Good questions. I think the reason it’s (usually) only the x-height that’s adjusted is that the adjusting other parts (cap height, ascender height) would be more noticeable. The whole point is for it not be noticeable, that the different weights simply appear to be the same size. It’s sort of like overshoots.

Also: If the range of weights is narrow (i.e., they don’t get very bold), it’s just as well to keep the x-height constant.

“adjusting other parts (cap height, ascender height) would be more noticeable”
But why should this be?

Below is the “h” from Adobe Caslon Pro and Adobe Garamond Pro, each in Regular and Bold. The Caslon compensates both at the x- and the ascender line, the Garamond only at the x-line. Doesn’t the ascender look shortened in the Bold Garamond?

Yes, the ascender on Adobe Garamond looks shorter, but but I don’t think that’s necessarily a problem. The bold is inevitably going to change shapes, so question is how it looks in context—eg in a line with the bold as a side head to a regular text line, or as a title, etc.

“The bold is inevitably going to change shapes”
Yes, but my point was to not make it change apparent sizes too much.

Now you can argue that ascenders don’t enclose volumes the way x-line features do, so they’re maybe not as relevant wrt the resultant apparent glyph size. But then what about the cap line?
Garamond Pro again. Why would you want to compensate in the lowercase – the x-line shifts by 6 units in this case – but not in the caps, where the Bold looks clearly smaller?

I suspect it might really have something to do with usage context rather than just optical effect, but I’m not getting it.

If you show pairs of glyphs out of context, the differences will be more apparent.

The reason I think this is done only in the short part of the lowercase has to do with relative scale. When you make letters bolder, the small letters take on much more weight for their size compared to the caps. So the optical effect of adding weight (making the shapes look smaller) is more pronounced in the small letters. And letters like the “s”, “e”, and “a”, which have a lot going on in them (three horizontal strokes each), are more compromised as you add weight. Increasing the x-height as the weight increases helps alleviate both of these problems.

I’ve done this in several of my font families to a greater or lesser degree because of the reasons stated above. The thing that always worries me is the idea of someone creating a logo that uses two different weights side by side, where the difference in x-height would become apparent. Even so, I think it’s the best way to handle a range of weights. I wouldn’t mess with the cap height, ascenders or descenders because I don’t think the effect would be as great as with the x-height.

Well, I suspect the whole exercise of bumping up the x-height a bit doesn’t have so much to do with making it look the same x-height as the regular. In my limited experience I found you just don’t have much much room to maneuver in the between the x-height and base line—particularly on the double stories a, e, s, g—and pushing up the x-height a bit helps you to distort the shapes of the regular a bit less.

Since we don’t normally mix bold and regular side-by-side, I’m guessing that Slimbach preferred to keep the same vertical metrics on the ascender and cap height, so the white space between lines wouldn’t be crowded.

edit: Cross-posted with Mark. He has had the same experience, which makes me think this may be the story.

“The thing that always worries me is the idea of someone creating a logo that uses two different weights side by side, where the difference in x-height would become apparent.”
Hmm, but don’t you think that a marked difference in apparent size would be much more jarring than a mathematical, barely perceptible shift of the x-line (since we’re talking about a range of maybe 5–10 units)? FWIW, it was actually on jobs like that that I was taught to increase the point size on the Bold if it “looks” too small.

Also, one would hope that graphic designers who use text fonts for logos should be aware that some things may look funky up close… maybe too idealistic.

Also, what about shifting the baseline? Doesn’t anybody do this?

“Since we don’t normally mix bold and regular side-by-side”
Um… how do you normally use Bolds?

And about this:

“Well, I suspect the whole exercise of bumping up the x-height a bit doesn’t have so much to do with making it look the same x-height as the regular.”

Should it not?
Meaning, why should one not want/try to achieve equal apparent sizes?
The apparent size difference, when there is no x-line offset, tends to look like a “mistake” at least to my eye. Am I seeing things?
Or does it not matter for some reason?

>side by side

I meant we normally don’t have a bold and regular adjacent to one another in the same word, as in your examples above. They are separated by a word space. I think that is also the concern Mark mentioned, that in a logo the could well be adjacent, and hence your concerns would be relevant. However, in the case of a logo, things can be custom shifted or redrawn.

In the normal case, to do the comparison, you need to have eg a side head in bold, and the sentence continuing in regular. In the case of Adobe Garamond, the ’bold’ relative to the regular is actually the semi-bold, so for a bold in the middle of the sentence that would be the comparison.

I suspect that there you will find that baseline shifting of rounds may create more problems than it will solve. I haven’t tried it, and it would be interesting to see—but in any case I think that’s the more important comparison, not adjacent glyphs of different weights.

I make the x ht the same for regular and bold of a text design. Overshoots could vary, but not x. If you feel the need to make a bold taller to look right with regular, there may be something else wrong, which could be as simple as you not looking at text, or as complex as the bold contrast is too low or the bold bodies too wide. Shifting the baseline is a size mastering technique that does not help companion faces stick together visually at a common size .

Cheers!

{Working backwards...}

So/but if you increase the x-line overshoot for a darker cut to make it not
look shorter, shouldn’t one also do that for the baseline (and elsewhere)?

I think Nina has observed that -as a rule- people don’t do this.

hhp

Thanks for the input David –
“Overshoots could vary, but not x”
Why would you exclude the option? Is it because of the loss of optical consistency that Mark mentioned?

Hrant, good question re the baseline shifting/overshoot. I just haven’t found any examples of fonts that do this successfully, at least not ones that seemed meaningful (sorry Miso). But I haven’t looked around for years either. So I’m curious.

1. As far as I know, there is no statistical survey.
I’ve done it on a few faces.
One reason is that when I did a lot of editorial design, I often increased the size of bold subheadings because I felt it looked too small next to the Regular. So that feature of my type design is based on my own taste as a typographer, which I’ve subsequently rationalized.

I would say that apparent size is dependent on more than one indicator or cue:

- Distance between extrema of letter outline
- Distance between optical centre of horizontal strokes
- Distance between extrema of negative counter

The gestalt of height is a synthesis of these height assessments.

2. Compare the density (overall tone) of the x-height zone with that of the median-to-cap zone. If the x-height is constant from Regular to Bold, then the density of the x-height zone will increase with font weight (in all but the didone genre), giving a relatively bottom-heavy appearance. Raising the x-height counteracts this. But raising the cap or ascender line as well would not.

3. Because letters stand.

>Why would you exclude the option?

Because it doesn’t work very well for a companion text bold. You can try and then exclude it or include it yourself.

Cheers!

This becomes an issue mostly in sans-serif designs or designs with very low contrast. When the horizontal elements of a letter also get heavier, the counter sizes suddenly shrink, unlike with serif designs, where they typically just get narrower, which just accentuates the height of a counter. You’ll see much more of this x-height variation (look especially at Futura) in sans designs than in serif ones.

Bingo Carl. I’m currently living in monoline land so that might explain [some of] it.

I’m still confused, for the record. This thread, as well as the fonts that I looked at, sorta gives the impression that some people do it and others don’t; some designers think it helps, others think it’s misguided; and even the purpose isn’t clear. :-\

“3. Because letters stand.”
But Nick, how does overshoot feature in that theory?
And if overshoot is allowed, why not let darker weights dip a bit further down – so that they’ll look like they’re the same size, as with an overshoot?

“You can try and then exclude it or include it yourself.”
David, I have, and before I started this thread. I like it, and my eye thinks it helps. I’m just not sure *why* I should be doing it and/or why not, and then it’s hard to just trust my [inexperienced] eye.

But Nick, how does overshoot feature in that theory?

It doesn’t. I’m thinking primarily of side-heads in Bold, with Regular on the same line, or boldfacing in Regular text, to pick out the names of celebrities or companies. In these situations, the reader’s expectation is that the letters stand on the baseline grid, not that they are hung on a harder-to-conceptualize optical centre of x-height, with a baseline that is stepped.

>This becomes an issue mostly in sans-serif designs or designs with very low contrast.

Ahhha. Fooled again. The examples were old style text faces.

Cheers!