I agree the changes make for better legibility, but why change the display cut and not the text cut? I'd have thought it would make more sense to do it the other way round.
I can see how my idea might be counterintuitive, inasmuch as you'd think at small sizes you might pare down the inessential details rather than emphasizing them. But my idea is that (like hairlines) the transitive serifs get lost when scaled down unless given a little oomph, whereas at display sizes they needn't be so emphatic to perform their linking function.
This is more a hypothesis than a conviction, I should say.
Since building the letters out from symmetrical counters has become a strong theme of the roman, it occurred to me to try out how that would look in the italic. Here's a rough try at symmetrical counters in the italic bowls (/a/b/o/ of second word):
Those counters probably still need to go clockwise a little more.
This "modernizes" the typeface noticeably.
(For comparison's sake here's where the roman is:
It's certainly true that generally an Italic should be at least as funky as the Roman.
Looking at italic /b/ and letters like /n/
/b/ in 1 has round bottom, /b/ in 2 has spur at lower left
/n/ in 1 is smooth s-shape, /n/ in 2 has a different counter including a corner near the lower right. (/i/ is changed correspondingly).
I probably need the spur on /b/ to anchor it.
As for the /n/, I think this new, stiffer counter fits better with the newly geometricized round counters and with the ideas behind the roman companion. But does that corner become something your eye trips on?
I prefer 1 for the /n and /i, but I like the spurred /b. It pins it to the baseline more. I think the top part of counter in 2nd /n is a better fit with the geometric counters, but the corner is a bit jarring and, IMO, unnecessary.
Thanks Dave, I think you're right.
3 shows a first go at the new counter but with the corner removed.
The new counter approach throws a wrench in my efforts to have an ascending italic /p/. Previous post shows one solution (taken from an earlier version) which I kind of like but it gets too busy esp. at text sizes. Here's a simplified version of same:
but I don't know that that's working well either.
Here (bottom three lines) is what the nonascending version might look like.
That indeed may be the way to go.
I'd go for the conservative approach. To me the /p/ acquires certain stiffness with the so far considered ascenders. The /p/'s you presented back in February a year ago looked reasonable to me. My personal preference would be the baroque, Caslon-style /p/.
To me the /p/ acquires certain stiffness with the so far considered ascenders
I guess you won't much like my carpet-tack variant! :-)
The Caslon-style /p/ works okay but I wonder if it's too quaint, and it's become an outlier with my turning the counters of the other bowled letters to ellipses.
How about this barbed finial? I think this is the best of the ascending non-Caslon ones so far.
I like it!
New pdf specimen. Changes since the last include:
- settled on italic /p/
- added crossbar to italic /Z/s
- changed counters of italic shouldered lowercase
- made fives more consistent
- widened /V/s
- adjusted some italic serifs
Yet to be done:
- draw small caps in all masters
- respace lining figs
- expand character set
Relaxed the curl at the bottom of oldstyle 3 and 5. Rejiggered weights of 4's strokes. Simplified tail of Q.display (took out the little loop) and tweaked other Q tails too.
Shortened crossbar of the italic Zs. Still might be a little busy, but I like having something there in the middle to add heft to that long hairline. Adjusted last serifs of VWY: still planning on having the "windblown" versions as normal with lowercase, and the more rigid ones on the right triggered by CASE.
Gave symmetrical counter treatment to sixes and nines.
I don't know if that "3" has been like that for a while, but: wow. Even though -to me- it's a novel shape, it still manages to say "I've been doing this for centuries."
Thanks! The top row shows the way it has been for a while until these last changes, which mainly were relaxing the curl of the bottom.
Middle row is current state of old style numbers which will be the default.
Bottom row is the lining figures. I like that the structure and weight of the figures (and not just the height/vertical alignment) differs between OSF and LF.
Then for the LF I'd consider a "7" with a double-serif base. And if you're feeling brave curve the diagonal of the "4" so it's not so lonely. Patria:
BTW I'd still have a spur on the OSF "2".
Yeah, that was on my to-do list.
Interesting idea on the seven, but I can't get it to be persuasive. The way my seven's stroke tapers and bends, it looks weird to me to grow out of a stem serif.
It looks weird in Patria too. But it feels right. :-)
On the other hand my "7" (along with "2" and "5") has "rationalist" stroke contrast, so it might not work the same for traditional stroke contrast numerals.
But try "1874" too.
I think I like it but not as much as the lachrymal original. My eye catches on that serif a bit.
Hmm, what about straighter, with a serif on the other side?
I really like the first one-way-serif 7, but the original one fits very nicely too. I guess that means it's just a matter of taste rather than necessity.
The spur on top of 5 could be more definite, I think, looks a bit shy.
I prefer that 7 too, more so than the lachrymal one. I think the serif holds it to the baseline much better. The teardrop version seems a little flighty.
While I'm here... Italic /y feels too light in all weights, especially next to the /p. For what it's worth, I much prefer the non-ascending /p over all versions with ascender, but if you must have an ascender, I think the current treatment is best. Bottom left of /i and corresponding part of /m and /n still a bit kinky. I still can't make myself like /a and /c in the Roman, but maybe that's just me. Display Black /w is gorgeous!
Of those I think I like the leftward-serif one best. But please tell me what's wrong with this:
I guess that means it's just a matter of taste rather than necessity.
Which is exactly the best opportunity to deploy some culturally-progressive ideology! :-)
Yeah, once I straightened out the stroke, a bilateral serif becomes workable (E, below, is my latest version). But I think I might like this new one, which is the rightward serif and straightened stroke (D).
Bendy, I agree on the five, don't know why I was so wimpy with that.
Dave, I'll look at all those, thanks for the feedback!
I like the last three (and the last one the most).
And now try bending the diagonal of the "4" to match. :-)
BTW that blob in the middle of the "3" looks like an aneurism. Just make it straight - it would keep the "4" company too.
Yes, I was just thinking that bubbly 3 is not keeping up with the dequaintification this is undergoing. (That kind of little loop is what I just cut from the tail of Q.display). Might even have a go at a straight-topped version, since this seven has gotten more rigid.
top: oldstyle figures (terminals of 2 and 5 are revised)
middle: the old lining figures (3's loop chopped, 5's terminal revised)
bottom: lining figures with new flat-topped 3, leading new 7 candidate, and same 5 as middle row
I'm not sure if the 3 is there yet but I think this might be the right direction to go.
All good, except the flat-top "3" is too mannered (as they almost always are, at least to me). And one nice thing about a round-top "3" is that it keeps the "2" company.
BTW you might want to shear the bottom of the OS "4" (maybe the same angle as the middle of the "3").
BTW you might want to shear the bottom of the OS "4"
It already is, but clearly not at enough of an angle!
BTW the spine/base join in the OS "2" is probably too dark.
And why not use the ogee form like in the LF?
Isn't the vertical in the "5" too thin?
And why not use the ogee form like in the LF?
There's less vertical space available for the curve to have room to change directions.
Dave, if I lessen the tilted effect on /a/ and /c/ could they win you back over? This is a revised version in all three masters:
Glad you got me looking at italic /y/ again. This was a letter that I loved the look of, but had to admit it didn't play well with others. So I've straightened out the SW/NE thin stroke, which allowed me to bring the intersection leftward, which ameliorated the spacing problems the older version was posing.
I can live with that new /c, but /a just seems weird. It looks taller at the start of a word than it does in the middle. Maybe shifting that node up a bit on the transition between vertical and curve on the top right would help? Or nudging the topmost node to the right slightly?
y looks better now in terms of weight. the top left of the black one is not as nice a curve as it was before though.
5 May 2013 — 11:41am Old Style Figures
The lower counter of old style 2 is rather small. I would draw 2 a little bit higher without enlarging the upper counter, make the join thinner on the lower side and maybe flip the serif on the end of the horizontal stroke downwards.
It looks as if old style 4 is positioned too low. In many typefaces this is solved by positioning 4 a little bit higher, as a result, the horizontal stroke of 4 does not align with the baseline anymore, but I do not think that causes any trouble.
Thanks, those are good ideas that I will try out. It's funny - I've known for quite a while that that baseline stroke of four will probably need to be raised but it's difficult to force my compulsive self to move it off that line!
In the meantime a different take on the top of lining two and three. I think I like the three but the two isn't as satisfying. And if I adopted this change I guess maybe the teardrops of six and nine have to be revisited.
a shows the oldstyle two from before
b shows my best effort to even out the counters keeping it at "zero-height," which definitely looks better than a.
c shows a version that is allowed to be taller, but not full "six-height." Was this what you had in mind Albert Jan? I wouldn't have thought of this, but I can see how, though vertical alignment doesn't match anything else, the counterspace feels more comfortable with the other oldstyle figures. I might let the oldstyle four go up to that level too (Actually I'd already done something similar with the italic oldstyle four.).
(I tried a full "six-height" version that looked terrible. Also couldn't get a serif on the bottom side of two's terminal to work in this style.)
Note that a properly ascending "2":
- Does have precedents, most notably in some of Gill's fonts.
- I think works better, because: it helps strings of numerals sit less low*; and is more distinct from "z" and "s".
* Which is why I prefer the French-style ascending "3"/"5" scheme.
Ah thanks, Perpetua is an instructive model, with its "three-quarter" height OS 2. Interesting that the italic 2 doesn't ascend.
LF 2 and 3: I prefer the ball terminal versions, but maybe they'd be better less curly, like the top of 6.
I like the ascending os 2, but maybe somewhere in between b and c would be a good compromise - as it stands, certain numbers might look a bit wavy on top (think 12820) which *could* look like a hinting error. On the other hand, a shorter 2 could make that effect more pronounced. Need to see it in running text to decide.
Oldstyle figures in running text
This shows that two shortened by 1%. It also shows a four that has not only been raised, but also the convexity of its diagonal has been flipped to open the counter.
Maybe my oldstyle extenders need to be reduced a bit overall (lowering top of 68 and raising bottom of 3579)? 5, 6, 7, and 9, at least, probably don't need to be so tall.
From the point of view of an absolute newcomer, I definitely prefer the "closed" OS 4 and the "non-ascending" OS 2. To my non-expert eye, the current OS 2 and 4 provide too much volatility within OS figures and also with respect to running text. Additionally, I think the lower counter of the OS 5 should be proportionally larger that the upper part; in its current form, the OS 5 seems a bit "loose".
And, of course, congratulations for your titanic work on this face.
Unless you do lower the "6" and "8", I think the "2" is too short. Make it mean it.
Love the new "4".
BTW have you considered a zero that doesn't look like an "oh"? But certainly not the stupid "perfect circle" one. My favorite is the one in Whittingham: http://luc.devroye.org/Berthold-GGLange+DieterHofrichter-Whittingham-200...
At the risk of sounding like a spanner, or a broken record, I don't think this design needs to push boundaries by adding quirks gratuitously. I'm not convinced about the appropriateness of the new 4, and the extra levels of alignment that the 2 and 4 introduce make it feel discordant to me.
How about this:
2 is back to the same height as 0/1.
Descending figures descend less deeply.
Ascending figures ascend less high, except for 8.
Also reined in the new 4 a bit.
Yes, I've tried a bunch of unconventional zeroes (I think we may have even discussed this a few months/pages back). I have a monoline zero alternate currently, as well as an I-like one alternate,
but frankly those might wind up on the cutting room floor I think.
I'm no fan of gimmicks either. It's just that sometimes our conservatism is not justified, and there are benefits to be had in features that might seem unconventional. I remember a discussion on Typophile many years ago where I asked Jonathan Hoefler whether he would consider implementing the French-style ascending "3" and "5"; his reply was something to the effect of "Why would I bother?" That sort of conservatism is very alien to me personally.
Bowing the diagonal of the "4" gives it more apparent size (which is one reason I've recommended Craig do that to the LF "4" as well); making the "2" ascend has at least three functional benefits; and things not lining up is not so bad in a text font.