Isaak

Glyph's .notdef glyph might not be ideal, but I do think the approach of automatically adding a .notdef glyph if the designer hasn't supplied one is an excellent idea which other font software really ought to adopt.