Although I have since mellowed a bit and made the Eszett of Ysabeau slightly narrower, and improved the alternation of thicks and thins (thanks Andreas!). It's probably time for an update to that tweet.
salut christian, thank you for your comment! i have seen your bad/better/good sketch but thought the design depends more on the general typeface style. like, your Ysabeau is excellent for what you aim for (good). bad and better indeed seem misfit Ysabeau.
what you labeled “Orangeville“ is in fact the Dresden form. No need for a further distinction here.
In your lightweight Roman design, the right half of capital O is far too heavy. In the same specimen, the idea of ß being a ligature of ſ-s is a conceptual mistake. A roman upright ß is a long s with a curly something on the right side.
In your sans specimen (Dresden form) the top right oblique stroke shall be more steep and the beginning of the bowl does not want to be so stiff.
Don’t worry, a bit more thorough study of good examples and you’ll soon get the hang of it.
thank you for commenting, Andreas. i re-designed the two "sans" uc ẞ:
…and the antiqua lc ß. i looked up some ß online, to figure out how curly/not-s-ish this might should be (the last shown ß is the previous version):
@Andreas Stötzner what you labeled “Orangeville“ is in fact the Dresden form. No need for a further distinction here.
I beg to differ, Andreas. It seems to me that the distinction between a flat and curved top is extremely significant. However, perhaps a more Teutonic name than “Orangeville” would be appropriate, I just chose that because it’s where I live.
@Andreas Stötzner what you labeled “Orangeville“ is in fact the Dresden form. No need for a further distinction here.
I beg to differ, Andreas. It seems to me that the distinction between a flat and curved top is extremely significant. However, perhaps a more Teutonic name than “Orangeville” would be appropriate, I just chose that because it’s where I live.
I agree with Andreas in that the design labeled «Orangeville» is part of the Dresden parameter space and not an alternative to Dresden.
I disagree with «no need for a further distinction». The Dresden parameter space is huge and includes a lot of butt-ugly designs. It is good to have terms for specific aesthetics within the Dresden space to help guide type designers toward successful designs for their typefaces. People's aesthetic preferences differ, so we can't just call them «good Dresdens».
In fact, both the «bad» and «good» examples in my infographic above are Dresdens, yet I would argue it's a very significant distinction!
Maybe our subclasses would be better accepted if we didn't pick more and more city names to represent them. They're pretty opaque to outsiders, and sort of imply equal importance to Dresden, Frankfurt etc., which is not intended.
How about calling Orangeville a plateau Dresden and Zürich a triumphant Dresden? (As in, proportioned like an triumphal arch...) Zehlendorf would then simply be a triumphant Leipzig.
Donat: Your antiqua solution for capital Eszett is already quite workable in my opinion. I would make the transition from heavy stem to thin roof more gradual, though, like in /U/, and widen the gap at ground level in the Bold version (without widening the glyph as a whole).
I think your sans solution needs a bit more work. I recommend reining in the horizontal motion of the right-hand structure (see «vertical stacking» in my infographic), making the gap in the bottom much more pronounced, and ironing out the kink in the top left shoulder. The Regular strikes me as a bit narrow, while the Bold is significantly too wide. Overall, I feel like there's a confusing mix of different curvature styles in the glyph.
Your antiqua lowercase eszett is probably a bit too excited; calm down the horizontal back-and-forth in the right side and perhaps narrow the counter a bit. I personally don't mind it when eszetts stick out their arse as your second design does, but I know Andreas considers it uncouth.
Why are the lowercase eszetts cap-high rather than ascender-high? The right one has a better bottom terminal.
this lc ß is at cap-height to match my grotesk/sans designs. when i did those, i felt ascender height makes them too prominent, cap-height seemed just fine.
i will have to review all of this, since Andreas points out: "the idea of ß being a ligature of ſ-s is a conceptual mistake. A roman upright ß is a long s with a curly something on the right side".
the prior bottom terminal is from the "s". looking up ß with curly-somethings made me aware of designs that relate to the their respective italic designs. which is a nice differentiation.
I don't see much change from the previous in the serif font. I would reduce the interior serif and align the right-hand structure to be more vertical. The interior terminal of the lowercase terminal is still weird; I would use the terminal style from /s/. The spur on the shoulder also strikes me as a bit fussy; it's not wrong, but perhaps removing it would calm things down a bit.
As for the sans, I believe the designs have gotten better, in particular the more humanist style. The DIN style is still a bit pot-bellied and clumpy. Both would profit from a wider bottom gap and a calmer (more vertical) right-hand structure. For the DIN, you might want to consider raising the top right corner to cap height (à la Nick).
Here's a quick mouse-drawn mockup of the Zürich style applied to your typefaces (bottom version in each case). Apologies for the poor graphical quality.
the serif looks good to me overall, but the transition from stem to arch is still a bit unnatural. Check out the /U/ for reference.
Sans is much better, but most versions are still a bit potbellied (upper Regular seems fine). For the others, tuck the lower curve in to align it visually with the top corner («vertical stacking»), but without narrowing the bottom gap.
Additionally, the Sans Bold is much too wide.
As for the style distinction, at this point I'd just use the top version for both font styles, it's the nicer one. In the bottom version, the arch and the bowl seem to be fighting over the curve philosophy.
These designs may be arranged in a logical sequence, and only one.
I would switch the order sequence of Leipzig and Zehlendorf, on the basis that the angle of the top left descending curve is closer to that of the preceding forms with straight diagonal stroke.
Doing so also suggests a possible intermediate form between Dresden and Zehlendorf, in which the top left descending stroke curves inwards rather than being straight, but that has a sharp corner into the lower bowl.
These designs may be arranged in a logical sequence, and only one.
That's because you cherry-picked a few designs from the multi-dimensional parameter space to make it look one-dimensional, thereby implying some non-existing dependencies. For instance, you make it look like the move of opening the counter and narrowing the right-hand structure requires Leipzig as a basis, whereas it can just as easily be applied to Dresden (yielding Zürich) and even to Frankfurt all the way to the left. Likewise, the flattening of the Dresden's roof could just as easily be applied to Leipzig.
You are both correct, the last two may be switched. In which case, they are topologically identical (were the top right corner of my Leipzig to be sharp, not blunt), so one may be dispensed with. Therefore only three intermediate “steps” are required, in order for a variable font to create all the different kinds of Eszett. Assuming that a Leipzig/Zehlendorf with a sharp top left corner (like Frankfurt) is not included; but if you really had to have that, then Frankfurt would be at both ends of the spectrum, making it a circle.
(The reason I included both Leipzig and Zehlendorf is because they are named instances.)
Here's how I see it. This should properly be a cube, but 2D will have to do for now.
Perhaps a three-way distinction between hard plateaus (left column), soft plateaus (middle column), and domes (third column) would make more sense, but «historically», the latter two are grouped together in Andreas' and Adam's four-way distinction.
(Or maybe the terms should be rectangular/half-domed/domed instead, that's more instantly descriptive.)
This “Eszett circle” comprises six topologically distinct variants. A variable font would require repeating one of these instances at both ends of the axis. It’s not perfect, but it would put a number of variants at the user’s disposal in an organized manner, and provide some potentially useful ’tweens.
This circle is just an obfuscated version of my diagram — the axes are still present, just a bit rotated, and you've imparted an arbitrary walk-through path on it. Keeping the axes clear and visible is much preferable IMHO.
(Leipzig and Zehlendorf are topologically equivalent.)
Necessarily so, since Zehlendorf is a subclass of Leipzig.
Certainly my path choice through the options was discretionary, but not arbitrary. The idea was to see if the variants could be arranged along one variable axis.
I presume this is primarily a thought exercise, and you’re not actually planning to make a variable font with an ESZT axis? Variations technology is useful when intermediate interpolated shapes are useful, which I don’t think would be the case here.
Yes, it’s an experiment, an investigation. Some of the ’tweens might be interesting, you never know. The experiment throws some light on layout app interface design: rather than a collection of specific Stylistic Sets, it might be better to arrange such multiple glyph variants on a slider with pre-sets.
However, there are very few characters with so many variants possible in the same font, as Eszett! It’s as if this new (to Unicode and general usage) character has not yet settled down into a default form.
Two-storey “a” is comparable, in the variety of topologically-different variants.
Comments
…and the antiqua lc ß. i looked up some ß online, to figure out how curly/not-s-ish this might should be (the last shown ß is the previous version):
am i going the right path?
what you labeled “Orangeville“ is in fact the Dresden form. No need for a further distinction here.
I beg to differ, Andreas.
It seems to me that the distinction between a flat and curved top is extremely significant.
However, perhaps a more Teutonic name than “Orangeville” would be appropriate, I just chose that because it’s where I live.
i will have to review all of this, since Andreas points out: "the idea of ß being a ligature of ſ-s is a conceptual mistake. A roman upright ß is a long s with a curly something on the right side".
the prior bottom terminal is from the "s". looking up ß with curly-somethings made me aware of designs that relate to the their respective italic designs. which is a nice differentiation.
Andreas Stötzner
thank you guys for your critical eyes! i tried to implement corrections. the ß still is on cap height for now:
Reviewing the graphic I made illustrating the “spectrum” of Eszett styles, it occurs to me that a typeface may contain more than one.
For instance, just as OpenType enables alternate forms of /a and /g in the same font.
And also, the spectrum could be enabled as an axis in a Variable font.
These designs may be arranged in a logical sequence, and only one.
There would be “steps” in the axis, a situation similar to what occurs in the weight axis of a variable Futura font.
Doing so also suggests a possible intermediate form between Dresden and Zehlendorf, in which the top left descending stroke curves inwards rather than being straight, but that has a sharp corner into the lower bowl.
That's because you cherry-picked a few designs from the multi-dimensional parameter space to make it look one-dimensional, thereby implying some non-existing dependencies. For instance, you make it look like the move of opening the counter and narrowing the right-hand structure requires Leipzig as a basis, whereas it can just as easily be applied to Dresden (yielding Zürich) and even to Frankfurt all the way to the left. Likewise, the flattening of the Dresden's roof could just as easily be applied to Leipzig.
In which case, they are topologically identical (were the top right corner of my Leipzig to be sharp, not blunt), so one may be dispensed with.
Therefore only three intermediate “steps” are required, in order for a variable font to create all the different kinds of Eszett.
Assuming that a Leipzig/Zehlendorf with a sharp top left corner (like Frankfurt) is not included; but if you really had to have that, then Frankfurt would be at both ends of the spectrum, making it a circle.
(The reason I included both Leipzig and Zehlendorf is because they are named instances.)
This “Eszett circle” comprises six topologically distinct variants.
A variable font would require repeating one of these instances at both ends of the axis.
It’s not perfect, but it would put a number of variants at the user’s disposal in an organized manner, and provide some potentially useful ’tweens.
Certainly my path choice through the options was discretionary, but not arbitrary.
The idea was to see if the variants could be arranged along one variable axis.
Some of the ’tweens might be interesting, you never know.
The experiment throws some light on layout app interface design: rather than a collection of specific Stylistic Sets, it might be better to arrange such multiple glyph variants on a slider with pre-sets.
However, there are very few characters with so many variants possible in the same font, as Eszett! It’s as if this new (to Unicode and general usage) character has not yet settled down into a default form.
Two-storey “a” is comparable, in the variety of topologically-different variants.