"TT" and "tt" ligatures

[joaopaulo1511]

It would be nice to have "TT" and "tt" ligatures, joining both letters, as the "Th" ligature does.

Would it be a case for discretionary, historical or standard ligature?

[skalyan91]

I second this request! And I'd also like to see "Ti", "Tj", "Tl", "Tb", and "Tk" (among many others). (Actually, I was going to add these myself once I'm done with the IPA characters—to be reviewed by Georg, of course.) You can tell that I'm a ligature addict…

I've heard that the "tt" ligature is mainly a German thing, and is obligatory in German but discretionary in other languages. See Linux Libertine (which also has other German-specific ligatures such as "ch" and "ck").

I'm pretty sure that "TT" would be discretionary (in any language). --  Siva Kalyan (♂) Sent with Airmail

On 19 September 2013 at 05:14:17, Siva Kalyan (sivakalyan.princeton@gmail.com) wrote:

It would be nice to have "TT" and "tt" ligatures, joining both letters, as the "Th" ligature does.

Would it be a case for discretionary, historical or standard ligature?

— Reply to this email directly or view it on GitHub.

[georgd]

First of all, I’d be very careful with ligatures besides the well established ones, especially in the roman font. The cursive nature of the Italics allows for a bit more playing, especially in combination with swashes, but it shouldn’t hinder readability. The primary purpose of these fonts is text not decoration.

The following is about the Roman font:

• I could imagine T_b and T_k to work fine, like T_h (and could live in the same feature). T_l however is rather problematic in my opinion, I doubt it’d be harmful to legibility.
• A TT-Ligature might cause problems (Pi with a long top-bar?) but might be ok. in a decorative context.
• I can not imagine a T_i or T_j. Simply omitting the tittle is not acceptable for me, but a drop on the T’s right arm would look rather funny IMO.
• EBG12-Re has a t_t but it’s not enabled in any lookup (it was, but I didn’t like it) so this is simple, we just have to agree on a feature. The thing about obligatory ligatures in German is quite simple: they don’t exist, at least not in a way that concerns us. In blackletter typesetting (predominant in German speaking countries until 1942) more ligature pairings were common (like t_t) and four of them were obligatory in a way that they wouldn’t be broken up for letterspacing. They are namely ch, ck, ſt and tz. Of these four, t_z is only rarely found in roman fonts, usually with a ʒ-shaped z. c_h and c_k are (if at all) a bit tighter kerned, but together with ſ_t they have their own cases in the standardized German lettercase for roman fonts (http://de.wikipedia.org/w/index.php?title=Datei:Setzkasten_Antiqua-Schriften.jpg&filetimestamp=20061025102031&). Thus they were produced even if they didn’t look different from the simple pairs.

Besides the ones accessible, there are currently t_t, t_z, c_h, c_k and longs_s (in my current working directory glyph IDs 1957 etc.).

About the features: none of these ligatures should be default. I want to reserve the hlig feature for the group of s_t, c_t etc. which are historically attested and very common in French typography. dlig currently contains T_h, there I could imagine putting T_b and T_k (and perhaps t_t). For all other playing with ligatures I’d rather use some styleset.

[joaopaulo1511]

Thank you for both responses!

I am not a typography expert, just a typography lover, and will share my thoughts:

I agree with Georgd on the following points:

• T_b and T_k, like T_h, are good, and T_l not so good to readability;
• T_i or T_j would be funny, maybe if the right T's arm would be a little bit extendend besides just dropped on top of "i", but is still a little bit weird. There are some samples at http://typophile.com/files/Ti%20ligature_0.jpg (bad) and http://typophile.com/files/Ti2_5116.png (maybe good, if set as a discretionary ligature?).

But I can't see how a T_T ligature would be bad. A Pi symbol isn't used inside a word unless the language is greek, and I think OpenType allows to set ligatures that work for a language and not work for a another. Also, the T_T ligature could have the arms ending on both T's visible in the middle of the T_T ligature (I tried condensing the letter spacing on 2 points to achieve a similar effect to what I was thinking).

Could you please consider restoring the t_t as a dlig? And T_T as a dlig (preferred) or as a styleset? I know you are the guy who is bringing the Garamond font to open source, and so you know better if the font should have or not the ligatures as liga, dlig or hlig (historical?)

[khaledhosny]

As someone who just hates Th ligatures (a quite new invention of a well known Adobe designers), I think adding new ligatures should not be taken lightly. Latin ligature are used to fix specific typographic problems, but I feel to say what the problems any of those combinations has that are to be fixed with ligatures. Purely decorative and playful ligatures are for display type (like the Times one linked above, which seem like a tightly kerned T and dotless i rather than a ligature, BTW), for text fonts like EB Garamond they are likely to annoy the readers with their unusual form. Good typography should not get in the way of the reader.

[skalyan91]

My impression was that the point of T_h is to save space, and make the black/white distribution more even. Otherwise, Th can look like it has a "hole" in it, especially in a Garamond where the top bar of the T is very wide, and "shelters" a large amount of white space. The same goes for any instance of T followed by a lowercase letter with an ascender (or an accent mark). I would make the same argument for r_a, c_a etc., but I don't want to be too radical :)

In my view, this is the main useful purpose to be served by ligatures: to make the colour of the text more even. (I thus agree with you that purely playful or decorative ligatures are best avoided for text typefaces; but I am arguing that T_h et al. are not "purely decorative and playful", but are highly functional! I am also not convinced that such ligatures hinder readability; personally, I would be more distracted by a hole in the text than by a ligature which, as it were, "melts into the background".)

I remain convinced that T_i, T_j and T_l could be made in an aesthetically pleasing and readable way, but I will need to try making these for myself. What I have in mind for T_i is something like the "Title" linked above, but with the dot a little higher (so that it is vertically centered with respect to the top bar of the T), and a little to the right (so that it is in the same horizontal position with respect to the stem of the i as in the standalone i). Likewise for T_j.

I agree that it is a little harder to make T_l readable; there is the danger of it looking like a mutant Pi. However, similarly to what João suggests for T_T, the serif on the right arm of the T could be retained, and the arm could simply be made shorter (but still with a visible connection to the top serif of the l).

(As an aside, I think T_l should be a higher priority than T_b or T_k; think of all the Aztec names beginning in Tl-! The only uses I can think of for T_b and T_k are Tbilisi and Tkinter.)

On 20 September 2013 01:34, Khaled Hosny notifications@github.com wrote:

As someone who just hates Th ligatures (a quite new invention of a well known Adobe designers), I think adding new ligatures should not be taken lightly. Latin ligature are used to fix specific typographic problems, but I feel to say what the problems any of those combinations has that are to be fixed with ligatures. Purely decorative and playful ligatures are for display type (like the Times one linked above, which seem like a tightly kerned T and dotless i rather than a ligature, BTW), for text fonts like EB Garamond they are likely to annoy the readers with their unusual form. Good typography should not get in the way of the reader.

— Reply to this email directly or view it on GitHubhttps://github.com/georgd/EB-Garamond/issues/54#issuecomment-24769095 .

[skalyan91]

(Tbilisi being the capital of Georgia, and Tkinter being a Python package.)

On 20 September 2013 10:00, Siva Kalyan sivakalyan.princeton@gmail.comwrote:

My impression was that the point of T_h is to save space, and make the black/white distribution more even. Otherwise, Th can look like it has a "hole" in it, especially in a Garamond where the top bar of the T is very wide, and "shelters" a large amount of white space. The same goes for any instance of T followed by a lowercase letter with an ascender (or an accent mark). I would make the same argument for r_a, c_a etc., but I don't want to be too radical :)

In my view, this is the main useful purpose to be served by ligatures: to make the colour of the text more even. (I thus agree with you that purely playful or decorative ligatures are best avoided for text typefaces; but I am arguing that T_h et al. are not "purely decorative and playful", but are highly functional! I am also not convinced that such ligatures hinder readability; personally, I would be more distracted by a hole in the text than by a ligature which, as it were, "melts into the background".)

I remain convinced that T_i, T_j and T_l could be made in an aesthetically pleasing and readable way, but I will need to try making these for myself. What I have in mind for T_i is something like the "Title" linked above, but with the dot a little higher (so that it is vertically centered with respect to the top bar of the T), and a little to the right (so that it is in the same horizontal position with respect to the stem of the i as in the standalone i). Likewise for T_j.

I agree that it is a little harder to make T_l readable; there is the danger of it looking like a mutant Pi. However, similarly to what João suggests for T_T, the serif on the right arm of the T could be retained, and the arm could simply be made shorter (but still with a visible connection to the top serif of the l).

(As an aside, I think T_l should be a higher priority than T_b or T_k; think of all the Aztec names beginning in Tl-! The only uses I can think of for T_b and T_k are Tbilisi and Tkinter.)

On 20 September 2013 01:34, Khaled Hosny notifications@github.com wrote:

As someone who just hates Th ligatures (a quite new invention of a well known Adobe designers), I think adding new ligatures should not be taken lightly. Latin ligature are used to fix specific typographic problems, but I feel to say what the problems any of those combinations has that are to be fixed with ligatures. Purely decorative and playful ligatures are for display type (like the Times one linked above, which seem like a tightly kerned T and dotless i rather than a ligature, BTW), for text fonts like EB Garamond they are likely to annoy the readers with their unusual form. Good typography should not get in the way of the reader.

— Reply to this email directly or view it on GitHubhttps://github.com/georgd/EB-Garamond/issues/54#issuecomment-24769095 .

[georgd]

My problem with T_l is, that the "l" would loose one third of its already few distinctive features. This is already a problem in f_l and longs_l (but being used to it it doesn’t matter so much) but in a combination of UC_LC this matters much more IMO. Also, as you said it’s mainly found in Aztec words which for non-Aztecs (or at least non-Mexican) may already be difficult to decipher without placing a glyph there that is not so easy to identify doesn’t help. I don’t know what Aztecs say about it.

@joaopaulo1511 your first example is ok as kerned Tıme and I don’t like the second one.

Such playing with T_i and T_j (and many more) where you place the tittle somewhere else, may be an interesting add-on for bigger design-sizes for display, but not for 12pt.

I think, it’s difficult to make a T_T-ligature at this size that doesn’t look like a keming-accident. However, I could imagine such for a "titling"-set of letters.