Inferior characters

Andreas Stötzner

I received an enquiry about scientific inferior capitals, both Greek and Latin, which seem to be used in some matters. I haven’t been aware of this up till now, I only did .sup capitals for one typeface occasionally. The official feature definition does not mention them either. So I wonder how common or uncommon this may be, and if anyone knows of any noteworthy implementation in text fonts, or even samples of usage in running text.

John Hudson

Brill have asked me to extend the subscript sets in their fonts to include a full complement of upper- and lowercase Latin and Greek.

[Deleted User]

The user and all related content has been deleted.

Mark Simonson

full character sets

Be sure to leave out the super and subscripts, though, or you'll never finish the font.

Denis Moyogo Jacquerye

I’ll never understand the difference between the sinf feature and subs feature.

The text for sinf only mentions substituting figures and lowercase letters but the text for subs says it can substitute or shift down any glyph.

A quick search shows subscripts capitals are used in electronic notation, for example in Crecraft and Gorham (2008), Electronics, p. 68 [on Google Books]:

A capital letter with a capital subscript refers to a d.c. quantity (e.g. V_AB).
[...]
A lower-case letter with a capital subscript referes to the instantaneous value of the total voltage (e.g. v_AB).

Andreas Stötzner

Denis Moyogo Jacquerye said:

I’ll never understand the difference between the sinf feature and subs feature.

Nor do I.

John Hudson

I’ll never understand the difference between the sinf feature and subs feature.

Oh, I can explain that. The sinf feature was intended for subscripts that sit partly below the baseline, as used in e.g. chemistry, while the subs feature was intended for subscripts that sit on the baseline. That’s how Adobe and Microsoft originally defined the features, and how they implemented them in some early OT fonts.

Then awkward people like me started coming along and pointing out that there was, in fact, no established typographic use for subscripts that sit on the baseline, and that all subscripts sit partly below the baseline: that there is, in fact, only one kind of subscript.

So after a couple of years, Adobe decided that they'd been wrong and that the subs feature should be used for subscripts that sit partly below the baseline, and the sinf feature was unnecessary. I think it should have been formally deprecated, but instead people mostly just stopped using it.

Thomas Phinney

John’s version is close. Except it was some person or people at Microsoft who were sure they needed 'sinf' to have subscripts on the baseline. Nobody at Adobe at the time (late 90s) had any clue why this would be needed. I remember those discussions very well, because it was so baffling to “us” at Adobe, especially when they changed their mind.

Thomas Phinney

Or maybe *less* after they changed their mind. Heh.

Not trying to throw stones btw. Adobe made plenty of dumb mistakes in early feature registrations, and certainly more than Microsoft. ('crcy' anyone? sigh.)

Andreas Stötzner

John Hudson said:

The sinf feature was intended for subscripts that sit partly below the baseline, as used in e.g. chemistry, while the subs feature was intended for subscripts that sit on the baseline.

Why is there no hint towards this concept in the definitions available? Have I overlooked something?

Thomas Phinney

Hmmm, I may have switched 'sinf' and 'subs' in my description earlier. Not meaning to contradict John on that part—I honestly don’t recall which one was which, since in the end they became ~ synonymous! Ah, good times.

Andreas Stötzner

So the two are basically the same; what is the preferable one to apply? Are there known issues with either of them in certain applications?

John Hudson

I favour subs, and really only include sinf (pointing at the subs lookup) because it wasn't formally deprecated.

Thomas Phinney

Same.

Piotr Grochowski

You also have to be careful about including superscripts and subscripts recursively. Be sure to test them out when including them.

The css style would be:

<style type="text/css">
sup {
    font-feature-settings: "sups";
    vertical-align: 0;
    font-size: 100%;
}
sup sup {
    font-feature-settings: "ordn", "sups";
    vertical-align: 0;
    font-size: 100%;
}
sup sup sup {
    font-feature-settings: "numr", "ordn", "sups";
    vertical-align: 0;
    font-size: 100%;
}
sup sup sub {
    font-feature-settings: "dnom", "ordn", "sups";
    vertical-align: 0;
    font-size: 100%;
}
sup sub {
    font-feature-settings: "sinf", "sups";
    vertical-align: 0;
    font-size: 100%;
}
sup sub sup {
    font-feature-settings: "numr", "sinf", "sups";
    vertical-align: 0;
    font-size: 100%;
}
sup sub sub {
    font-feature-settings: "dnom", "sinf", "sups";
    vertical-align: 0;
    font-size: 100%;
}
sub {
    font-feature-settings: "subs";
    vertical-align: 0;
    font-size: 100%;
}
sub sup {
    font-feature-settings: "ordn", "subs";
    vertical-align: 0;
    font-size: 100%;
}
sub sup sup {
    font-feature-settings: "numr", "ordn", "subs";
    vertical-align: 0;
    font-size: 100%;
}
sub sup sub {
    font-feature-settings: "dnom", "ordn", "subs";
    vertical-align: 0;
    font-size: 100%;
}
sub sub {
    font-feature-settings: "sinf", "subs";
    vertical-align: 0;
    font-size: 100%;
}
sub sub sup {
    font-feature-settings: "numr", "sinf", "subs";
    vertical-align: 0;
    font-size: 100%;
}
sub sub sub {
    font-feature-settings: "dnom", "sinf", "subs";
    vertical-align: 0;
    font-size: 100%;
}
</style>

Denis Moyogo Jacquerye

@Piotr Grochowski, you should use font-variant instead of font-feature-settings, or specifically font-variant-position for superscripts or subscripts.

Piotr Grochowski

Denis Moyogo Jacquerye said:

@Piotr Grochowski, you should use font-variant instead of font-feature-settings, or specifically font-variant-position for superscripts or subscripts.

I don't get how that would help. How would you stack superscripts and subscripts?

Denis Moyogo Jacquerye

How would you stack superscripts and subscripts?

With a font with a MATH table I guess.

Are there actually any fonts that hijack the ordn, sups, subs, sinf features to do superscripts of superscrips, superscripts of subscripts, subscripts of superscripts or subscripts of subscripts?

John Hudson

I’m doing some updates to the Brill types, and one of the things requested is secondary super- and subscript numerals. The approach we've taken is that the numerals have to be encoded as the Unicode superior and inferior numeral characters, and then have the appropriate sups or subs feature applied to them. 

Andreas Stötzner

• We have Latin lowercase superscript a–z completely encoded (although scattered over several places)
• We have most of Cyrillic lowercase superscript and subscripts encoded, mainly in the 1E030 block
• Of Greek lowercase we have only a very few letters encoded as superscript modifiers.

I understand that these char. groups have got their encoding with the background of phonetics in the first place. But now I stumble across Greek superscript letters in mathematics, which are not encoded.

Wouldn’t it be logical and sensible to complete the sup.s and sub.s sets in the UCS once and for all, instead of labouring about it again and again over decades? – Just wondering (not the first time) why there is so much inconsistency in the UCS.

John Hudson

The math equations you show can all be handled using math fonts and software without direct encoding of superscripts and subscripts. The multi-line equations you show in the first image require such fonts and software.

Andreas Stötzner

Thank you John, I see your point and I was aware of the fact that math composing is different. However, what if such letters appear also in a bit of running text (like in the 2nd image)? What if they occur in matters other than mathematical equations? What if adressing those sub.s and sup.s by OT function is not ideal?

John Hudson

It is the usual question about where one should draw the line in encoding sets of characters, as distinct from encoding discrete attested characters. When it comes to superscript and subscript alphabetic characters, Unicode has taken the approach of encoding only those that are attested with semantically distinct meaning, which means e.g. the subset that act as modifier letters in phonetic notation.

Andreas Stötzner

I know that reasoning of Unicode, of course.

But knowing about it still doesn’t solve the obvious conundrum.