Paramond — an extreme display serif

Christian Thalmann

Steve: Better now?

Jan: Thanks! Good catch — I had adapted the /i and /j to the newly corrected x-height, but forgot to apply the same correction to the similar tops of /u. Better now?



This is all very helpful, I appreciate it.

Steve Gardner

Hi Christian... I think the fuller terminal looks better, so my view is to have the /a/ match the /c/. I think this would result in greater harmony with the /f/ and /r/.

Steve Gardner

You beat me to it! The /a/ terminal still looks a touch light to me in comparison.

ETA - I think this is because of a difference in the taper rather than the size of the terminal ball itself.

LeMo aka PatternMan aka Frank E Blokland

When you look at the g, the freedom of its design comes at a price. As I understand it, the top part is deliberately small and leaning to the left, and this generates quite some space at the right side of the character. The equilibrium of white space is clearly obstructed by the g in the pattern below.



Because the bottom part of the g is subsequently sticking to the right, this will result in problems when it’s followed by for instance j or y. Below is a blackboard demo from the time I lectured in the first year at the KABK (I found this photo on the web). The g is more balanced because its optical axis is more vertical. This axis can be represented by a line that connects the extreme of the top of the g with the extreme of the bottom part.



For a display type your spacing is quite wide IMHO. This is more something you would expect for point sizes meant for texts purposes. Basically your g alone will make it impossible to make the spacing more tight.

Christian Thalmann

Hi Frank,

thanks for the insight — you're right, the /g still feels like a foreign object in the flow of the font as it is now. I'm actually quite fond of my first rendition of the /g earlier in the thread, with its extremely low bottom, but it's just really impractical to use.

Your hand-written /g is very pretty and well balanced, but I guess it would collide with itself when geminated. I guess that could be amended by making the lower half narrower in a symmetric fashion. I'll try to import that design structure into Paramond.

My main concern, though, is how to make the /g fit into that concept of "tiny counters" that the /a and /e embody, and that is reflected to a smaller degree in /A/B/R. I can't seem to come up with a design for /g that both has small counters and fills up its space in the word in a balanced way.

As an expert in proportions and styles, do you have a suggestion how that problem could be solved? Or are the two requirements mutually exclusive?

As for the spacing, yes, I find the airy rhythm from wide spacing pleasing and in tune with the light cut of the glyphs. I figure it might be a bit tightened for the heavier weights I intend to add later. Or do you find it hurts the word shapes in the light weight?

Steve Gardner

Christian - what about something like the attached image as a possible solution for the /g/?

The top line is your original; the bottom shows slightly 'compressed' /g/s. You original counter sizes and shapes are retained, but the white space is reduced.

Christian Thalmann

Hey Steve — interesting proposition. Wouldn't that look weird among the deeper descenders of /p/q/y and the generous ascenders, though?

Steve Gardner

If you decide the /g/ (however gorgeous it is) doesn't work in its current form, you have little choice but to modify it in some way. Maybe you can find an alternative, equally attractive form that will address the white space issue. If not, maybe shortening the /g/ will work (maybe less than I suggested?), if you're also able to balance it by shaving a bit off of the /p/, /q/ and /y/ descenders. Given the typical differential between ascender and descender values, these characters will probably stand a bit of tinkering without having to do likewise with the ascenders.

Christian Thalmann

I noticed Frank's design has its neck rather far under the head rather then looping out to the side, as I had it in my previous designs. I had also made the head itself rather flat based on previous comments, which also generates more whitespace beneath it. I stole the round head, straight neck, and overall balance from Frank to make this new /g, which I like rather much... and it even plays reasonably well alongside /y, especially with a hint of positive kerning:

Craig Eliason

I dunno, that just looks very droopy to me.

Christian Thalmann

Hmmm... I see what you mean. The new equilibrium and economy of space comes at the price of reduced æsthetics. The "eggy" version up there look much prettier. I wish I could make it work instead.

Frank, what exactly do you mean by "the equilibrium of whitespace is obstructed" in the previous version? It doesn't look mis-spaced to my eye; is the problem only that I can't make the spacing tighter (which I don't foresee for this version)?

Other suggestions...?

Christian Thalmann

Maybe a wider version of that last design would look less droopy?

Or with smaller heads:


Although those /g barely have enough space next to each other again, making the font "incompressible". If it's going to be like that, maybe I should stick to the "egging" design after all, which seems to make the best overall æsthetic impression. I've tweaked the sidebearings a bit, maybe that helps address the impression of unbalancedness?

Craig Eliason

Those look pretty good to me.
How do gj and gy look? If gg is literally the only problematic pairing, a contextual alternate or ligature could be a reasonable solution.

Christian Thalmann

Wait, could this finally be it? It's similar to the upper two in the previous post, but the lower counter is narrower and taller. It's neither droopy nor ugly, less asymmetric than the "egging" design, and uses its space sensibly...

Christian Thalmann

Craig: The /g/y pairing appears in all the samples. ;o) The /g/j is completely unproblematic, since I'm using the tapered Garamond /j.

It's not all about collisions, though. Some of those designs may hurt the rhythm of the face with their whitespace.

LeMo aka PatternMan aka Frank E Blokland

Christian: ‘[…] what exactly do you mean by “the equilibrium of whitespace is obstructed” […]?’

I think that the original g took/generated too much white space within the x-height. In the image below the first three rows are made with this version. In the fourth and fifth one the right side bearing of the g is narrowed (which would result in a collision of the bottom parts) and the left one slightly widened. The last one shows Adobe Garamond for comparison (of which the fitting could be improved also by applying the original spacing system that I measured and distilled at the Museum Plantin-Moretus).

Martin Silvertant

This is also why I recommended a more horizontal top of the loop. It wouldn't necessarily be a proper solution, but it reduces the excess white space in between the top and bottom parts.

Also, moving the top part to the right so it's centered more above the loop would have compensated for a bit of the access space on the middle right.

I think the new /g works, though I would still make a contextual alternate for g_y and g_g.

Martin Silvertant

By the way, I don't know what typeface this is but you might like to see it for its characteristic A.

Christian Thalmann

Thanks for the illustration, Frank. It makes sense that the x-height should be the important region to evaluate.

Martin: I'm not sure I'll make a contextual alternate or ligature for /g_y and /g_g — while I generally enjoy such things, it can also introduce an unwelcome element of inconsistence, in particular in the geminate /g_g. Well, maybe if it were so subtle people wouldn't notice it at first sight...? In any case, I think the current combinations don't work any worse than they do in professional Garamonds.

Nice catch on that /A! Very pretty. Though I really don't care for that deconstructed /S.

Martin Silvertant

In any case, I think the current combinations don't work any worse than they do in professional Garamonds.

Perhaps not, but I don't see why Garamond is your goal and restriction in terms of quality. For book typefaces I always try to improve the spacing by introducing subtle contextual alternates in the same vain as Mota Italic's Vesper: http://www.motaitalic.com/typefoundry/fonts/vesper/contextual-alternates

With the contextual alternates off Vesper certainly looks good, but the contextual alternates push the consistency of rhythm and texture just a little bit further. Perhaps that's not what you're looking for in a Garamond, but I think it's worth to think about. I certainly wouldn't make contextual alternates that look obvious though. In fact, you shouldn't notice it at all, but you should notice a slight improvement in the rhythm of the text. It's probably just about ever so slightly decreasing the width of the loop of the left /g in g_g or move the top part of the left /g a bit to the right. Right now you have even spacing but with tighter spacing you're going to run into problems with g_g.

Christian Thalmann

Vesper is very pretty indeed, and I do like the idea of subtle OpenType wizardry for improved rhythm. I'll keep it in mind for when I have a first complete character set done. Do you have any experience with such things yourself, Martin?

Meanwhile, how is that /Germandbls working?

Steve Gardner

That /Germandbls/ is delicious, Christian. Out of interest, do any other capitals terminate with a ball?

Also, does the stroke on right side of the /A/ get thinner toward the baseline, or are my eyes playing tricks on me?

Martin Silvertant

Do you have any experience with such things yourself, Martin?

I'm currently still working on a sans typeface for a client where I've included some automatic ligatures using alternate versions of f, t and r. The change in the alternate /t is especially subtle, but it's an improvement nonetheless. For serif typefaces I like to include a few more contextual alternates because the forms are more dynamic, so depending on the letter combinations I find sometimes more subtle changes are required to make the letters work as optimally together as possible.

Perhaps it's not so much necessary for the typefaces as it is necessary for me.

Really beautiful eszett.

Also, does the stroke on right side of the /A/ get thinner toward the baseline, or are my eyes playing tricks on me?

Optical distortion or not, I'm seeing it too.

Christian Thalmann

You guys were right about the /A — the top was 2 units too wide. Nice catch! I see my eyes still have a lot of training to go through.

Steve: You're right, the ball terminal is out of character among the caps. I've replaced it to match the /S instead.


Martin: So your /f/t/r retract their protrusions when followed by, what, /v/w/y? What other combinations to you feel require special attention in a serif? (Or is that a trade secret of yours...?)

Jack Jennings

Seems like B, cap ß, and S are pretty regular, when they could certainly play into the theme of small counters. You've already shifted the counter really high for A, why not consider doing the same for S and B (smaller upper counter, larger lower) and cap ß (tighter "bowl"?)

Martin Silvertant

So your /f/t/r retract their protrusions when followed by, what, /v/w/y?

Actually the alternate /f is for all the common ligatures like ff, fff (not common, but okay), fi, fl etc. Right now I have an alternate /t for an automatic t_t ligature where the crossbar of the alternate /t is a bit shorter on the right side. It's just to bring the two letters a little bit closer together while keeping the gap in between the two crossbars open.

Right now I have no alternates for v/w/y combinations as there is no need for it due to the lack of serifs.

What other combinations to you feel require special attention in a serif? (Or is that a trade secret of yours...?)

It probably differs per typeface and its intended purpose. During a study I did into horizonal dimensions and rhythm of typefaces I noticed going from humanist (Jenson) to garalde (Garamond) to transitional (Baskerville) the proportions become more constant and mechanical and vertically less and less sticks out; Baskerville consists of pretty neat blocks. As you probably know the garaldes are preferred for optimal reading, which is due to the diagonal weight distribution, the medium contrast and probably the horizontal dimensions and its rhythm influence the reading experience as well, though I haven't been able to conclude to which extent there should be a constant rhythm versus variety. Variety in spacing helps distinguish between letters. I personally see the process of adding contextual alternates as applying the finishing touches. You may not necessarily do it at the end of the process, but what I mean is that with contextual alternates you can apply the final subtleties you feel your typeface needs. This can be to make the spacing more consistent, but you might as well do the opposite. For example, if you design an /r with a relatively long arm to go before /n, you help the reader in avoiding interpreting r_n like /m.

What combinations do I think require more attention? I don't have specific pairs but I just look at my typefaces and decide where I want a letter to interact with an other letter in just a slightly different way. It's probably unnecessarily obsessive, but I don't usually take it too far. An example of taking it really far is this:
Let's say you condense /T a little bit to compensate for the big spaces underneath its arms. Let's say you have an S with a small top and a big lower part that extends beyond the top terminal, like a fat swan. I can imagine in the combination STA you might make an alternate /T with slightly extended arms to compensate for the space in the top part of /S and the big space /A always has.

You could make the design colder or more lively depending on how you bring emphasis with contextual alternates, but I often feel like striving for too much perfection kills a typeface. It becomes cold and lifeless. I often want my typefaces to have little quirks and inconsistencies that bring some warmth, as long as they're not obtrusive. I'm still designing typefaces in a way I feel is too rigid, but lately I've been looking at late renaissance typefaces a lot which often have little quirks as well, if not by design then often due to the printing.

Christian Thalmann

Jack: I'll consider that for a series of alternates, but I don't expect it to work for the defaults. The small counters I currently have still preserve the expected Garamond overall shape of the glyphs, whereas for /B/S/Germandbls, it would upset their gross proportions.

Martin: Thanks for the insights! I made a /g_g, /g_y, and /g_g_y ligature, along with some accented versions — does that work?


Also, as usual, I have trouble judging whether my /Schwa/schwa work. I like the way they look, though:

Steve Gardner

Christian... I know nothing about the design of a /schwa/, so I may be wide of the mark here, but the terminal on the lower case /schwa/ looks out of place. It looks, in context, as though it should be either a ball or else the stroke should terminate in the same way the /e/ does.

The cap /Schwa/ looks great to my eye.

Christian Thalmann

Hi Steve,

to my knowledge, a serif on the top right of the /schwa is quite customary, given that the stroke begins rather than ends there. Ball terminals, too, strike me as more of an ending feature that a starting one. The one in /a is an obvious exception, but perhaps that's another reason not to use one for /schwa as well, since it would increase the (already considerable) risk of confusion with /a. The serif I chose fits very well with /z, though, which I think is the most natural role model for the /schwa's serif, since it starts with a horizontal motion from the top left corner as well.

Steve Gardner

Yes, that makes sense, Christian. The only other observation I have is that, when comparing the angle of the serifs on the /Zs/ and /Schwas/, there appears to be agreement in the lower case but not in the caps.