Superior and ordinal numbers

mauro sacchetto

Should there be a difference between the glyphs of superior numbers and ordinal ones? In some fonts I see that they are set in different lookups, but the glyphs are exactly the same (better: the same double glyphs with different extensions). I instead created only one lookup with two subtables, one with ordn and one with sups with reference to the same glyphs. What is the most correct practice? Thank you

John Hudson

If the glyphs are indeed identical, then there is no reason not to use the same lookup in both features. Some styles of ordinals have a small line below the superscript letter, so in that case they would require separate lookups. My understanding is that in contemporary Spanish and Portuguese usage the line is considered old-fashioned, and the preference is for ordinals without the line.

Personally, I think supporting old-fashioned typography is sometimes appropriate to the design of a typeface.

mauro sacchetto

Thank you. I considered, for example, EBGaramond12: it has sups and ordn numbers completely identical, but two separate lookup. In my inexperience, I find it redundant 

Thomas Phinney

Agreed: it is preferable, albeit only mildly so. When one doesn’t see it done, that may be because many type designers are not quite savvy enough about the inner workings of OpenType layout to make a lookup and reference it twice. Or maybe they think it isn’t very important—and I wouldn’t argue that point.

Not only is your approach more efficient, but it also reduces the likelihood of error by only having one set of data. Yet... given the small number of glyphs typically involved in 'sups' and 'ordn', I suspect the file size savings is just a handful of bytes. So, an efficiency, but not a terribly important one.

notdef

Wait … what do you mean by ordinal numbers? Figures reduced in size and raised to match the ordinal indicators (i.e ªº)? What kind of language uses those?

John Hudson

Oh, good point, Frode. I'd quickly read the original post and assumed it was referring to ordinal numbers e.g. 2ª and 8º, not to superscript numerals.

mauro sacchetto

My question was about both kinds of numbers. The superior numbers are those used in superscript to indicate the footnotes. I mainly use my native language, Italian, and usually in Italian, the ordinals are printed in very small characters (and without any small line below). 
Since, however, some fonts have both types of numbers, I went to see, as I wrote above, the EB Garamond: it contains two sets of absolutely identical glyphs, one for superscript and one for ordinal numbers. I wondered if in general they must be the same or if it is more correct to distinguish them typographically, and in what way: the ordinals even smaller? And higher?

John Hudson

the EB Garamond: it contains two sets of absolutely identical glyphs, one for superscript and one for ordinal numbers.

Are you sure that's the distinction? It may be one set for superscripts and another set as fraction numerators.

Nick Shinn

As per Frode, I’m not aware that such a thing as ordinal numbers exists.

I put superior letters in the <ordn>feature, and, if a full alphabet, code them for English (just like Word!—but not as default).

I often make two sizes of superior figure, with those for <sups> being clearly smaller than those for <frac> and <numr>.

George Thomas

Re: Ordinal Numbers -- They are not a special set of numbers

http://www.personal.psu.edu/ejp10/blogs/gotunicode/2009/11/glyph-dujour-romance-ordinal-a.html

mauro sacchetto

Thank you for your broad answers.
2 last questions:
1) I saw that different fonts present different solutions, but is it appropriate or not that the superscript numbers (those that indicate the footnotes, for instance) are the same as those that make up the fractions? In some fonts they are the same size, in other fonts the footnotes numbers are smaller ...
2) Do superscript letters and numbers have the same baseline? In GaramondPremierePro the don't have, in EBGaramond they do. Is there a standard, or are there some reasons of opportunity to marry one or the other solution?

Andreas Stötzner

I get confused by the term ordinal numbers. In typography, I thought it is best (and unambiguous) to speak of figures (1 2 3 …) and, in this case, letters (a b c …).

The superior figures and letters I make so that they match in height and size. The inferior figures are the same outlines as the sup.s, positioned low.

The fraction figures I make larger than those.

notdef

I usually match the size of denominators, numerators, inferiors and superiors. Dnom/numr have to relate to the cap height, and not appear to large combined as fractions within it. Sups/subs/sinf may in one project benefit from being relatively large and noticable, and in the next from being as small as possible. Depends on context.

My ordinal indicators hang from the cap height, and share baseline with the superiors. They are scaled proportionally from the lowercase letters and lining figures, so the superior figures stick up above the cap height slightly.

mauro sacchetto

In the absence of a shared general rule, do you think this arrangement is plausible among those possible?

notdef

Historically, they were just a smaller point size, right? I think it makes sense to keep roughly the same relationship between x-height and figure height in superscripts.

mauro sacchetto

In other words, to increase the size of the figures so, as to say, that "1" has the same height as the "t"? My solution supersedes old style numbers with lining ones, which is why I made them the same as lowercase x

Bhikkhu Pesala

I recommend that superscripts should be 65-70% of CapHeight. Yours are about 64% of xHeight, or only 41% of CapHeight, which is too small for legibility at normal pointsizes.  Superscripts of 12 point text would be less than 5 point. 

I have always designed the tops of superscript figures to align with the top of Capitals, but I now incline more to align them with the top of ascenders to make footnote references more visible when scanning a line of text. 

One could also align the top of superscript ascenders with the top of ascenders, and the bottom of superscript figures with the bottom of superscript letters. 

mauro sacchetto

I attach here a small example from a book of Italian publisher Einaudi, which uses the well known Simoncini Garamond. In your opinion, are too small the footnote numbers?

notdef

This illustrates my approach.

George Thomas

In your opinion, are too small the footnote numbers?

Not too small; those are a typical size.

Nick Shinn

Is there any typographic situation where both figures and letters appear together in <sups>, and therefore require a common “baseline”?

Bhikkhu Pesala

mauro sacchetto said:

I attach here a small example from a book of Italian publisher Einaudi, which uses the well known Simoncini Garamond. In your opinion, are too small the footnote numbers?

I would say so, yes. They are 50% of CapHeight. The superscripts illustrated by Frode are 71% of CapHeight, which is just slightly too big IMO. 

Craig Eliason

Nick Shinn said:

Is there any typographic situation where both figures and letters appear together in <sups>, and therefore require a common “baseline”?

In math you might see b to the power of n+1.

notdef

Footnotes may very well be annotated [1a], [1b] etc.

John Hudson

Is there any typographic situation where both figures and letters appear together in <sups>, and therefore require a common “baseline”?

Yes, this can occur in apparatus critici of scholarly editions. This example is from the Biblia Hebraica Stuttgartensia.



I always design all my superscript (and subscript) on the assumption that they should share a common baseline and, regardless of the actual set included in a particular version of a font, I proportion them so that they can be reasonably extended to full uppercase, lowercase, numeral and common punctuation and math symbol sets.

This means, of course, taking lowercase descenders into account when determining the size and alignment height of the superscripts.

mauro sacchetto

Ok, than to all!

Bhikkhu Pesala

I reuse the superscript glyphs for an ordinals feature and for the numerators in fractions. The subscripts and denominators are composites of the superscripts, and sit on the baseline. 


Scientific inferiors (0-9) also reuse the superscript glyphs as composites, and bisect the baseline. 

Nick Shinn

Here is one of my more scholarly faces, Goodchild.

Note that the superior figures are smaller than the subscript and fraction figures. I make the subscripts larger, because they are often read as part of immersive text, not merely for reference.

Claudio Piccinini

Nick Shinn said:

Here is one of my more scholarly faces, Goodchild.

Note that the superior figures are smaller than the subscript and fraction figures. I make the subscripts larger, because they are often read as part of immersive text, not merely for reference.

Nick, I am reasoning on this right now, to decide a standard for my typefaces.
Aside from the choice of making them slightly different in size, by "subscript" you mean the superscript/subscript (onesuperior-u00B9, oneinferior-u2081 et al.), which you make the same height of figures which will compose fractions i.e. numerators and denominators (one.numr, two.numr - one.dnom, two.dnom)? Thanks much!