List of glyphs that remain upright in italics

Thomas Phinney

I am generally of the belief that some glyphs ought to remain upright even in italic fonts.

I have long used lists of such glyphs when working on italics—even reverse engineering a list at least once, looking at my own previous fonts.

Anybody else keep such a list they are interested in sharing or discussing?

My previous list as a text file (FontLab .enc format) is here: https://github.com/tphinney/science-gothic/blob/master/lib/encoding/upright-in-italic.enc

I am just about to update this for my current project, Science Gothic.

Paul Hanslow

Thomas, this is an interesting topic and hopefully it will be a fruitful one.

I've noticed that 'upright conventions' for italics can differ between foundries—though math symbols remain upright in most cases, so this follows standard practice.

I'd be interested to know if there are established conventions for rarely seen, encoded glyphs, which are found in extended Latin typefaces with huge character sets. i.e. Gentium, Brill etc. I say this with a large degree of nativity and personal interest. 

And a side note why/where did the conventions come from as to what remains upright in an italic? 

Thomas Phinney

I can’t say where the conventions came from! As best as I recall, I formed my ideas and list by a combination of:

• discussions with other staff at Adobe, including Robert Slimbach
• looking at a few fonts both from Adobe and a few other major vendors

Craig Eliason

Do you keep plus upright but slant minus/multiply/divide?

K Pease

An asterisk should not be skewed, but I feel that when it has upright symmetry in the roman, it is appropriate in the italic to tilt it by rotation instead.

Paul Hanslow

Craig Eliason said:

Do you keep plus upright but slant minus/multiply/divide?

Keep /minus /multiply and /divide upright in italic and only alter the metrics.
i.e. change RSB and LSB of mathematical symbols to match italic slant. 

Thomas Phinney

I think those particular glyphs were simply not in my character set at the time. I certainly am not inclined to slant minus, multiply or divide. Indeed, math operators in general, I would tend to keep upright.

Ray Larabie

My list of glyphs that shouldn't be slanted is

1. notdef

Here's a thread.

Artur Schmal

Thomas Phinney said:

I think those particular glyphs were simply not in my character set at the time. I certainly am not inclined to slant minus, multiply or divide. Indeed, math operators in general, I would tend to keep upright.

Nice topic! I'm curious to hear why you believe that math operators like plus, minus, equal should be upright. 

André G. Isaak

I suspect I am in the minority, but I’ve always strongly preferred upright brackets and parentheses even in italics (at least where text fonts are concerned).

Mike Wenzloff

Thomas Phinney said:

... I certainly am not inclined to slant ... multiply or divide ...

For myself, I slant these and other math symbols if the slanted version is viewed as an informal expression of the upright.

In typical text fonts I don't generally italicize or otherwise slant math operators with at least one or two exceptions, and then I provide alternates that are not slanted.

Theunis de Jong

After reading that (long) thread I agree with Ray: whether to use slanted or upright symbols should be at the typesetter's discretion, and not the designer's.
Even if that typesetter might be lazy and may use the font "wrong".

K Pease

For some symbols, the issue is that a vertical line is so key to their identity that the line has to read as vertical to be recognized. In a very square face, obliquing a plus sign to match everything else will not harm the implicit verticality. In the chaos of a text italic, it becomes unclear, and readers may wonder if "not minus" is another symbol that means something else, and if so, how.

This concern can be balanced with the user's desire to see something happen when they italicize. In a display family with a prevailing italic angle of 10°, I slant the vertical bar, broken bar, and up and down arrows only 2°. This is a detectable slant but still reads as "vertical" and the bar won't be mistaken for a slash.

John Savard

It is certainly true that one never sees slanted math operators, but usually that's because the math operators aren't italicized. So it wouldn't matter what they look like in the italic font, because those glyphs are never seen.

Scott-Martin Kosofsky

I agree with André. The tradition in metal typefounding was to make a set of mathematical symbols and operators in each size, in a single weight, to be used with any font. These included parentheses, brackets, and braces, each of which carries a mathematical instruction. While braces had no use in literary work, parentheses and brackets did, so text fonts that were “put up” (the term used in the U.S. to describe the standard assortment of letters arranged for sale) were furnished with parentheses and (sometimes) brackets, though if you look closely, you’ll find most often that their design was the same from font to font. The weight of parentheses varied with the weight of the font, but brackets were always the same: one weight, upright. In Anglo-American and most Western European typographic styles, the only use for italic parentheses is in texts set entirely in italics, but many foundries didn’t offer them at all. Morison’s program at Monotype was one of the first to regularly include italic parentheses, likely because he envisioned the first italic in the program, Blado (companion to Poliphilus), would be used to set texts in the Aldine manner. 

Speaking as a typographer, I would urge designers to NEVER make italic brackets, because, as I see it, there is no such thing. As for italic parentheses in text fonts, I would prefer having them in a Stylistic Set, with upright ones in the main font, though my opinion on this might represent a minority.

Andreas Stötzner

Scott-Martin Kosofsky said:

… NEVER make italic brackets, because, as I see it, there is no such thing.

…

Hm.

Just done a very quick case study out of curiosity …

Can’t see where the problem lies.

Scott-Martin Kosofsky

Sorry, Andreas, but I was speaking rhetorically, with an eye toward typographic tradition. My point was not whether italic brackets exist in fact, but whether, as a new development without a history or any obvious usefulness, they should exist at all. Your examples look very orderly, but I do not think they are valid. I have never seen a mathematical text or formulation set entirely in italics—have you? The meaning of italic characters in mathematical formulas are quite specific; operators are never in italic. The lines you show in italic would be, at best, very rare exceptions. In my opinion, glyphs intended for exceptional usages should be part of alternative Stylistic Sets, not the principal set.

In Anglo-American usage, brackets are used in literature and scholarly work to denote the following: an interpolated text; an editorial remark within a quotation; or parenthetical text within parentheses. While one (not me) might argue that italic brackets would have a use in texts set entirely in italics, such an occurrence would be very rare. And even so, I do not believe that italicization applies to brackets. Just because you can doesn't mean that you should.

André G. Isaak

The above examples don’t provide anything to compare with.



I prefer the second examples in each instance. [n.b. I haven’t adjusted the spacing here. Were upright forms included in the italics, presumably they would be shifted slightly to the right to provide better spacing].

John Savard

Adobe's Arno Pro by Robert Slimbach: a typeface designed to lie between Aldines and Jensons, if I understood the description of its rationale correctly. To applaud it for its stunning originality, when compared to the many Jenson revivals and Aldine revivals we have... might seem even pathetic to those who would seek greater originality from type designers.

But then, even Optima was firmly grounded in classic roots.

Did anyone ever attempt a typeface that was intermediate between an Aldine and a Garalde, I wonder?

John Hudson

When making the types for Brill, I had a specific conversation with the client about parenthetical signs ( ) [ ] { } in the italic fonts, and whether these should be slanted or upright. I was quite pleased that they opted for upright, as this concurs with my own preferred typographic style, but I'm not sure I would make this the default in typical retail fonts, because I know some users prefer slanted parentheses, etc..

In the 2.01 release of the STIX Two fonts, I added a set of upright ( ) [ ] { } to the italic text fonts as stylistic variants. This seems preferable to having users switch to the roman fonts for these signs, since they need horizontal offsets and some significant kerning when used with italic letters, which can be automated if the upright parentheses etc. are in the same font.

_____

The vertical bar | is another character that I personally prefer to set vertical in italic text, but again I know that some users disagree.

Thomas Phinney

Thanks everybody for all the thoughtful (or sometimes brief, but still relevant) responses—even if many of them ended up being more about whether to make glyphs upright in the italic at all, rather than which glyphs.

This has been very helpful for me in thinking through the issues. For my own work, I think my future choice of whether to make everything slanted will depend on the case at hand.

For a text typeface intended for sophisticated settings and users, keeping some glyphs upright still makes plenty of sense for me. I find André’s examples of the effect of that quite convincing.

But for a general-purpose display face intended for a broad range of general users, I am at least open to the possibility of slanting more things or nearly everything (or everything but the notdef, nod to Ray).

When I look at the core system fonts that most people are used to, many of them do keep plenty of glyphs upright in the italic. Not just oldies (like both the Lino and Mono versions of Times), but also slightly more modern typefaces such as Verdana.

But then again, Calibri goes all-slant. Some prominent open source options are also all-slant or close to it, including Source Sans and IBM Plex Sans.

For this particular project, for a variety of reasons, I may go all-slant. But in general? ... ¯\_(ツ)_/¯   I will go with “it depends.”

John Savard

I tend to agree with your conclusion. People can always use non-italicized characters by not italicizing them. Not all word processing software provides access to advanced features of fonts like stylistic alternates, so making access to the italic glyphs dependent on having these features will lock them away from many users. (Of course, I expect many here will say that this is the wrong reason, even if your conclusion is correct.)

You and Ray Larabie are correct, though, about notdef! Italicizing it would only confuse people, leading them to think this was a real character instead of a placeholder that should not be used.