Speaking of tabular figures...

charles ellertson's picture

Why is it that the superior figures, if present, are always on the same body? Whether you're using them for footnote calls or scientific superiors -- or anything I can think of -- they need to act like proportional numbers, not tabular. It wouldn't hurt if they were kerned, either.

Miss Tiffany's picture

All typefaces? Or specific?

charles ellertson's picture

Not to say there aren't some out there, but all the fonts I have that include superiors have them on a common body. Anyway, Miguel gets tired of me being specific -- but to Adobe's credit, at least they include them in fonts that could be used for text.

Or are you asking if specific typefaces should have "proportional" superiors, whereas others should have "tabular" superiors? I can't think of any reason to have "tabular" superiors, but maybe I'm missing something.

Nick Shinn's picture

Surely when superior figures appear in a footnote, they are neater if the same width.
(The same is true of the traditional reference marks.)

The space on either side of the one is part of the character, and helps distinguish it from l and I.
Nonetheless, the number 11 as a reference number in text may need a bit of kerning!

Numerator and denominator figures may be kerned, for use in arbitrary vulgar fractions (sounds pretty punk, eh?)

charles ellertson's picture

Surely when superior figures appear in a footnote, they are neater if the same width.

Only with one style, not much used. That style is flush & hang with superiors numbers. If you use paragraph style, with either on-line or superior numbers, it matters far less. Yes, you can make a case for tabular numbers, esp. if the book uses notes a lot of reference-type notes -- ibid & title-only citations create a lot of short lines, so the numbers do form a sort of list.

I can't remember the last time we set superior note numbers for footnotes proper. It does happen, in once or twice per every 500 books.

Note calls are an entirely different matter. Whether you are setting footnotes, end-of-chapter notes, or end-of-book notes, they have to be signaled in the text. Those are always superiors. Frequently, they fall after quote marks, periods, or commas. The way everything is spaced today (from most foundries), you get a very tight sequence of letter, period, quotedblright, and then a very gappy set of superiors. And the numbers themselves need kerning against the punctuation, they are too tight against a quotedblright, and too far after a period or comma. Then, you also run into a few authors who call a note after a word without punctuation.

But if you really want to do it in perfect harmony, have two sets of superiors for notes, one for the calls in the text, one for the few times when superiors are used in the note proper. And I suppose you could add another set for scientific superiors, if you think a particular font is good for setting science or math.

Thomas Phinney's picture

Any reason one couldn't have both proportional and tabular superiors (and inferiors) in the same typeface, no?


charles ellertson's picture

No reason, I suppose. I'm sure your boffins can do it, but I can't think the feature code for that one through.

(And if you implement it, don't forget the kerning with proportional!)

Nick Shinn's picture

Thanks Charles, this is valuable information.
Could you post some specimens which demonstrate your preferred spacing in the situations you mention, showing where you have had to add or remove kerning?

Any reason one couldn’t have both proportional and tabular superiors (and inferiors) in the same typeface

Don't spoil the lad Thomas, he'll have nothing left to do if the fonts practically set themselves :-)

Thomas Phinney's picture

It's much more insidious than that, Nick. If, through complexity, Adobe can drive all non-Adobe type designers mad, we'll be the only ones left making new typefaces! Mwahahahaha!

Or, maybe not.


charles ellertson's picture


OK, examples using Arno Pro, from the foundry, set in InDesign:

k.l.'s picture

The little-descending slash actually is a very good trick to avoid -- more kerning trouble with descending letters to the left!  ;-)

Nick Shinn's picture


1. Yes, superior figures should be kerned to follow lower case letters, if for nothing else than "f". However, some negative kerning may be useful when following letters such as "a". So, for an OpenType font, the designer should create a kerning class for the superior figures, and then kern one of those, the two, against the lower case. Question: is there ever a need for superior figures to follow capitals or small caps?

2. So, you're suggesting the 7 is too close to the quote mark? I would agree. However, in this and the following example, I don't have a problem with excessive space between superiors--it is of the same order as the space between lower case figures, and surely the goal is to create evenness amongst the different typographic elements? The 7 4 sequence is most gappy of figure combinations, but occurs so rarely, let's leave it to the typographer/comp to kern, if they consider it necessary.

3. The problem with introducing kerning between a period or comma and following superiors--it backfires when the mark itself follows f, r, v, w, or y, and is kerned negatively with them. In those situations, the arm of the f, for instance, would touch the superiors following the mark. Therefore, as letter-to-mark is a more frequent occurence, mark-to-superior kerning must be left to the typographer/comp. There, still some work for you guys to do, at least until Adobe invents three-character contextual kerning!

4. In an oldstyle face, where capitals and ascenders have different heights, and the f is frequently tallest of all, surely it's not a problem to have a variety of vertical alignment. As with accents (diacritics), it's a judgement call for the type designer, based on the face and the particular accent/superior, whether to align by centering along a horizontal axis, or "flush bottom".

5. I'm afraid I have to disagree. IMO, the slash should, as much as possible, disambiguate from the vertical extremity of all other characters. Generally, this means it's a little above cap top, and a little below baseline, only lining up with ascender top in old style faces. I mean if you're OK with the irregular depth of old style figures, which is never the same as the descenders of letters, then surely having the slash somewhere else again fits in with that scheme. After all, if the bottom of the slash is lined up with the bottom of the descender, the question arise, why isn't it lined up with the bottom of the old style figures? It's better to have it line up with neither, which provides greater consistency.

Nick Shinn's picture

[Duplicate post deleted.]

k.l.'s picture

[An addition. Nick, I only saw your post after clicking the "post" button!]

Spacing superior numerals and letters is tricky and hard to address with the current sidebearings+kerning approach: superiors are smaller and need less space "within" (inbetween two superiors) but more space to the outside (between normal glyph and superior). Solving this by pair kerning alone would require kerning them with almost the entire glyph set. In future this could be done contextually.
The same, btw, is true for all caps spacing. Uppercase letters usually are spaced to fit best in lowercase context. This implies that with the current pair kerning approach, uppercase-uppercase pairs requires a lot of kerning to make the white space in A L T V W Y disappear. Again, contextual positioning can help by not only do pair-wise corrections but by giving uppercase letters completely different sidebearings in all caps context.

Your examples uncovers a weak point. Not so much of Arno or any other particular typeface, but of the underlying font technology.

charles ellertson's picture

First of all, I'll abandon the comment on the slash -- it is as deep as an old-style figure, and that's enough. And it is off-topic.

Back to superscripts: Obviously, the clashes & too tights should be fixed, either in the font, or by the compositor as set.

As to kerning the superiors themselves: whatever balance you want should be achieved. But look at the space between 7 and 4. Just like any other set of numbers, if not kerned, it is far greater than the space between 6 and 9. And 69 & 74 as note calls could well fall in the same paragraph Then look at 100. The space between the one and zero is large; the space between the two zeros quite small. etc. Just like with any tabular number set.

The example I gave does not show the usual situation, justified copy, with all the compromises of word spacing that occur on the page. But these spaces between the word and the note call are fixed, in that they result from the sidebearings of the characters themselves. How you space them needs to accommodate a tight and a loose line. Ideally (I believe), there should be less space between the number & what it refers to than in the following wordspace, whether that space be tight or loose in a particular line of justified text. Nothing new here, it is simply that these numbers must be made to space correctly just like any other set of numbers.

For what it is worth, when I rework a font, I make these numbers proportional (varying setwidths). I then kern them against themselves. I next kern them individually (pairs kerning) with close quotes, periods, and commas, because that is the most commonly found placement. When I get the sidebearings just right, I probably could use class kerning for period/comma & quoteright & quotedblright and the "super" class. I suppose if you are a foundry, you'd consider it sloppy not to do so. For me, since as a comp I have to go back into the file & find my place, I use pairs kerning for important stuff.

For the rest, these are the characters I start with:

quotesingle asterisk exclam bracketright c d e g i j k l p r s t v w x y question uni02BE uni02BF];

adjusting the class as needed for particular fonts.

And typical kerns:

pos @_quotesingle @_zero.super 50;
pos @_f2 @_zero.super 130;


dberlow's picture

I think the safest thing for now might be to put in two sets, one proportional for setting 'numbers', and one tabular on a 1/4 em or so and shifted right until the biggest right kern in the Family, is cleared, just for setting 'notes'.

"OK, examples using Arno Pro, from the foundry, set in InDesign:"
The alignment is bad, the weight, width, kerning, and spacing too. Are these drawn, generated or borrowed from Optima?

I agree with Nick on the slash. I try to make it embrace from below the l.c. to above the U.C., regardless of descent.

"Uppercase letters usually are spaced to fit best in lowercase context."
A good point. Uppercase in many old style faces, are also 'designed', to compose best in a lowercase context.


charles ellertson's picture

The Arno was set in InDesign, with nothing done to the font, just used as it came from the foundry, with default ID settings.

BUT. I exported a PDF, brought it into Photoshop, and had to resample for 600-wide pixels. That probably fuzzed it up. I'm not use to making screen dumps, or posting images, for that matter. I may not be all that good with words, but sure better with them than with all this new stuff.

Set it yourself & see what it looks like.

eliason's picture

Question: is there ever a need for superior figures to follow capitals or small caps?

Academics can and will footnote anything, so, yes!

charles ellertson's picture

Sorry, I see that Nick wanted an example of a font I'd worked over. For good or ill, here is a worked-over Adobe Garamond set just as the previous example:

Nick Shinn's picture

Thanks, they all look good to me, except the long slash.
And I don't believe you made any change to vertical alignment in item 4.

I'll be putting character-to-superior kerning in all my future fonts, and making the superior and inferior figures proportional and kerned. I just hope there's not some other important usage of superiors/inferiors we've overlooked where's it's necessary for them to be tabular!

Thomas Phinney's picture

"at least until Adobe invents three-character contextual kerning!"

Already exists, and is supported in the AFDKO, and in InDesign. FontLab and other retail tools don't do it yet, however. Not sure about VOLT.

As it works just like other contextual stuff, I doubt that three glyphs is a limit in either AFDKO or InDesign, but I'm not sure how large a context has been tested.



Nick Shinn's picture

Here's a usage where tabular superiors are required: price lists which use the "superior cents" style.

It seems to me that the option of Tabular or Proportional figures should be available for ALL figure styles.
At the moment, the "tnum" feature would seem to allow this, but how do Quark and Adobe interpret this--if I create an OT class of tabular superiors and another of proportional superiors, and put the appropriate coding in my "tnum" feature, is this something that could be implemented by a layout application?

Another instance is "cap figures".
I am producing several typefaces which have three-quarter lining figures as the default or default lining figure, but which also have cap figures (cap height and weight) for when "case" (all caps) is applied. At the moment, I am making most of these figures proportional and kerned, but that doesn't cover all potential uses. Ideally, and in honour of the great god Proliferation or his pal Becozwecan, why not have tab cap figures too? in fact, the font shown above IS with tab cap figs, because the superiors align with the top of the caps and cap figures, and not with the three-quarter figures.

k.l.'s picture

and put the appropriate coding in my “tnum” feature, is this something that could be implemented by a layout application?

This should work. The application doesn't care which glyphs your tnum or pnum cover. The important thing is feature order. If your superior feature precedes tnum/pnum, then your addition will do. If they are ordered the other way round, then your superior feature needs to replace tabular numerals with tabular superiors, and proportional numerals with proportional superiors, because the tabular vs proportional selection has been done already.

It seems to me that the option of Tabular or Proportional figures should be available for ALL figure styles.

In a way, yes. Unfortunately every addition does not add to but multiply the number of numerals ... I really hate doing numerals. They cost me at least half the design time.  :(

Nick Shinn's picture

OK, the simplest method is to have one set of superior figures which is both monowidth AND kerned.

The kerned version is of course the default according to the layout application, and that's good, because that's what's mostly used.
But on the rare occasion when tab superiors are required, just enter "0" in the kerning field and voilà--tab figures!

k.l.'s picture

Or use two (identical) sets of superiors, both tabular, but one is kerned while the other is not. Then 'pnum'/'tnum' would care rather than kerning settings.

Earlier I wrote -- ... superiors are smaller and need less space "within" (inbetween two superiors) but more space to the outside (between normal glyph and superior).

As simple solution: Imagine that superiors are spaced rather tight so that a string of superiors looks well spaced, but superiors and surrounding non-superior letters are a bit too close. Now the latter can be adjusted like this:

   @SUPERIORS = [zero.sup one.sup two.sup ...]
   @NON_SUPERIORS = [A B C ...] # contains all glyphs not being superiors

   feature kern {

      # here the content of your standard kern feature:
      lookup standardKerning {
         pos V odieresis 10;
         pos X one.sup 10;
         pos @UC_V @LC_O -50;
      } standardKerning;

      # increase space between non-superiors and superiors:
      lookup spaceBeforeSuperiors {
         pos @NON_SUPERIORS @SUPERIORS 15;
      } spaceBeforeSuperiors;

      # increase space between superiors and non-superiors:
      lookup spaceAfterSuperiors {
         pos @SUPERIORS @NON_SUPERIORS 15;
      } spaceAfterSuperiors;

   } kern;

The trick is to use two additional lookups for adding space leftside of superiors and rightside of superiors independently. Space inbetween two superiors is not touched.
An advantage is that effectively you do a global adjustment -- and are still free to also kern superiors the usual way. Kerning values resulting from different lookups will add. E.g. the pair "X one.sup" will be kerned at 25.
A disadvantage is that older and most current applications would ignore this kern feature completely because it contains more than one lookup. Adobe CS2/CS3 applications can handle this, and hopefully other applications will add support too. Considering that many Word users may not know how to switch on kerning anyway, I don't think this is a real issue.  ;-)

(There's a more elegant solution with contextual positioning which would get more matches, but it does not work yet.)


Nick Shinn's picture

Or use two (identical) sets of superiors, both tabular, but one is kerned while the other is not. Then ’pnum’/’tnum’ would care rather than kerning settings.


However, isn't your kern feature code a bit of "overkill"?

Firstly, I wonder if kerning after superiors is ever necessary? For closing quote marks perhaps?--but shouldn't a font already have enough of a left sidebearing on its quotemarks anyway?

Secondly, isn't it best just to go through the alphabet and proportional figures, and kern each superior figure individually?

k.l.'s picture

Hello Nick!
This is as tiny as it can be.  :)  You should see the code for uppercase spacing which is built on the same basic idea ...
Even though these additional lookups are located in the kern feature, better consider them as spacing adjustment rather than kerning. They make sure that else too narrow sidebearings get 'normalized' in context of glyphs other than superiors. It is like adjusting the fundament, based on which you would do the fine-tuning: Indeed you would go through relevant pair and kern them individually.
(Also I don't say that such an adjustment is required in every typeface. It's just an offer. As trivial as this solution is, it can be helpful for similar general spacing adjustments too.)


Syndicate content Syndicate content