what makes it text vs display?

Amado's picture

Hope the title wasn't misleading.

So I was studying text faces and display faces for a face I want to design, theorizing about xheights and things like that, when suddenly... see attached.

(I know you'd never set large blocks of text in a bold, but bold makes this illustration more clear.)

I'm wondering about vertical metrics. What I wish to understand is, what makes Parkinson a display font that you'd never use for blocks of text, and what makes Whitman such an awesome text font? Also, more fundamentally, I wish to understand what the hell point size has to do with the actual vertical metrics of typefaces?

Parkinson Medium @72pt and Whitman (Roman or Bold) @88.5pt have damn-near the same xheight and cap-height. @72dpi, the Parkinson @72pt measures 72px from desc-line to asc-line, while the Whitman @88.5pt measures 83px from desc-line to asc-line.

I know the answer is "that happens sometimes" (QED), but my hope is to understand what's happening.

IS there a true-ish statement that can be made along the lines of "Usually, about ____% of the time, in a good text face the cap-height is about ____% of the point size and the xheight is about ______% of that, and the ascenders and descenders do such-and-so..." ...?

AttachmentSize
parkinson-and-whitman2.jpg26.74 KB
pica pusher's picture

I wish to understand what the hell point size has to do with the actual vertical metrics of typefaces?

It doesn’t really have anything to do with it. The point size defines an imaginary block of vertical space, which the type designer has divided up to suit his whimsy. Within that em you'll get an ascender height, cap height, x height, baseline, and descender height (often even in that order!), but they could be in any proportion to one another and in any proportion to the em. What remains constant as you change the point size are the proportions.

For example, here are an M from Perpetua, and an M from Calisto, both at the same point size. The squares have height and width equal to that point size.

The designer might use only a tenth or a hundredth of the vertical space available (although, that would be a painful way to work).

Why were you interested in Parkinson and Whitman for studying text v. display faces?

Amado's picture

Parkinson for pedigree (1470 by way of ATF Jenson "on acid") and cultural associations (Rolling Stone, rock-and-roll).

Whitman because it is a great face designed (and successful at) body-text -- plus I find it beautiful and well-made.

I'm looking at other text faces too, to try to get a sense for what makes them work well. I thought it was, in part, the proportion of cap-height to xheight. The attachment in my original post blows that hypothesis away.

edit: Something else just sunk in. XHeight has been a red herring for me. Alone, it doesn't really tell you anything about the vertical proportions of a typeface. If you don't also know CapHeight, AscHeight, and DescDepth, you don't really know a face. Also: vertical proportions don't really tell you anything about whether a face is a good text face. Contrast Sentinel with Whitman with your favorite cut of Jenson.

So, what does?

charles ellertson's picture

I think I' going to regret this...

Your mathematical/mechanical comparisons are the wrong tree. Take a look at Trump Medieval. A good text font. Now look at Adobe Garamond, another good text font. Sorta knocks the wind out of the cap heigth/ascender heigth/x-heigth/descender depth/small cap height as a matter of ratio, doesn't it?

The one think you can say about text type is it has to be comfortable for a 300-page book.

& my current rant, which is not enough attention is paid to how the type appears. Printed offset, on a certain kind of paper -- coated versus uncoated. Or printed on a laser printer on an 8.5 x 11 sheet of office paper. Or as pixels on a screen. And, uh, which screen?

Whitman, printed offset on an uncoated sheet, is a bit light. Kent knows this. It is worse on a coated stock. It is beautiful printed with toner, as with a laser printer.

Never used Parkinson. Doesn't look very good to me at

http://fontzone.net/font-details/Parkinson-Medium/

but that's not a fair way to observe the font. I think some of the letterforms suspect, but in the final analysis, it's the use that counts.

It will probably print like Merlo, I suspect. Merlo doesn't print well offset, on an uncoated sheet set larger than about 10-point (& that's an iffy combination), but is wonderful on a coated paper where the ink doesn't spread much.

http://www.felicianotypefoundry.com/cms/fonts/merlo

The thing about type is, it hasn't much value until used. So look at that.

Amado's picture

Charles, you don't have to regret it on my account. All you'll get from me is a sincere thanks for taking the time to point me in some fruitful directions.

What I'm left with is that my question has pretty much devolved to: "hey y'all type designers, how much room do you typically leave above your ascender line and below your descender line?"

After that, I just have to design something that I think/hope looks nice, and try it out on as many media as I can.

Nick Shinn's picture

Approximate your vertical metrics to those of a face which you consider to be a benchmark for the genre to which the face you are designing belongs.

For instance, for a news text face, Utopia; for an oldstyle book face, Minion or Garamond.

J. Tillman's picture

Charles_e,
With regard to Merlo, are you saying that it looks too dark on uncoated paper, or is there something else going on? And so would it look even more too dark on a personal laser printer?

And the $64,000 question... How can you judge if a font is going to be okay, lets say printed offset on uncoated paper, by looking at the laser-printed copy. People with limited or no experience are making these judgments, for paperback books. Is there any way to judge, particularly for newer fonts? I feel like I'm flailing around here, but someone has to be the first to look at a font and say "yes it'll look okay, let's do it".

Thank you.

Amado's picture

Nick,

Of course. Solid suggestion; in-front-of-my-nose solid. I should be ignoring what http://www.cs.tut.fi/~jkorpela/x-height.html tells me and just study the actual proportions of /H/x/l/p/. And look at the weights too, b/c vertical metrics alone really mean little. Then just shoot for the moon and see what happens.

charles ellertson's picture

Mr. Tillman:

Yes, Merlo will look very dark if set over 10-point and printed on an uncoated sheet. But even 11-pt will look quite good printed on a coated stock.

And the $64,000 question... How can you judge if a font is going to be okay, lets say printed offset on uncoated paper, by looking at the laser-printed copy.

Well, experience helps, but there is no substitute for actually making the test. Worth reading is

http://www.amazon.com/Counterpunch-2nd-Sixteenth-Designing-Typefaces/dp/...

It may seem out of date, but the process one goes through, looking at how your "punches deliver ink when pressed into paper" has a modern equivalent.

People with limited or no experience are making these judgments, for paperback books.

Tell me about it. And they look like $hit.

Is there any way to judge, particularly for newer fonts? I feel like I'm flailing around here, but someone has to be the first to look at a font and say "yes it'll look okay, let's do it".

Again, experience is a good guide. But try this: take a magnifying glass and look at the characters as they are printed on paper, using the technology appropriate. Look how the ink spreads. Notice that the contrast (fine strokes to heavy strokes) changes. Or if you own the typeface, make a scan of the printed letters (laser versus offset/uncoated versus offset/coated) and bring all those up on the screen next to each other -- big, so you can see how the ink spread is different on the offset, and character formation & inking different on the laser -- and finally compare the scan images to the glyph that was used in the typeface. Should be an eye-opener.

You can see the same thing on the screen, sort of, by looking at text rendered at 10-11 point on a new I-pad versus a Kindle Fire. You might get an interesting answer in which is "better." I tried this with Utopia. The I-pad has the most faithful reproduction, but to me, the type looks better with the Kindle Fire. And who knows what Slimback was anticipating when he set the contrast of the glyphs? With this font, maybe the accuracy of the I-pad is a problem.

kentlew's picture

What I'm left with is that my question has pretty much devolved to: "hey y'all type designers, how much room do you typically leave above your ascender line and below your descender line?"

Amado — FWIW, when I designed Whitman I was under a good deal of historical influence from my study of Mergenthaler Linotype faces. In hot-metal, there were limitations for how close you could get to the top and bottom of the body without risking casting problems.

Somewhere I have a chart with the production specifications that Linotype used for the various sizes. I think the average head/foot space was around 0.004 inch. But I don’t think I had that in front of me at the time, and I believe I undershot even that spec with Whitman.

Anyway, by any standard, Whitman is rather small on the body, around 930 (not including f/g over/undershoots). Nowadays I tend to work to around 960–80.

Of course, there’s no physical limitation with digital type production, and no one sets solid any more anyway. So these things are more about convention and expectations.

In my opinion, the overall size on the body has less impact on Text vs. Display questions than the rest of your internal relationships — x-height, cap, ascender/descender, stems, set widths, fitting, etc.

William Berkson's picture

Traditionally, type design intended to be used at smaller sizes have the contrast between thick and thin reduced, compared to a similar design at larger sizes. This is because at small sizes you seem to be bumping up against some kind of visual threshold problem, and the thins seem to look a lot thinner, compared to the thicks, than when the same design is viewed in a large size.

This is already true in normal book text sizes, say 9-12 point. Below 9 point, designs generally 'cheat' on visual size by increasing the x-height, and having shorter ascenders and descenders. This makes the physical size of details of the design big enough that the eye can still recognize them without too much discomfort. For the similar optical reasons, the designs intended for small size also increase the inter-letter spacing.

There are other more subtle effects, but these three: contrast, x-height and letter spacing are consistently attended to in traditional letter press fonts, and in current digital designs sensitive to optical size.

Personally, I think the readability of text type benefits from some of these adjustments at larger text sizes (10-12 pt), which is why I made my Williams Caslon Text the way I did.

Amado's picture

Kent -- dumb question #1: that's out of 1000, I presume? dumb question #2: does that leave enough room for diacritics? (I suppose I will learn that when I try it.)

William -- thanks for your reply. Your work on Williams Caslon Text is another inspiration to me. In particular, your oft-repeated reminder that there's no such thing as a "true" revival... I just gotta do what seems right to me, and the result will be judged on its own merits (if any).

p.s. I adjusted my vertical proportions per Nick's suggestion. It's time for me to draw some curves on my pixelated sketches.

kentlew's picture

#1 — Yes. Sorry for the shorthand. Easy enough to interpret as a percentage and translate to a different em size, if you want.

#2 — Depends upon how you set your x-height, cap-height, and how you handle your diacritics for each. Fortunately, there is no physical limitation to the body any more. I doubt there are many ĺ (lacute) glyphs that manage to stay in bounds.

Syndicate content Syndicate content