Turkey: a new symbol for the lira

[evertype]

OK, everybody, it's time to get your design caps on.

1) The character will be encoded. No point in arguing about it. Turkey submitted a proposal to encode it, as did I. Mine is here.

2) So far for the ballot, the glyph for the character will be the "official" one as seen at the top of the page here. Now as you know, in the code charts the more generic shapes (in this case based more or less on Times) are generally used. Now despite the fact that I had proposed to use one of Craig Eliason's forms in my proposal, the UTC felt that it was too early to know how this would be designed and implemented in OS fonts or other fonts. So I said I would come to this forum and ask the wider community for designs so that se might see what sort of consensus there would be. I see John and Bendy have come up with some. It looks a bit like people are turning a P and lopping off part of its bowl. In a serif font, the real questions are, what to do about the top serif (I followed Craig's), what to do about the horizontals, and what sort of finial should go on the anchor's fluke.

Start designing!

[Riccardo Sartori]

I think the design of uppercase Psi already presents much of the same issues.
(and there could be worse historical references)

[kentlew]

Regarding the top serif, it seems to me the primary relation of the sign is to the L for “Lira”, so the top serif treatment might as well follow that.

I don’t think the starting point for the lower curve should be so much a turned P. Since the official spec derives from the quarter circle, I think you need to look at the “circular” elements in a face, primarily the O — which naturally leads to the D, which is already a half rounded form with a straight left stem. So, I think the lower curve could probably derive from general pattern of the D.

(Obviously, there is usually a close relation between the P and D. So maybe I’m making a trivial distinction, since neither is usable outright.)

Now, what happens with the terminal of the truncation — the anchor fluke — that’s where the interesting design happens. And I expect this element will be the most variable, echoing the more salient aspects of any particular design’s “style.” Will it admit of a ball terminal, for instance? There’s an intriguing question.

I expect the treatment of horizontals will be driven by the proportion of the other elements and the targeted size and display environment.

[ahyangyi]

@Riccard0

[Riccardo Sartori]

Yes, thank you, that's exactly what I had in mind.
I think it could be the best solution for the terminal of the upward curve in a serif style.
Especially if one manages to balance the relationship between the slanted crossbars and the "exit stroke" of the curve.

[eliason]

How about this "rocker bottom" alternative?

[Riccardo Sartori]

Fancy! ;-)

[hrant]

The following didn't turn out as coherent as I could manage, but I hope it will still serve to temper our -commendable- enthusiasm and provide food for thought.

Michael, nice to see you spearheading this. But I'm actually surprised that Unicode is proposing to pay attention to what we do "on the ground". Is this usual? Is it that they need to figure out what the actual Times-flavored rendering will be in their codepage renderings? That makes sense (but not much beyond that does).

And I have to take issue with the first half of Kent's post. Yes, formal harmony in a set of glyphs is important, but I personally always strive for a character to be itself and basing the TL too closely to existing structures risks shortchanging it (pardon the pun). Also, most currency marks are derived from alphabetic structures, but all currency marks end up abutted to numerals... So maybe we need to think in terms of [diverging from?] the numerals, which for example traditionally have thick horizontals.

Also, as much as I admire Craig's inventive -and polished- work, I think we need to talk about this in the abstract before we resort to renderings. This is one of the biggest responsibilities that Typophilers can have - let's go slow and tread carefully.

hhp

[John Hudson]

I agree with Kent regarding the top serifs: this should refer to the L in the same typeface, in the same way that the treatment of the yen symbol refers to the Y.

I don’t think the starting point for the lower curve should be so much a turned P. Since the official spec derives from the quarter circle, I think you need to look at the “circular” elements in a face, primarily the O — which naturally leads to the D, which is already a half rounded form with a straight left stem. So, I think the lower curve could probably derive from general pattern of the D

I think a turned P -- sic, actually inverted -- would be very wrong. What I did for the Cambria signs was to take the lower right portion of the P bowl and modify it to flow into the serif construction of the bottom of the 'L' stem, i.e. actually making it less round. I think making it more of an arc of a circle is a mistake unless the design of the typeface already includes such forms in some way. Kent suggests referencing the D instead of the P, but in my types the D and the P usually reference each other, and the P is proportionally a better match for the Turkish lira sign.

What I think is important is that the arrangement of stroke modulation in the sign should reflect that of the rest of the typeface, in order to maintain a sense of stability -- surely desirable for a currency --, and that it makes sense to make use of the same construction as used elsewhere in the design in similar situations of transition from horizontal to vertical bowl segments. Because of the bars extending to the right, I also think it is important to keep the terminal of the upturn quite simple and not too high, with a good amount of clearance between it and the bars without the latter being placed too high on the stem. What I did in Cambria is to optically centre the crossing of the two bars on the vertical stem, and to roughly align the top of the upturn with where the lower bar emerges on the left side of the stem. This seems to me to be the sort of handy rule that suggests a general model.

There will, of course, be plenty of room for more flamboyant designs, as there is for any glyph, especially in display types, but what is needed at this point is a kind of 'basic construction model' that can be applied to as wide a range of styles of type as possible and produce harmonious results that fit well with the other glyphs, especially the numerals. This is the approach that various people took to the euro symbol when it was instroduced:notably Linotype, who had the largest library of fonts in the target market that needed updating, and who produced a euro symbol design guide that we found produced suitable results for all but an idiosyncratic handful of the 500 fonts that we were hired to extend. Indeed, so good was Linotype's basic construction model for the euro symbol that Adam was able to semi-automate the production of new euro symbol glyphs from elements of existing glyphs, which I had only to fine tune. To me it simply makes sense that the best way to ensure that a new sign harmonises with an existing design is to reference the construction of like-elements in that design. It also accords with my belief that once one has established a core set of glyphs with characteristic shapes and constructions, the rest of a typeface should mostly design itself, and my job is careful refinement of proportions and spacing, not endlessly wondering what shape something should be.

[eliason]

who produced a euro symbol design guide

Is that publicly available anywhere?

Did you reject a serif terminal at the end of the hook because of overall busyness? Does that stroke truncated at the thick part find any analogues in other glyphs?
(My first thought of an existing model for that terminal would be capital /G/, but maybe a typical ampersand could offer some guidance.)
Ben reversed the contrast to go thick to thin (which I had floated earlier, too): the consequent pointier terminal might seem less of an oddball. The thick horizontal, as noted, might fit fine with figures like 2/5/7. I found the thick/thick join at the bottom hard to deal with though Ben's is pretty persuasive. With higher contrast typefaces will that still be workable?

I think we need to talk about this in the abstract before we resort to renderings.

I dunno, when the conversation is about "what should this look like" I find renderings a good way to converse.

[hrant]

Visible language has multiple layers, with different mentalities required for the proper design of each. And when a character has an "official" rendering it gets even more complex. I worry that if somebody produces a very attractive rendering that works well in one style, people will blindly gravitate to that and try to extend it to every other style, and we'll pay the price in the long term.

What I personally find important right now is the sort of thing John did in his most recent post: try to talk about what the symbol -and its parts- need to be doing. This is parallel to the advantage of performing rough sketches of an idea before actually making things that look final.

Specifically, the biggest question seems to be: assuming "stroke" contrast, should the bottom be thick or thin? Although it would be OK to go against such a decision in a given case, it's a sort of decision that does have a place, and on a level higher than making glyphs.

hhp

[John Hudson]

Yes, Craig, I avoided a serif terminal on the upturn to avoid busyness. It isn't reflected elsewhere in the basic Latin set, but it is in some of the mathematical sorts for the same reason.

I can't say whether the 'reversed constrast' is successful without seeing it in context of the rest of the design, but unless that construction carries across elsewhere in the type, what is the point of such a departure?'

I think your hairline-to-lachrymal upturn can work in an appropriately styled font, as could hairline-to-ball terminal.
_____

Re. the Linotype euro design guidelines, see pages 36 & 37 of my Same Difference lecture slides. Pages 38 & 39, which shows Andreas' suggestions for uppercase eszett construnction are also germane to this discussion.

[Due to a font embedding problem, some of the specimens of student work on the subsequent pages are messed up.]

[evertype]

Michael, nice to see you spearheading this. But I'm actually surprised that Unicode is proposing to pay attention to what we do "on the ground". Is this usual? Is it that they need to figure out what the actual Times-flavored rendering will be in their codepage renderings? That makes sense (but not much beyond that does).

Hrant, what happened was that when the UTC accepted it to put it on the ballot, somebody said we should use the Turkish Bank glyph. I pointed out the hullaballoo which ensued when the ECB said that the euro's glyph should be inviolate. One of the UTC people (reasonably enough) said that it was too early to know how typographers would shake this character out, so they shouldn't pre-empt that discussion by mandating a harmonized glyph (like the one based on Craig's that I used in my proposal). I said, fair enough, I'll go prod the typographers and see what it is that they intend to do.

I would rather see renderings alongside abstraction.

I think we have seen consensus that the top of the vertical bar should have the same serifs as the capital L though.

[Andreas Stötzner]

It seems we’ll have to sink or swim together …

Craig, this looks promising.

[Andreas Stötzner]

Is it to be derived from the L or from the anchor?

I hink this question is rather important. An L is not an anchor and vice versa. I don’t like to vision a bastard of both … :-)
Perhaps the official “anchor” interpretation helps to make the newbie sufficiently different looking from the Lira sign?

Here’s my first take.
.

[kentlew]

I think a turned P -- sic, actually inverted

Yes, I misspoke — inverted, not turned — thanks.

But I see what you’re saying about the bottom portion of the P instead. Which makes the distinction between D/P more minor.

That said, as I’ve played now with this glyph, I’m moderating my view about this arc. As much as I admire John’s specific solution, in my own explorations along these lines, I keep feeling too close a kinship between the result and the D and less affinity for an underlying L.

This bothers me in the same way that it bothers me that so many Euro designs seem to look more like a C than an uncial E.

John’s approach seems to ameliorate this by flattening the arc and curling the end slightly — yielding something that reminds me more of a gondola prow, perhaps, than an anchor.

But I wonder if there isn’t some way to retain more of the intention of the quarter circle anchor (as misguided as the specific geometry itself might be) while still referencing an underlying L. To that end, I’m thinking that some kind of simple terminal treatment echoing the leg of the L, where possible and appropriate, might help stave off a truncated D feeling.

Just musing out loud.

[evertype]

I suppose just for the sake of the elephant in the room:

[hrant]

Michael, don't go there! :-)

I'm not opposed to a full double serif up top. But: why? For one thing, the need to relieve crowding that John mentioned should possibly be extended to the top = no serifs. Also, numerals don't have head-serifs.

The "metaglyph" we've been given clearly deviates dramatically from an /L (something I personally value). We shouldn't ignore that intent; it's like the Turkish government and people are our clients.

I can't say whether the 'reversed constrast' is successful without seeing it in context of the rest of the design, but unless that construction carries across elsewhere in the type, what is the point of such a departure?

- To make it less a letter in general, and less an /L in particular.
- The numerals have reversed contrast (which however could actually just as well lead to the opposite decision, although the TL seems difficult to confuse with any of the numerals - except maybe a flipped /7).

Is it to be derived from the L or from the anchor?

It is to be derived from what it needs to do. And to me the best /@ is one that doesn't look like an a-with-something. So part of it is what it shouldn't do.

hhp

[eliason]

I think we have seen consensus that the top of the vertical bar should have the same serifs as the capital L though.

For argument's sake here's some dissent on that point:

1) The glyph retains sufficient "L-ness" regardless of the serif treatment there. What that bilateral serif may do most is not lend "L-ness" but lend "capital-ness" to the glyph, but there is no pressing reason to associate the lira symbol with the majuscules, is there?

2) A unilateral (to the left) serif treatment at the top makes visual room for the rising crossbars, which the original proposal specifies as longer on the right and higher than many of the serif-topped proposals have them.

3) Settling on the top first may be constraining options as we work out the problems of the other parts of the glyph. The truncated thick stroke at the end of the curve in the John Hudson design perhaps needs some terminal, such as you might find on a /G/, but the busyness of the glyph kept him from using serifs there. Could a different solution on top make serifs possible? If there's no serif on the right side of the stem at the top, the crossbars can be higher and a G-like serif can have more room. At the same time, a unilateral serif means just less ink overall, and less visual clutter with the crossbars right of the stem. And if it is an angled serif on the left, being closer to parallel with the crossbars should also make the glyph less busy. (Though I suppose that angle might vary with the angle [if any] of the ascender serifs in the typeface?)

[Riccardo Sartori]

there is no pressing reason to associate the lira symbol with the majuscules, is there?

No, except that most major currency symbols have this sort of "capital-ness" (leaving the "lowercase-ness" to sub-units, like ¢). And it seems clear that one of the goals of Turkish government was to establish the Lira as a "world class" currency.

[John Hudson]

I think the whole 'anchor' notion is a red herring. In the press release introducing the symbol, it has all the marks of a post facto interpretation of a design that is clearly based around an uppercase L, a lowercase t -- more on that below --, and the conventions of other major currency symbols. And this is what it is important to remember: it is a currency symbol among currency symbols, not a pictogram of an anchor. The whole point of the exercise is for the Turks to have a symbol for their currency that takes its place among the other internationally recognised currency symbols, so we'd be doing them a disservice if we adopted an approach that makes the sign visually out of synch with other currency symbols and risks exoticisation. As far as I'm concerned, the anchor notion is something to be completely ignored, along with any other external shape correspondences; hence I am neither seeking to make the upturn look like a gondola prow nor fussed if it turns out to suggest that, any more than I care if my uppercase A reminds someone of a Swiss ski chalet. I'm designing a sign within a writing system according to the construction of the other signs in that system and the particular typeface design. As Gill might say at this juncture: a currency symbol is a thing, not a picture of a thing.

Hrant:

I'm not opposed to a full double serif up top. But: why? For one thing, the need to relieve crowding that John mentioned should possibly be extended to the top = no serifs. Also, numerals don't have head-serifs.

I didn't find any problem with crowding or busyness at the top of the symbol, but then I also didn't place the bars as high as in the official logo sign. Why not? Because that's not the convention of most currency symbols, and I didn't want to make the Turkish lira sign unnecessarily distinct in this regard (although I consider the higher bars a valid alternative that should be considered relative to the shape and size of the upturn terminal, as in Craig's lachrymal example).

I favoured a double serif at the top because, with few exceptions, currency symbols are based on uppercase letters, and the other Cambria currency symbols follow this pattern. Once again, I think it makes sense to follow what one has already done in the other currency symbols in a typeface, in e.g. the yen symbol. Now, that said, it should be obvious that if one has deliberately crafted currency symbols that utilise terminals referencing lowercase forms, maybe for use with oldstyle numerals, then it would make sense to apply the same approach to the Turkish lira sign. In other words, the shape of serifs to be used is not part of a basic construction model, but a design-specific determinant of the general approach to currency symbol design.
_____

Most of the discussion so far has focused on the relationship of this symbol to the uppercase L, with some reference to the anchor notion. It may be helpful to consider that the sign also resembles a lowercase t, and this is presumably intentional. This is why I think Craig's 'elephant in the room' is potentially a legitimate variant, although I think it would benefit from longer and slanted bars.

[eliason]

Another option for the lower join would be a little closed loop to the left. Starting point here was thinking about £, but you'll also sometimes see this structure in other letters, e.g. some italic /z/s.

By "informalizing" that intersection, it seems to me the curve of the glyph changes character from an arm/leg (as in /L/) to a tail (as in /Q/). I think there may be promise in that reconceptualization, since the curve has to do things we don't expect arms/legs to do.

[dezcom]

What in the origin gives it flavor akin to Turkish? The Latin L may say Lira but what tickles the national origin? Is the anchor part of this? If so. would it look so much like the typical European anchor?

Craig's upturn feels more Turkish to me as a westerner but I have no idea how it feels to aTurk?

[John Hudson]

Craig, I think that construction is a valid option if that's the sort of thing that happens in the typeface design. Note that the closed loop reversal in £ is usually a factor arising from the slant of the descending stroke -- the more vertical the stem, the more likely a simple stop-reversal or corner -- and is specifically a cursive construction, i.e. in ductus terms a result of the pen not ceasing to move. So I can see this form of construction being particularly appropriate in italic fonts that follow cursive ductus models. And, of course, there is no reason why the roman and italic forms of the symbol should have the same construction, just as one might vary the treatment of terminals and stroke transitions between roman and italic £ $ etc.

[evertype]

John, the "elephant in the room" was mine, not Craig's, and it was a turned Armenian dram sign.