thanks, Stephen. I think i'll go with proportional as the default and tabular as one of the OT features.

how common are tabular oldstyle numerals? are they worth messing with?

Alvin Eisenman wrote a piece for Font & Function some years ago entitled REINTRODUCING LOWERCASE FIGURES & SMALL CAPITALS, which is well worth reading if you can find it. (Sorry, I don�t have the volume & issue). BTW, I believe it was Paul Rand who first used the term �lowercase figures.�

�Modern� can be a deceptive term -- I once had a book about Thomas Pynchon to design; the editor remarked �this is a modern book, it should have modern figures.� I pointed out to her that people are apt to use the term �modern� to describe their own time, and �modern� figures became common in Victorian times, so they are �modern� just like hoop skirts. The image of Pynchon in a hoop skirt won out; we used �oldstyle� figures.

Old-style figures have ascenders & descenders, just like lowercase letters. �Lining� figures are like capitals, & should always be used with them. But a text peppered with lining figures visually calls attention to the figures, giving them a greater visual weight, which is usually not appropriate. Even with mathematical expressions, �old-style� figures were common in the 19th and early 20th centuries. Having said that, I�d guess most scientists today are more comfortable with lining figures, so I�d use them for certain kinds of books. Ideally, I�d mix them, using �oldstyle� for dates & common numbers, with lining for technical constructions. But you always get into odd situations, and it is far easier to use just one style of figures.

As to tabular old-style figures, how common they are depends on the typesetting system. Most Monotype metal fonts, and certainly most Linotype linecaster fonts, did not use a common body for oldstyle figures. With the introduction of the Linotron 202, almost all the supplemental fonts that had os-figs had them on a common body � i.e., no �proportional� figures. You did a lot of kerning. Most PostScript Type 1 fonts seem to have reverted to the hot-metal days, and the os-figs were not on a common body. For many fonts, it was easy enough to construct them. Matthew Carter constructed a special set for Galliard; you can get them from Carter & Cone if you ask � they are drawn differently than the proportional figures; he didn�t just put the proportional os-figs on a common body.

As for using oldstyle figures in tables, I use them when the tables are straightforward data, and when the book isn�t one I think the audience would be more comfortable with lining figures.

And I�d note that it is sometimes necessary to make a tabular comma and period, for either os or lining tabular figs.

FWFW

Charles

Bloody thing gave me an error message so I re-posted it. So it turned out the first one did go through. We don't need it twice, that's for sure.

I can edit the second posting, but not delete it. You just can't find a teenager when you need one.

Sorry,

Charles

thanks for the info, Charles. that was rather insightful, especially the part about actually redrawing figures for tabular settings. (it was too much common sense for me think of that!) do you know of anyone else who follows this practice?

oh and you can usually just ignore those error messages. just make sure you copy your posting, back out of the message and then go back in to double check that it actually posted.

Paul,
It is hard to read your user's intentions when designing a typeface. If your intent is to make your font more marketable to everyone, I would suggest including both lining and oldstyle figures. If you must choose one, I would make it lining but at a height somewhere between cap height and x-height. This makes the numbers less obtrusive in the text but still allows for the tables found in annual reports and price catalogues. Like Tiffany, I think business cards set in oldstyle can be disruptive when you have so many phone numbers and addresses and zip codes all packed into such a small size card. This again is the users choice and why having both sets is always the best solution.

ChrisL

IIRC the new FF Unit does have special tabular figures, plus lining figures and oldstyle figures. And Adobe OT fonts usually have the 4 styles of figures: tabular lining, tabular oldstyle, proportional lining and proportional oldstyle.

[update: see Mr. Spiekermann's post regarding FF Unit's figures here.]

and then there are numerator figures, denominator figures, scientific inferior figures, scientific superior characters... ad nauseum ;^)

It depends on the style/usage of the font.
In general, for a text face I would make one or both of two main styles: hybrid proportional, and 3/4-lining fixed-width. And unless the caps are very short I wouldn't align the latter with them (in general). This will cover you 90%. BTW, see de Groot's numerals for the best in the west.

As for non-numeric characters that might benefit from having a relationship with a fixed-width numeral width, certainly the currencies should be that width - and here's a trick I learned from JF: make stuff like commas and periods (and the en dash) exactly half that width, and make the blank space character that width too. That way you can easily pad the widths as necessary.

* Actually, you would of course determine the space first and then double it to get the fixed-width for the numerals.

BTW, Paul, dude, are you guys gonna give Typophile.com credit when the Lanston stuff is [re-]released? You've been mining our heads like there's no tomorrow! ;-)

hhp

You've been mining our heads like there's no tomorrow!

Just delving into the dark mysteries of text fonts. I've always been more of a display guy myself, but you know how one thing leads to another...
Anyway, I've been relying on your collective knowledge for almost a year now, this is no new development! ;^)

Paul, my suggestion would be to include every possible numeric format in the OpenType font. Then, the user has everything to choose from. If that means ten different numeric styles, then so be it (as long as you can program that)!

And I, for one, enjoy seeing all of your questions. Keep 'em coming!

>BTW, Paul, dude, are you guys gonna give Typophile.com credit when the Lanston stuff is [re-]released? You've been mining our heads like there's no tomorrow! ;-)

This wouldn't really be necessary, IMO. We all use Typophile to get our questions answered.

I don�t think too many font designers re-draw their figures for tabular use � The �oldstyle� (I really prefer the term �lowercase�) figures in Carter�s Galliard seem strongly influenced by Van Krimpen�s work. The figures were far from robust, & I confess I never liked them � we modified the tabular figures for a proportion set for Galliard in PostScript Type 1. Back in the days of the Linotron 202 when you couldn�t modify fonts, but could select what supplemental font set you used in a job, we used Sabon figures with Galliard. For American audiences anyway, that worked quite well.

If the figures in the font you are designing are more robust than Galliard, you probably don�t need to redraw them � at least not much � for tabular use. Of greater importance I think are the period and comma. In text, when a period or comma is used, a word space will usually follow, so their position in the set of the character usually assumes that space. In a table, no space will follow the period or comma. Moreover, you don�t kern characters in a table; to do so would lose the alignment.

The trick is to open the Metrics Window in FontLab or the Kerning routine in Fontographer, and key in something like 7,000 and 7.012 using your tabular figfures � Actually, all numbers with periods & commas before & after them. Usually, you�ll discover that you will want a wider set for the comma and period, and the left sidebearing will be smaller.

FWIW

Charles