Good question. To a companion roman, an italic should have consistent heights of upper and lowercase, as well as figure height and other character alignment. The spacing and weight of the two should be optically equivalent, which means they appear to be about the same, (the trickiest part). The italic should be offset so an italic word set amongst roman will balance between roman words. The design of the uppercase forms should be within the same class, i.e. you would use the same serif structure for uppercases in both. the lowercase structure can have more variety between roman and italic, by tradition, but you wouldn't e.g. make the lowercase tails, terminals or serifs much wider, heavier or contrasting than the roman. Whatever kind of stress, or pen angle, that is apparent in the roman should follow in the italic, despite the angle of the italic stems.

You could think of this as if you were writing the fonts, and didn't want anything but emphasis to change, or not. ;)

Cheers!

My position is a pain to defend in public, but I stand by it like some crazed Japanese soldier in a tree: in terms of "belonging" to its roman (which is what an italic in a text face needs to primarily do) the best an italic can do is a be a slanted version of the roman, albeit with two important twists:
1) Aesthetics: some features (like double-sided serifs) don't look good slanted, and some glyphs (like your average bicameral "a") look lousy too. Those need to be fixed.
2) Differentiation: some slanted glyphs (especially if you like your slants slight like I do) need extra help not to be confused for a roman glyph.

So the slanted-roman I'm proposing is nothing like what ATF did and Morison proselytized; it's much more like what the Deberny foundry was doing in France towards the end of the 19th century and some German foundries were doing in the first quarter of the 20th century.

I've done something in this vein as well:
http://www.themicrofoundry.com/other/nour&patria/dev/nour-latin.gif

Essentially, I would caution against the mainstream parroting of "an italic must be cursive"*, because cursiveness is almost guaranteed to lend an atmosphere to the italic that the roman does not share. We need to find something more sensical, and really slant capably does all the heavy lifting.

* Which leads to dysfunctional banalities like "upright italics". Gag.

hhp

hrant, in your view does it matter if the slant is to the left or the right (and if so, why)?

(And dare I ask, where does the crazed Japanese soldier in a tree simile come from?)

Hrant, just curious: do you feel that Jeremy Tankard's Kingfisher italic fits your bill, or did he "fix" too much? (I realize that the f probably kills it for you.)

-- K.

"We need to find something more sensical and really slant capably does all the heavy lifting."

It could, in the right class, (like swiss sans). It can't in the wrong class, as you show.
But generally you are 1/2 trying to redefine something that's been worked out over 100's-o-years.
The other 1/2, aesthetics and differentiation you're talking about, are not found in pure slant, Hrant.
After all, how popular is that work from Deberny or the German foundries?
Users are not so much interested in theoretical writing, as they are in actual reading.

And who do you mean by "we"?

Cheers!

Craig:
http://www.wanpela.com/holdouts/
http://www.bbc.co.uk/history/worldwars/wwtwo/japan_no_surrender_02.shtml

More soon.

hhp

To a companion roman, an italic should have consistent heights of upper and lowercase...

I'm sure David meant to say visually consistent heights. There is an optical effect that can make a slanted letter look taller than an upright letter when they are mathematically the same height. This is because the actual stroke length of a slanted form is longer than that of a vertical form. In particular, one may want to make the actual height of the italic lowercase slightly shorter than that of the roman, to compensate for this effect. This must be judged visually.

Whatever kind of stress, or pen angle, that is apparent in the roman should follow in the italic, despite the angle of the italic stems.

That's a solid approach, but I think a more basic observation would be to say that the stroke construction model should be the same -- even if the actual 'pen angle' were allowed to vary. Note that this observation doesn't imply a strictly chirographic approach: whatever 'idea' is used to regulate the construction of the roman letters should be applied to the italic letters in order to make the two work together.

It's not the the differences or commonalities between an italic and a roman, per se, it's the ratio of one to the other. This is particularly true if one is jettisoning the cursive model for an "oblique and tweak" approach as the reader has just lost a lifetime of learning that cursive equals italic.

A related and interesting historical study might be how italics were married to romans in the centuries before families (font families, not mom, dad, and that short aunt who smoked too much). Upon what did they base their choices? I fear much of it was simple availability, but even then the printers were probably going down the same criteria checklist.

"Oblique and tweak" is more fun to say than to read, by the way.

Tongue is firmly in cheek, jaws not moving because it would hurt . . .

Type designers are control freaks. The key word here is "belong." So, does Blado belong to Poliphilus? Does Arrighi belong to Centaur, or only to Monotype? What about the italic for Bunyan (or is that Pilgrim)?

Maybe a better word would be "associate." To be effective in most cases, they have to work well together. David has outlined that pretty well. Anything more is just duplexed mats or computer code.

Thanks John. Yes I meant "visually", but I was kinda scared of trying to describe that in 2500 words or less to a novice without knowing the answer to "text or display?" or "what kind of Roman?". I was not sure how to describe the overlapping characters, like bottom of n, h and i, top of k and most of the diagonal l.c. But when the square at the x-ht are limited to f and t... what do you do? I also don't apply your suggestion to uppercase, where the italic length of stroke is also greater than roman, but composition often demands Roman and Italic in proximity too close for a height difference of any kind or the serifs clink. I got screwed on that one twice by clients mixing the two in words... :)

Cheers!

> Thanks John. Yes I meant “visually”, but I was kinda scared of trying to describe that in 2500 words or less to a novice
> without knowing the answer to “text or display?” or “what kind of Roman?”.

Mr. Berlow,
Are you suggesting a greater discrepancy between mathematical (actual) heights of a roman to an accompanying ital when the faces in question are display? Is this strictly a question of size?

I'm not a novice and was well aware itals should be a bit short to optically appear right; but I had considered that the amount of difference would be more or less when designing display fonts.

David: I also don’t apply your suggestion to uppercase, where the italic length of stroke is also greater than roman...

Nor do I. It is only the x-height optical difference that seems to cause a problem, e.g. when you have one word set in italics in the midst of roman text. Cap height and also extender length don't seem to matter much, and variance might be accepted as a stylistic feature of the italic. But if the x-height is not optically adjusted it can result in something that looks like a change in font size.

thanks everyone for your replies.

xtianhoff: though i agree with you regarding the display fonts, especially with hand-rendered feel. But i must have not phrased my question right. I was genuinely inquiring about the relationship of italics to the romans, not necessarily for display faces.

This question has been bothering me for awhile, and i am very grateful for all your thoughts on the subject.

>I’m not a novice...
My apologies, I made an incorrect assumption.

>Are you suggesting a greater discrepancy between mathematical (actual) heights of a roman to an accompanying ital when the faces in question are display?
I believe this is true of all alignments. The lowercase should be shorter, as most people know, the overlaps diminish and the italic will not appear taller amongst roman as it will in text.

Cheers!

I believe what David means by "overlaps" is what I'm used to calling "overshoot" (and "undershoot"). [DB, correct me if I'm misinterpreting.]

That's how I make sense of what he's saying. In larger scales, the amount of overshoot required for rounds to appear the same height as squares/flats is proportionally less than at small text sizes. This also affects the relative alignments and perceived heights of italic l.c. (where there are relatively few flats at x-ht, for most traditional styles, with f and t being the main determinors).

Also, I would say that the degree of slant makes a difference. The greater the angle from the vertical, the more the optical illusion that John describes will come into play (since the strokes become proportionately longer as the diagonal becomes greater).

-- Kent.

>>I’m not a novice...
>My apologies...

No, no apologies necessary. You hadn't been referring to me anyway and I only mentioned it so you wouldn't feel the need to "dumb down" your response.

Between the kind responses of both you and Kent I feel comfortable that I understand this isn't a question of Display vs. Roman so much as a question of scale. And I'm sorry, Vic, that I took your thread down a tangential avenue.

"That’s how I make sense of what he’s saying."
yes overshoots, sorry.

Cheers!

Related:
http://typophile.com/node/60506

hhp

> how popular is that work from Deberny or the German foundries?

Not very. I had long wondered why those fell totally off the map, and
I never believed it was because of unusability. I think I figured it out
when I started preparing my TypeCon-LA talk: what happened to the
German foundries was WWI; what happened to Deberny was Charles
Peignot, who killed off what he saw as the "stagnant" elements of the
D&P library, except he replaced them with superficial (if marketable)
display fonts; and what happened to Monotype and Linotype is that
-as they were replacing the "cold" foundries- they needed to mimic
the establishment (read: traditional Italics), just like Gutenberg at
the very beginning.

But history, shmistory - I'm happy to report that the
slanted-Roman is back, winning hearts, minds and money.

hhp

"but surely there must be a reason why garamond italic belongs to garamond family and not caslon."

Does it? If i'm not mistaken, Claude Garamond never did italics at all. Robert Granjon's italics are used as a reference in most contemporary designs, i think.

This roman-italic pairing thing is just dogma. Of course, if you are designing a text typeface, most people will expect you to follow the rules, one of them is to provide the basic four styles: roman, bold, italic, bold-italic. Anyway, you said it is a display face, so maybe the rules don't even apply. Anyhow it's always good to review dogma from time to time, least we became obsolete.

Talking about obsolete, i like the fraktur-roman pairing:

Looks groovy and makes sense, you know, German script for German text, Roman type for Roman text.