Scarbo — an attempt at a text face

Jan Pietkiewicz

It is with some trepidation that I submit my design for critique, since several of this forum's esteemed contributors  expressed some strong opinions about essentially derivative typefaces, superfluous rehashes of the old-style model, as well as beginners punching above their weight. I therefore feel obliged to preface this by saying that I am not a type designer and that I don't have type designer's education, perspective or ambitions. While working on my design I wasn't thinking in terms of innovation, self-expression or "cultural contribution". What I am is a moderately competent type user and all I wanted to achieve was a modest text typeface for my personal use which would have all the features I like and none of the features I don't. Even though I realize such aim might not be particularly laudable, I do hope it can be accepted as legitimate.

Scarbo (provisionally named after the sadistic dwarf from Bertrand's Gaspard de la Nuit) was conceived as a book typeface suitable for body text set at about 9 to 11 pt. Since the books I specifically designed it for may very well never be published, I can also imagine using it for correspondence, business cards and the like. At the present stage I was concerned only with the support for the Polish language.

I believe the design bears its influences on its sleeve, but I will enumerate them briefly. The roundness and generous width of the characters, together with the elongated, flat-ended ascenders and descenders were derived from the family of faces inspired by Griffo's type (Bembo, Dante, Iowan Old Style etc.). Since literal Renaissance revival was not my goal, I turned to Plantin and Times for several more "contemporary" (by the way of the early 20th century) letterforms, including the large-bowled /a and large-eyed /e, as well as several other glyphs (e.g. the /?). While the use of slab serifs places my typeface in the tradition of Joanna, Whitman and Scala, I wanted the effect to remain subtle, coming off almost as an afterthought; Štorm's Jannon 10 provided the initial impulse.

One of my main goals was to arrive at a streamlined, smoothed out and slightly generic look. I wanted to avoid any too-obvious vestiges of handwriting (beyond the broad-nibbed terminals). Whenever possible, I went for the simplest or even "naïve" letterforms: /k's and /K's arms and legs meet at the stem, /M has parallel stems, /R – a straight leg, and both /w and /W – only two serifs.

The only eccentricity of Scarbo that isn't a result of my lack of skill but rather of a conscious decision is the use of angled serifs at places where most humanist slab serif designs would have straight vertical ones, namely /C, /E, /G, /L, /s, /S, /z and /Z. This decision was partly dictated by the fact that angled serifs in /s and /z play together much more nicely in Polish digraphs c_z, r_z and s_z:


At the current stage the font is spaced (using Frank Blokland's cadence method with some ad-hoc modifications) but unkerned. I plan to add small caps at some point, but first I'd like to make sure whether the uppercase is at all acceptable. I also have a vague vision for the italics, but I'm afraid it will be quite some time before I arrive at anything substantial.

Since I lack the benefit of formal type design education, mentors or even type-savvy acquaintances, I would be eternally grateful for any comment or advice on my design. As this is my first project of this sort, I feel rather uncertain about virtually all aspects of it, from the proportions and contrast down to the shapes of individual letters. The glyphs that are a particular source of concern for me at the moment are the /s, /S, /T, /Z, /z, /3 and /5.

Thank you in advance for any help or opinion!

Lastly, because I believe it is incumbent on a dwarf clumsily trying to climb onto the shoulders of giants to at least acknowledge the names of said giants, I would like to list the typefaces that to the greatest extent have influenced my decisions when working on Scarbo. Even if it wasn't my intention to replicate any of them in particular, their aspects inform my idea of what constitutes a good text font and it would be remiss of me not to mention them here:

Adobe Text, Alia JY, Aragon, Bembo, Concorde, Dante, Erato, Fournier, Berthold Garamond, Iowan Old Style, Jannon 10, Joanna, Kings Caslon, Logica, Lyon, Minion, Plantin, Poynter, Rawlinson, Rotation, Sabon, Sabon Next, Scala, Times Ten, MVB Verdigris, Whitman, Yale

Vasil Stanev

Never understood the aversion to using other peoples fonts to create something original - it can't be original of it hangs too much on it's mothers' apron strings, and using other people's Bezier and metrics information could get you into actual legal trouble - but past that it is legit way to understand how typefaces function. I feel that, at present, there are a limited numbers of ways one can achieve an original text or sans face, or any typeface for that matter. There are already so many players on the field that specimens are bound to look very similar. Most type designers start with a modular or geometric font and work their way up to a sans and from there - to a serif, you chose the other way and mashed serifs into something new. 

When you use other people's designs, you got to make the result not only effective, but more pronounced. Always look for size, proportion and contrast. The upper serifs of C, S etc. for example are too small and will get lost at smaller sizes. b, 6 and 9 look out of place, 3 is slightly leaning to the left. Always keep in mind that width should be consistent: you have a smallish a right before a very big b, a tight s and a wide, generous u. The tail of Q could go as low as the one of q. f looks like too much of a solo player with that hat. k should be redone. ew should have the stroke more pronounced, it should go out more boldly. Work on the ą where the ogonek meets the stem. g is strange, looks too much like a san-serif Times. Try some variants, especially on the spur. Ł -make the stroke bolder. 
All in all, a nice font, nothing special, but it does not have to be. Body serifs keep a low profile, so I consider the job done.

One more thing. Text fonts always go in quartets: a book, a bold, an italic, a bold italic. Create the other three players and you will see the inconsistencies yourself. What might have worked for the book weight might not work for the bold or for their interpolation, so there is such a thing as retrofitting the older design to the newer. Only when you have done this further exercise can the job be judged in full, IMO.

To sum up: good way to teach yourself some principles, a bad way if you want to be original.

Craig Eliason

If this is a first original design I’m impressed!
Your curves overshoot too much I think—look at the letter height of the first 5 letters of the big Polish word you posted. 
The tightness of the curl in /a/‘s upper terminal doesnt fit with the rest. 
/f/ ascender looks a bit flaccid. 

Jan Pietkiewicz

@Vasil Stanev, @Craig Eliason – many thanks for your very kind and helpful replies. I really appreciate your taking the time to look at this.

Because Vasil mentioned "using other people's fonts" twice, I would like to make it absolutely clear that while I make no claim to unqualified originality, and while it was important for me to "call it what it is" and to be upfront about the fact that my design was heavily referenced and that most of my choices can be traced back to specific sources, my process involved no mindless (nor mindful) tracing or stealing existing Beziers or metrics. I hope my initial post didn't give anyone the wrong idea.

I humbly take most of your points on board, so I will refer only to those suggestions of yours about which I'm not quite certain.

Vasil Stanev said:

b, 6 and 9 look out of place

I will take a look at /6 and /9 (although after many tries I more or less resigned to the fact that I'm unable to make digits which wouldn't look out of place), but what precisely is wrong with the /b? I constructed it using the same principles as in /d, /p and /q, using the form of the /c. Avoiding the spur was intentional and I would prefer to keep it this way.

Vasil Stanev said:

you have a smallish a right before a very big b, a tight s and a wide, generous u

Arriving at present proportions involved a lot of back-and-forth adjustments until I had to tell myself to stop for the fear of losing mental balance. When in doubt, I nearly always opted to err on the wide side. That being said, I now completely agree with you about the narrowness of /s. But can someone confirm that the /a is indeed smallish? I remember trying to make it wider at some earlier incarnation, but the result looked very unwieldy and I immediately reverted.

Vasil Stanev said:

f looks like too much of a solo player with that hat

Craig Eliason said:

/f/ ascender looks a bit flaccid. 

I realize this and yet I'm not sure if I can deny myself this little indulgence. I have a soft spot for Bembo's pronounced arched ascenders and descenders (on /f, /j and /J) and wished to implement them here in an almost parodic form.

Craig Eliason said:

Your curves overshoot too much I think—look at the letter height of the first 5 letters of the big Polish word you posted.

This is something I've been concerned about for some time now and your observation confirms my fears. Would it be right to assume, however, that even after the correction the overshoots should look slightly too hefty at larger point sizes (since the font is intended for body text)?

Vasil Stanev said:

Text fonts always go in quartets: a book, a bold, an italic, a bold italic. Create the other three players and you will see the inconsistencies yourself. […] Only when you have done this further exercise can the job be judged in full, IMO.

I'm fully aware of this, but I decided to ask for critique already at this point, encouraged by @Thomas Phinney's statement to the effect that it is better to seek advice early. Your present comments already helped me to eliminate some mistakes which otherwise I would have carried to other variants.

Again, thank you so much for your input. It is already enough to keep me occupied for quite a while. I will report back when I have something new and exciting to share.

joeclark

I therefore feel obliged to preface this by saying that I am not a type designer and that I don’t have type designer’s education

I like you already.

As various type designers have confirmed to me in person, here is a commonplace scenario. You open a thick envelope one day and find inside a well-typeset letter that describes the writer as a high-school senior who has followed your work for many years, but, while maintaining full respect for same, the writer feels you missed an important precedent from the early 18th century that would have informed the design of [typeface that sold millions of dollars’ worth of licences].

I hope you don’t mind, this letter continues, but I did a quick mockup – this means six months’ after-school beavering – of what at the very least the roman and italic would have looked like if this important tranche of typographic history had not been forgotten.

I really don’t mean any offence, the letter continues, but if I’m reading [scion of typography]’s original notes in Latin correctly, my improved design here is historically justifiable.

Yours faithfully, the letter, which contained no neutral quotation marks, let alone emoji, concludes.

So you’re really on the right track here, Jan. Plus I like your shit.

Christian Thalmann

Nice job! I find the figures particularly clean and striking — hard to believe you’re not happy with them! Only the /3 looks a bit unbalanced to me. 

I’m a fan of generous flags on /f, so I’m glad you’re not backing down there. I recommend drawing a «retracted» version of it for use before closing quotes, brackets, capitals etc., and even as a fallback solution to the ligatures. 

Btw, the /f_h-style ligatures strike me as a bit compressed. They don’t appear in your Polish text; I’d test them out in Latin. 

Have you considered making the typeface available open-source?

Kent Lew

While working on my design I wasn't thinking in terms of innovation, self-expression or "cultural contribution". What I am is a moderately competent type user and all I wanted to achieve was a modest text typeface for my personal use which would have all the features I like and none of the features I don't.

I personally think this is a perfectly legitimate basis from which to enter the discipline. As evidenced by your other statements, I think you have the right attitude toward your influences.

This is a good foundation. Input and feedback are valuable, but take the crowd-sourced comments with a grain of salt and weigh them all through your own eyes and experience.

Craig Eliason

Yes, overshoots should be adjusted so that all seems equal at intended size which may seem slightly off at other sizes. Use an excellent printout at text sizes, or maybe better view larger printouts from a greater distance (or through a reducing glass) to evaluate. 

On the /f/ I’m not opposed to a larger ascender tout court (so long as you’re prepared to handle the fitting issues it introduces). I just think yours leans too soon. Compare Bembo’s which stays rigid and vertical (especially on the right side) a bit above the crossbar before thinning and turning. 

I don’t see the /a/ as too narrow. 

Kent Lew

FWIW, I’m not convinced that the overshoot is too great. I will say that the prominence of z in the Polish language will give a very different impression of the overshoot than English settings. The constant establishment of that flat x-height plateau in a Polish setting makes the normally subtle overshoot correction more noticeable.

But reducing the overshoot in your design, I fear, might compromise the balance of many of the interior counters against that z.

I wonder if what Craig may be sensing in that szczebrzezynianin is as much a function of the very open c in that sequence as it is the actual overshoot amount.

Craig Eliason

Kent Lew said:

I will say that the prominence of z in the Polish language will give a very different impression of the overshoot than English settings. The constant establishment of that flat x-height plateau in a Polish setting makes the normally subtle overshoot correction more noticeable.

Interesting, @Kent Lew. I am seeing overshoot adjustment as accounting for optical illusions with regard to horizontal alignment. In which case, at least at a given size, if the judgment is made ideally the overshoot is unnoticeable (and character frequency irrelevant). 

Sounds like you see overshoot adjustment as a compromise between accounting for optical illusions with regard to alignment, and something to do with the balance of interior counters. Can you say more about what that second factor is? Would you want an /o/ to look slightly but noticeably taller than a /z/ at target size to achieve something else? 

Nick Shinn

Scarbo? I would have expected something a little more dramatic!
But perhaps I am overly influenced by Ravel’s take on the subject, I haven’t read the original work.

Andreas Stötzner

Jan,
I honour your humble attitude as a ‘beginner’. You’re on your way to leave this stage behind you, maybe more quickly than you assume. Your reflections about your design approach and reference designs you define as (sort of) models – are totally credible to me. With execution/glyph drawing you have done really well already. As common as it may seem at first glance, I find your project the more noteworthy the more I think about it.
1st, the idea of blending features of the Bembonian and Caslonian traditions is interesting in itself and, as far as I know, this concept is everything but over-explorated. 2nd, your key motivation for doing this as “a typeface just for my own use” interests me, because I have experienced myself that this can be a very strong motivation and it causes some sort of subconcious guidance for detail decisions induced rather by one’s heart and womb rather than from the brain.

Jan Pietkiewicz

I would like to sincerely thank you for your comments, which double as a very warm welcome to the forums. Even though I knew I could count on this community for sage technical advice, I was half-prepared to be thoroughly castigated for some rookie mistakes or my generally misguided approach. I certainly wasn't expecting this level of encouragement, for which I remain most grateful. I'm very glad, if surprised, to hear that my design is already looking half-decent (although let us be clear that I didn't raise the bar for myself particularly high).

@joeclark – As an occasional reader of your writings since the mid-noughties, I can only feel honored that my very own thread inspired this tidbit of your unique brand of snark. It is nice for a change to be appreciated for my lack of qualifications, a virtue I managed to achieve without much of a struggle.

Christian Thalmann said:

Nice job! I find the figures particularly clean and striking — hard to believe you’re not happy with them! Only the /3 looks a bit unbalanced to me.

I'm reasonably satisfied with the numerals on their own, I just have some doubts about their coherence with the letters. Maybe I'm also slightly apprehensive because working on them made me realize that I always took the digits for granted and felt lost when it came to deciding on their shapes. Glad to hear you liked them despite that! The /3 is definitely in need of urgent surgery and I feel there is something wonky about the way the bottom segment of the /5 begins.

Christian Thalmann said:

I’m a fan of generous flags on /f […]. I recommend drawing a «retracted» version of it […]. 

Btw, the /f_h-style ligatures strike me as a bit compressed.

A retracted alternative /f would undeniably help to prevent any bruising that this glyph is cruising for with its garish and provocative flag. Thanks for this suggestion! The ligatures are without doubt the most half-hearted component of this whole enterprise and I will need to reapproach them with more care.

Christian Thalmann said:
Have you considered making the typeface available open-source?

I would be certainly open to this possibility, but keep in mind that prior to starting this thread it didn't even occur to me that this typeface might be of any use to anyone but myself. I believe it would be extremely irresponsible to release it into the wild before it's complete (in the most basic sense), perfected and thoroughly battle-tested (even if the open-source model obviously allows for further development and polishing). I'm not ready to count the chickens before there aren't even any eggs to speak of; I do accept as an enormous compliment, however, the very suggestion that my design might qualify for a public release.

Kent Lew said:

This is a good foundation. Input and feedback are valuable, but take the crowd-sourced comments with a grain of salt and weigh them all through your own eyes and experience.

Thank you for this important reminder. It is very easy for a person as prone to second-guessing and self-doubt as I am to fall into the trap of unreflectively following the advice of the others, especially if they are incomparably more experienced professionals. That being said, this thread has already given me enough confidence boost to trust my own judgment for years to come.

Craig Eliason said:

On the /f/ I’m not opposed to a larger ascender tout court (so long as you’re prepared to handle the fitting issues it introduces). I just think yours leans too soon. Compare Bembo’s which stays rigid and vertical (especially on the right side) a bit above the crossbar before thinning and turning.

Thanks for elaborating on the sensitive matter of /f's flaccidity. I will make the relevant comparisons and see if I should pump some rigidness into it.

@Kent Lew, @Craig Eliason – I found your discussion of overshoots extremely interesting, but way above my level of expertise (up to this point my understanding of overshoots pretty much boiled down to the point that "some should be present, probably"). Before making any adjustments I would definitely like to read more about the theory at work here; I'm also very curious now whether any Polish sources corroborate Kent's intriguing observation about the influence of letter frequency on the perception of baseline and the x-height (a point so interesting that I would be perfectly happy to accept it as the only good thing to come out of my starting this thread).

Nick Shinn said:

Scarbo? I would have expected something a little more dramatic!
But perhaps I am overly influenced by Ravel’s take on the subject, I haven’t read the original work.

I'm afraid that the prose poems in question would also call for a design more characterful and fanciful than mine. Since the choice of the name served mainly my own amusement, I settled for "Scarbo", because it was the first thing that came to my mind when I was looking for a two-syllable, vaguely-Italian-sounding name to continue the sequence of Griffo, Bembo, and, say, Arno (the second thing, Pozzo, somehow seemed even less appropriate). The added advantage is that since Scarbo is a dwarf, the name helps to drive home the point about standing on the shoulders of type giants. If it is any consolation, the writings for which Scarbo was originally developed (yet to be appreciated by the readers at Polish publishing houses) are much more in tune with both Ravel's and Bertrand's works than the typeface itself.

@Andreas Stötzner – as I remain wholly skeptical about any noteworthiness of my first attempt at type design, I can only thank you for your very kind words.

Having received this much sympathetic and helpful support, I now feel obliged to continue my work on this face to my best ability. There are now plenty of things for me to think about and experiment with. Should I arrive at some next milestone, I will be most grateful if I could seek your further opinion.

Kent Lew

@Craig Eliason — Hmm. You’ve made me stop and think. You’re right to challenge me, as it does seem now that perhaps I’m holding two different things in my mind at once under the single rubric “overshoot.” Let me see if I can articulate.

It seems to me that I have frequently felt confronted by the conundrum, when evaluating lowercase vertical dimensions, that the answers to the two questions: “Do these appear to align at the top?” versus “Do these appear to be the same height?”, which would seem tautological on their face — end up being different.

In the classic illustration of the concept of overshoot, using solid squares, circles, and triangles, the answers tend to be one and the same. But the addition of counters, weight, and contrast may lead them to diverge — or so it seems to me anyway.

For whatever reason, this paradox seems to concentrate in the x-height region. I don’t experience the same dilemma as much with caps or along the baseline. And context seems to affect my perception — rounds in proximity with kxvwy can feel different than in context of ftz. So, I am in fact usually trying to find a balance between those two perceptions (hence my comment about z frequency).

And, as a result, I often have a different overshoot amount at the x-height than at the baseline or cap-height. Maybe on the order of 2–4 units difference.

Does that make any sense? Maybe I am alone in this.

Christian Thalmann

rounds in proximity with kxvwy can feel different than in context of ftz

Which is why I tend to lower the horizontal strokes in /f/t/z in the light weights:

Without this correction, I could well imagine that the prominence of /z in Polish would raise the overall perceived location of the x-height.

Craig Eliason

Kent Lew said:

It seems to me that I have frequently felt confronted by the conundrum, when evaluating lowercase vertical dimensions, that the answers to the two questions: “Do these appear to align at the top?” versus “Do these appear to be the same height?”, which would seem tautological on their face — end up being different.

Thanks for the reply @Kent Lew

I don't think you're alone in seeing this--your analysis reminds me of wrestling with how to adjust the x-height on a bolder weight of a type design, especially a low-modulation design. In that example it's easy for me to see how heartlines and counterspaces make your two questions have different answers. It makes sense to me that there would be a similar though more subtle effect glyph-to-glyph. 

Along those lines, standard practice for handling too-small bolds seems to be raising the midline (not also lowering the baseline), so the asymmetrical solution you're talking about has precedence there too. 

(Still think Jan's flat letters look too small though! )

Kent Lew

your analysis reminds me of wrestling with how to adjust the x-height on a bolder weight

Exactly. Similar phenomenon — apparent size vs apparent alignment.

Still think Jan's flat letters look too small though!

Fair enough. As a perceptual issue, opinions will differ. ;-)

[Hope to see you in Portland later this week?]

Craig Eliason

Sadly won't be there

Ori Ben-Dor

@Christian Thalmann

I remember reading somewhere that Adrian Frutiger once noted that, practical concerns aside, vertical and diagonal stems also need a little overshooting.

Maybe that's what's really happening here.

Ori Ben-Dor

Well, can't say this experiment supports that theory...

Christian Thalmann

Try removing the verticals in the boxes. I found the effect the most jarring when designing the capital /Xi of my hairline sans. 

Dave Rowland

Really interesting, and it's made me look at the font I'm working on now, and yes, the lighter weight z, Z, 5 and 7 all look too tall. Would you raise the bottom of z off the baseline too?

Christian Thalmann

Dave Rowland said:

Really interesting, and it's made me look at the font I'm working on now, and yes, the lighter weight z, Z, 5 and 7 all look too tall. Would you raise the bottom of z off the baseline too?

I usually don't, although the effect is visible there too. As Kent said above, it draws more undue attention at x-height than at the baseline for some reason. Maybe the baseline is so well established by the consensus of the other letters that the eye applies some internal hinting to «fix» such problems. 

That /Xi is really painful, though. I think I'm going to bring it down a little, like the /z. I hope this won't lead to hinting mistakes at certain sizes.

Robin Mientjes

This is echoed in Adrian Frutiger’s drawings (I think) for Frutiger (I think!) where the E, F, H and I shared a base template (as a lot of Linotype drawings did, because they often share metrics one way or another), but all had slightly different heights. I’m trying to find this reference, but I promise I saw it once. Maybe it was actually a phototype version? Someone’s gotta know.

Andreas Stötzner

Christian Thalmann said:

…

That /Xi is really painful, though.

Xi is always painful.

Craig Eliason

It probably makes more conceptual and hinting sense to set this up as overshooting the verticals/diagonals rather than lowering the flats. 

Craig Eliason

Robin Mientjes said:

This is echoed in Adrian Frutiger’s drawings (I think) for Frutiger (I think!) where the E, F, H and I shared a base template (as a lot of Linotype drawings did, because they often share metrics one way or another), but all had slightly different heights. I’m trying to find this reference, but I promise I saw it once. Maybe it was actually a phototype version? Someone’s gotta know.

Referenced in this thread:
http://typedrawers.com/discussion/2111/too-much-overshoot

Kent Lew

@Craig Eliason — Sorry to hear it. Been awhile. Maybe next year . . .