Design of @

Bendy's picture

So I got to thinking that the at sign really shouldn't be aligned with the uppercase, since its primary use is in e-mail addresses, which are always correctly written in lowercase.

MS Character Design Standards however says:
Height alignment : Visually centers on the uppercase height, dependent on the typeface design.

Spacing : Visually center between uppercase H and O.

As an aside, this doesn't help me know what size to make the glyph and so I'm not fitting its upper and lower extrema to any particular heights. Instead its top is around three-quarters of the distance between the x line and the cap line, and the bottom around the same between the baseline and the descender line. This leaves the /a/ part floating in space around the centre of the x-height, which looks fine to me. Any thoughts?

PS not sure what that flat curve was doing in there, it's been eliminated now ;)

.00's picture

Unless you are going to use it as originally intended: 25 Bails of Hay @ $.30 per Bail, you should probably align it so it works with the lowercase. Then you can make another one, name it at.case and have it work with the uppercase

theplatypus's picture

I had a similar issue a number of years back. Your optical alignment to the lowercase works for me!
best regards,
daniel

nina's picture

Yours looks good to me judging from this – mine is similarly placed FWIW.

Old usage definitely wanted it higher IIRC, which is why it looks funny when you print e-mail addresses in letterpress, using old fonts.

Bendy's picture

Rather than create a separate at.case glyph, wouldn't it be more efficient to simply gpos this one up a bit?

eliason's picture

which is why it looks funny when you print e-mail addresses in letterpress, using old fonts

besides the fact that it would often be a generic sort rather than part of a font, no?

kentlew's picture

Bendy -- One disadvantage of gpos-ing the existing at glyph is that you can't provide a separate set of kern pair values, which you may or may not want.

.00's picture

Not only is the kerning likely to change for the .case version of the at glyph, but the glyph itself may need a bit of weight as well. Or you can use an uppercase A form inside the ring for the .case at.

Bendy - from my point of view, the weight of the a portion of the at glyph you show seems a tad light and the ring appears a tad heavy. FWIW

nina's picture

Craig – good point.

"Or you can use an uppercase A form inside the ring for the .case at."
Hey cool. I've never seen this. Do you happen to have an example?

Bendy's picture

James, yes, I think you're right about the relative weights. I was thinking the whole glyph should be a slightly bit larger to make the a part darker, but I think instead I'll make the ring lighter. Thanks.

To me, a capital A inside would look pretty alien. I think I've seen it somewhere but can't remember where.

Here are some interesting ideas I've just quickly found:

hrant's picture

Forget UC. And the alignment you have up there looks solid.

As for what's inside the circle, for a text face I'm a very
firm believer that it should NOT look like an "a" (or "A").

hhp

nina's picture

"should NOT look like an “a” (or “A”)"

What, because of confusability?
(Seems potentially tricky in Italics, or other fonts with monocular "a"s.)

Bendy's picture

Agreed, that's why they're usually italic in a roman face, is it?

>Seems potentially tricky in Italics, or other fonts with monocular “a”s
That's where the creativity has to come in. You wouldn't want type design to be simple!

hrant's picture

Nina, because you don't want people to see an "a", you want them
to see an "@" (like how you don't want an ampersand to look like
two joined letters). In an Italic, I'd try to make it more symbolic.

hhp

Jongseong's picture

Hrant, your point about the ampersand makes sense, and you could add other examples: a question mark should not look like the letter "q", and an eszett should not look like a long s and an s or a z merely joined together. I personally don't like the W made of two Vs simply overlaid either. In other words, we shouldn't be taking the etymological origins of symbols and letters literally in designing them.

But to me "@" has always been nothing more than an "a" with a circular swash, and the statement that you want people to see an "@" instead of an "a" makes as much sense to me as wanting people to see a "©" instead of a "c" in a circle.

Is it just me? Are there people who think of "@" as an independent symbol with nothing to do with "a"? Come to think of it, I see how people who have only ever typed "@" and never wrote it by hand may not ever make the connection between "a" and "@".

eliason's picture

because you don’t want people to see an “a”, you want them
to see an “@” (like how you don’t want an ampersand to look like
two joined letters)

In the case of the ampersand (for English readers like me), the visual 'et' and the semantic 'and' are unrelated, but for the at-sign the meaning (or better, the phonetic component - whether read aloud or heard "in the mind's ear") does start with "a" so I don't see a problem with a-like @'s.

Which makes me wonder if French typefaces are more likely to have "et" ampersands.

hrant's picture

Brian, for any symbol we must consider what it does, not where it comes from. The "a" component of "@" should be actively discarded because all promoting that can do is confuse people; seeing an "a" in there is much more likely to do harm than good. Knowing that the "@" comes from writing "at" quickly is just cute trivia, not good design.

This reasoning is also what I apply in demoting chirography in type design.

hhp

Jongseong's picture

Hrant, this must be a difference of how we think of the "@" sign, because I think an "@" where the central component is stylized beyond recognition as an "a" will be more confusing to me.

Maybe this comes down to the fact that I'm more used to glyphs that may be confused with an overly-stylized "@". Koreans (and Japanese, I think) often make use of circled letters, and Thai numerals or Japanese hiragana contain many symbols containing a circular-swash-like element. So to me, what distinguishes "@" from an array of other similar symbols—indeed, the most characteristic trait—is the central "a" component.

I remember seeing other Koreans writing "@" as simply a circled "a". This is nothing more than anecdotal evidence, of course, but it shows that to some people the connection between "@" and "a" is really strong.

The situation may indeed be different for those who are not likely to use any other symbol that can be confused for the "@".

hrant's picture

> to some people the connection between “@” and “a” is really strong.

Indeed.
It's our duty to break it. :-)

hhp

quadibloc's picture

I would think that the at-sign @, in order to look like @, needs to look like a single story a, with a tail that goes around it. Yes, it needs to be very distinct from the lower-case letter a, and from a circled A, whether upper- or lower- case, but it also needs to not be an unfamiliar symbol which readers will not recognize at sight.

Of course, on the other hand, & would only look like an et ligature in a typeface which is patterned after a historical typeface or group of typefaces in which that was the case.

Jongseong's picture

Hrant, while I agree with the principle behind your suggestion, is there anything inherently unnatural about thinking of "@" as a swashy "a" symbol? Any more than thinking of "©" as "c" in a circle, for example? It's a convenient and intuitive way to think about the symbol, even for many non-English-speakers who have no idea that it is called the "at sign" in English. Take away the connection and we are left with a random scribbly hieroglyph that we'll never remember how to write. ;)

I'm also curious how one would move the design of the "@" away from the "a" component and still make it recognizable as an "@" to those who rely on the "a" component to distinguish it from similar symbols.

Just to make clear, none of the atypical "@" treatments that Bendy found (from Roadway, Jana Thork, Perla, and Letunical) look like "@" to me. Just because in my mind "@" is inextricably connected to "a" doesn't mean that just any "a"/"A" with a circular swash makes an "@". But Bendy's own "@" design to me is a perfect example of what an "@" should be shaped like. I'm curious how one would improve upon it.

hrant's picture

> in order to look like @, needs to look like a
> single story a, with a tail that goes around it.

What about in fonts with a monocular "a" (like most Italics)?

hhp

hrant's picture

Brian, not "unnatural", but not-ideally-functional.

> still make it recognizable as an “@” to those who rely on
> the “a” component to distinguish it from similar symbols.

Like what?

And always, think about the context: the "@" sign's main role now is email addresses. when you see it in a tightly-packed string of alphanumerics, what are the chances that it's going to be confused for the copyright symbol, versus an "a"?

For example if you see "nin@gmail.com" and the "@" has a honking "a" in it, you're much more likely to incorrectly see it as "nina@gmail.com".

> we’ll never remember how to write.

1) How often do we write [for other people] these days?
2) I myself have a hell of a time writing a "proper" "at" sign. It takes too long, I rush it, it comes out horrible. I'm better off writing something simpler, like a simple spiral.

> The circular swash means the “@” will
> never, ever be confused with an “a”.

Never say never!
When you're in a hurry you have accidents. (But you still hurry.)

hhp

Jongseong's picture

Hrant: What about in fonts with a monocular “a” (like most Italics)?

To me, that is nothing to worry about, at least no more than one would worry about the "d" looking too much like the "a" when you take away the ascender. The circular swash means the "@" will never, ever be confused with an "a". The focus should be on differentiating it from other circular-swashy characters, which admittedly isn't much of a problem for most westerners.

Jongseong's picture

Crossposted...

Hrant: For example if you see “nin@gmail.com” and the “@” has a honking “a” in it, you’re much more likely to incorrectly remember it as “nina@gmail.com”.

Thanks for this example. This makes much more sense.

I think we will agree that in the context of an email address, anything looking remotely like the "@" will immediately be recognized as such. But I still contend the expected shape for the "@" contains the "a" element, so that an isolated "@" without the "a" component may be tricky for some to recognize.

Some similar symbols, off the top of my head, include the Thai numeral one, one form of the Greek kai ligature, and possibly some Hiragana syllables. But really, you have to understand how fond Koreans are of using circled letters (both Hangul and Latin) and numerals to see why I am even talking about this. In everyday handwriting, these may well take on forms similar to the "@", with only the central components distinguishing one from the other.

hrant's picture

> you have to understand how fond Koreans are of using circled letters

Interesting - I never realized.
Any handy examples?

hhp

Jongseong's picture

There are some examples in this picture, although it is hard to see: http://pds13.egloos.com/pds/200901/23/53/e0041253_4979349bb31a4.gif

It's trickier than I thought to do an image search of samples of Korean handwriting.

Bendy's picture

Hrant, I was writing some e-mail addresses earlier; seems I'm with you!


Pertinent how I've aligned the first one to the Cap height and the second to the lc!

cerulean's picture

Well now, that just looks like a C to me. Best seek middle ground!

quadibloc's picture

Using the bass clef as the starting point for the design of a new at-sign... and drawing the conclusion that @ is not used much for its original purpose simply because it is now used more for a new purpose... both lack appeal to me, at least.

@ was used for E-mail addresses because its original use gave it a semantic link to "at", which has a range of meaning - at a price, or at a location. If a new at-sign were developed just for E-mail, and displaced the old one among the easily available characters on a computer (otherwise, E-mail would become more difficult to use on older computers!) then when one used a computer for word processing as a typewriter, suddenly ASCII would no longer contain the old at-sign for commercial correspondence.

It's bad enough that the Euro isn't $EU, but has its own brand new symbol that typewriters and fonts don't already have. While new symbols are a fun exercise in creativity, from a practical standpoint they're to be avoided unless desperately necessary.

Bendy's picture

I'm not aiming to redesign the @: I'm observing that its most common use these days is in e-mail addresses and in that context uppercase alignment looks eye-poking.

The reversed bass clef happens to be how the sign comes out simplified when I write quickly. Except in the most experimental of fonts, I wouldn't use it as a basis for type. Or if I tried to make a font from my handwriting :)

hrant's picture

> new symbols are a fun exercise in creativity, from a practical
> standpoint they’re to be avoided unless desperately necessary.

Well, often distancing yourself from a behemothic power is quite practical! If I were European (and to some extent I am) I'd like to avoid a dollar sign in my currency at all costs (pardon the pun). And for example the Mexican currency symbol situation is totally lame.

hhp

nina's picture

"It’s bad enough that the Euro isn’t $EU, but has its own brand new symbol"

?
And what is the "$", god-given? Or any other character for that matter?
Symbols are «made» as they are needed, and the fact alone that the "€" is newer than the "$" shouldn't give it less of a right to exist. And while repurposing existing characters may be convenient,* much like Hrant said it brings with it a whole slew of other problems, like confusability, or the loss of specificity (?) for example with regard to a cultural/regional context.

(* But the need really isn't as pressing now as it used to be. Typewriters? Should their limitations really still be a deciding force for our character repertoire?)

Jongseong's picture

Quadibloc: I think Hrant was talking about changing the design of the existing @, not replacing it with a new symbol or developing a new @ just for e-mail.

Which does bring us to the fact that what we think of as the @ has historically had a number of different uses, with slightly different designs according to use. Here is a previous thread on the subject, with a nice illustration towards the end:

http://typophile.com/node/57817

quadibloc's picture

"And what is the '$', god-given? Or any other character for that matter?
Symbols are «made» as they are needed, and the fact alone that the '€' is newer than the '$' shouldn’t give it less of a right to exist."

It's true that nowadays adding a character is trivial. Just go to Windows Update and downlaod a patch to your keyboard arrangement... and new fonts... and, presto! the new character prints on your screen and on your laser printer when you type ctrl-alt-5 or some such thing.

Back in the dark old days when people would have to run out and buy new typewriters - and when my habits of thought were formed, of course, bringing a new glyph into common use would create massive inconvenience and be an extremely slow process.

splittingfield's picture

Not exactly related, but lately I have found myself enamored with the alternate ampersand in luc(as) de groot’s theSans. It just feels right (to me at least) when using it in an email address in a more modern setting.


nina's picture

Wow, that's very cool.
Does it still work at sizes like you'd use for a business card though?

Tomi from Suomi's picture

From when I started makin typefaces, I've been wondering the @ glyph: it has been too similar from typeface to typeface, and I quess that the reason 15 years ago was that it was just another obligatory glyph to draw, like integral or logicalnot. I think the first new approach was in FF DIN. I took my queue from that, and (unless customer ensists) still make my @ like this:

It is basicaly 'd' with a loop around it.

I'm not saying this is right, but I think it's better than the glyph from early '90s, when @ was set on baseline and the height was everything under ascender line, and the basic form was just copied from earlier versions.

splittingfield's picture

Here is a quick card I threw together with xelatex to show the email address and ampersand at card size. I then took the pdf and converted it to a png for easy posting. Any aliasing is due to the conversion process.

nina's picture

Hm… borderline decipherable now, no? :-\

BTW, as far as I know:
"@" is just an "at" sign; ampersand is "&"

Tomi from Suomi's picture

Yes; ampersand is '&', but ampersand does not cause this sort of stir. Even though ampersand has a huge amount of variant forms.

Perhaps 'at' glyph creates so much interest because the basic form has been so rigid for such a long time. That was my motive to make a new one; to incorporate the glyph more to the overall design of a font, instead of interpreting a generic form to fit the design.

splittingfield's picture

Yes, silly typo there. I meant "at" and typed "ampersand". I guess life could be worse than having ampersands on the mind :-)

I think it could be readable at card sizes. However, I tend to use it at letterhead size
for my own work at least.

Bendy's picture

I think the novelty of it could make it useful at large sizes.

Jongseong's picture

I'm sorry, but to me that alternate @ in TheSans doesn't work at all. I'm all for experimentation, but this will be hard to decipher at normal text sizes. Did Luc(as) really design it to be confined to the x-height?

Besides, even if people are able to see the "AT", those who don't know that the @ in e-mail addresses is read "at" in English will be utterly lost; they would be left to try to decode this purely based on context, as they would for something like "name(골뱅이)typophile.com" in Korean.

Tomi from Suomi's picture

I must agree with Brian; turning a single glyph into two glyphs with something on top and below them can not be progress.

nina's picture

Brian – very good point that transforming a character into what is essentially a catch word glyph is quite limiting in terms of cultural/language context.
Still, in principle I think it's interesting and refreshing to see a designer going about making this character differently – who said they need to always share the same structure (that's certainly not "progress" either). Plus, it's an alternate.

BTW, I missed this the first time round:
"nin@gmail.com"
Reminds me of how people actually tried to capitalize on the formal closeness of "@" and "a" for a while, back when email was still a new and exciting thing. I even know somebody who registered the part of her name that follows the first "a" as a domain name, so she could have an email address that was simply her name with that "a" replaced by an "@". :-\
It's certainly very fortunate that that isn't really done anymore. The close association of "@" with "a" seems to have waned largely by itself, maybe simply due to the fact that the "@" symbol has grown to be seriously relevant, and is now largely "understood" as a character in its own right. All the less reason to make it look too much like an "a" if you think about it that way (but maybe also less reason to completely redefine its design…).

riccard0's picture

The close association of “@” with “a” seems to have waned largely by itself

I haven't checked lately, but, for some time, among some circles in Italy, there was/is the use of @ as a "gender agnostic" replacement for the usual female a and male o ending vowel.

nina's picture

Wow, so it's now associated both with "a" and "o"?! :-\
Maybe we should have another type battle…

henrypijames's picture

I second the idea: This is a great subject for the Type Battle.

.00's picture

The close association of "@" with "a" seems to have waned largely by itself, maybe simply due to the fact that the "@" symbol has grown to be seriously relevant, and is now largely "understood" as a character in its own right. All the less reason to make it look too much like an "a" if you think about it that way (but maybe also less reason to completely redefine its design…).

I disagree.

Jongseong's picture

Just yesterday, I saw a Korean write down her e-mail address making the @ an "a" in a circle. Now that I think of it, I've noted in the past looking at sign-up sheets and such that this is how the vast majority of Koreans write the @ in e-mail addresses; I was the type geek who insisted on a proper connected @ form. I can't even recall seeing another Korean who writes the @ like I do, although surely I can't be the only one.

The lesson, I think, is that we should keep in mind that how one sees a character is often more culture-dependent than one might think. Something to keep in mind when we discuss the designs of glyphs like the @ or numerals, asterisk, daggers, and such.

Syndicate content Syndicate content