30

In the default font (Computer modern) the ß looks ugly to me. It really sticks out on a page as if it would come from a different font. Here's an image:

enter image description here

What is the history of this character? Why is there a vertical bar sticking out on the left? Why is the bow on the top right much thinner than any other line in this font.

I'm using

\usepackage[T1]{fontenc}

but even if I disable T1, the ß still looks weird (but slightly differently so).

Concrete questions are: What is the history of this character? Was it part of Knuth's original Computer modern or a later addition? What are font options to get an ß that is more in line with the rest of the font? I've tried lmodern which is a variant of Computer modern. Its ß looks better to me, but it still has the little bar sticking out on the left.

  • 5
    Perhaps the font-designer hates the ß character? ;-) – user31729 Nov 28 '15 at 9:51
  • 3
    I mentioned two things that I find weird right below the image. I could go on and on about how ugly this letter is :) I mean, look at the wiggly line. Why on earth does it change to from super-thin to super-fat with an almost non-differentiable bend???? – Thomas Nov 28 '15 at 9:52
  • 1
    @Thomas you really need to ask the font designer if you want to know why the letter looks how it looks… – clemens Nov 28 '15 at 9:57
  • 8
    The realization of the “scharfes S” depends on whether you interpret it as a ligature between long s and short s or between long s and z; apparently, Knuth decided for the former and he's not alone. The authors of Latin Modern went for the latter interpretation. Several fonts sport the small bar at the left. – egreg Nov 28 '15 at 10:31
  • 2
    The little bar sticking out is not about fashion, it is because, as the answers say, the ß is a ligature of ſ - the long s. It has always been extremely common, at least in serif fonts, for the long s to be drawn with this "little bar" (see Wikipedia) and the letter is famously easy to confuse with f – Au101 Nov 29 '15 at 23:12
46

The glyph makes much more sense visually when seen as a ligature of long s and round s, one of the two traditional forms of the ß (the other, of course, being long s and z). Here's a comparison, using outlines from cm-unicode, version 0.6.3a:

f, ß, s

Here I've used f as a reference for the first part of the ligature, since I couldn't find a long s in cm-unicode. So the "super-thin" line you mention in your comment is thin because it's only acting as a ligature between the two sub-glyphs. You can see that the s part has been compressed horizontally to keep the total width reasonable, but on the vertical axis it matches up very closely.

Addendum: having discovered from other answers that Jörg Knappen was the designer of this ß, I was able to find a few contemporary Usenet postings relating to the design. The most concise explanation seems to be this one, from Knappen himself:

I consciously redesigned the sharp s to exhibit the ligature structure <long s> <short s>. Despite the popolar name eszett I find the arguments in favour of this analysis (as given by Tschiold) more convincing than the ones in favour of <long> <z>.

There's also a thread in German which can be summarized as a vigorous discussion over the relative merits of the ſs and ſʒ forms, in which Knappen firmly defends his choice.

One remaining question is why Knappen made the upper part so narrow, when many (most?) fonts with the ſs variant bring the ligature line out much further to the right -- Linotype Aldus Roman, for example:

Linotype Aldus Roman ß

In this case I don't think there's anything on record, and we have to assume it's a personal stylistic choice. This "compressed" style of ſs certainly isn't unique to Computer Modern. For example, here it is in Antiqua:

Antiqua ß

As to actually replacing this glyph with something you find more appealing: if your TeX installation is sufficiently recent, switching to Latin Modern should be enough. From Martin Schröder's answer and barbara beeton's comments, it seems that the Knappen ß was the default in some older releases of Latin Modern. So if a simple \usepackage{lmodern} doesn't do the trick, you could consider updating your TeX installation, or following Ulrike Fischer's instructions for selecting the Knuth ß from the cm-super fonts.

  • 4
    I've looked at ß in several other fonts and think that the design here is a rather mechanical combination of f and s. The small width of the stroke seems to be derived from the serif of the s rather than from bows otherwise found. The curvature of the stroke between the s and the f also seems forced - you wouldn't draw it like this with a pencil. So, I agree with the opinion of @Thomas: this is rather ugly. – Christian Lindig Nov 28 '15 at 11:28
  • 2
    @ChristianLindig From a sampling of other fonts that use the ſs (rather than ſʒ) form, it seems to me that CM's unusual feature is the compressed top part, leading to the sharp angle between ligature and s. Most designs seem to throw the ligature out to the full width of the lower "s" curve. I haven't seen any other design with a thick stroke for the ſ-to-s ligature (I suspect it would produce a top-heavy appearance). Personally I think the ugliness comes from the squeezed top rather than the thin stroke, but "it's ſs, not ſʒ" is at least a partial answer to "why does it look like that?". – Pont Nov 28 '15 at 12:29
  • 3
    The German Wikipedia (certainly not the best source) says “Die Theorie des Typografen Jan Tschichold, dass das Fraktur-ß auf eine ſs-Ligatur zurückgehe, hat sich seit den 1940ern weit verbreitet, ist aber nicht haltbar.” It seems the sz version is about 200 years older than the ss form. – clemens Nov 28 '15 at 18:53
  • 1
    @barbarabeeton Thanks for the correction. I've run some tests: with texlive 2013.20140215-1 on Ubuntu, I get the Knuth ß in both cm and lm! In ec I get the Knappen ß. The Knappen ß in my answer is from an OTF of cm-unicode 0.6.3a (2008-03-14). To add to the confusion, the LaTeX font catalogue shows the Knappen ß in cm. I'll edit the erroneous references to cm and lm out of my answer, since things are evidently more complicated than I thought. – Pont Nov 28 '15 at 21:45
  • 2
    @ChristianLindig This is an ſs ligature, not an fs ligature: no f’s were harmed in its making. :) – tchrist Nov 29 '15 at 16:35
18

I don't know why Knuth designed the ß as it is in the cm-fonts. I don't quite remember why Jörg Knappen changed the look of the ß in the ec-fonts, but I do remember that there was some quite heated discussions about the choice.

If you don't like both ß there is no much you can do (apart from redesigning the glyph yourself). But as the cm-super fonts contain both variants of the ß you can use the "old" version also with T1-encoding without switching to the lmodern fonts:

Find cm-super-t1.enc (in fonts/enc/dvips normally), make a copy e.g. cm-super-t1-alt.enc and store it where latex can find it (e.g. in the current directory). Open the new .enc file and change at the end /germandbls to /germandbls.alt.

Find cm-super-t1.map (in fonts/map), make a copy e.g. cm-super-t1-alt.map, store it where it can be found. Open it and replace every occurance of cm-super-t1.enc by cm-super-t1-alt.enc

Then load the new map file e.g. with \pdfmapfile:

\documentclass{article}
\usepackage[T1]{fontenc}
\usepackage[utf8]{inputenc}
%\pdfmapfile{= cm-super-t1-alt.map} % to replace ß in all fonts
% for the example only for one font:
\pdfmapline{= ecrm1000 SFRM1000 " T1Encoding ReEncodeFont " <cm-super-t1-alt.enc <sfrm1000.pfb}
\begin{document}
The original of the ec font (larger to be able to show all in one document):

{\large grüße}

The alternative germandbls.alt from cm-super:

grüße

In the cm-fonts: \fontencoding{OT1}\selectfont 

grüße

In the lmodern fonts:
\fontfamily{lmr}\selectfont

grüße

\end{document}

enter image description here

If you want to use the new ß also with dvips or dvipdfmx or don't want to bother with \pdfmapfileyou will have to activate the new map-file with updmap(-sys),

18

I can't do anything about the font you are using but would recommend to use the lmodern fonts, a modernised variant of the Computer Modern fonts.

\documentclass{minimal}
\usepackage{lmodern}
\usepackage[T1]{fontenc}
\usepackage[utf8]{inputenc}
\begin{document}
\begin{center} 
    Große Straße ließen gießen maßen heißt Spaß Fuß Maß Gruß reißend
\end{center}
\end{document}

enter image description here enter image description here

Addendum: Below is the ß glyph in some well respected fonts for comparison.

enter image description here

15

What you are complaining about is the "ß" of the dc fonts created by Jörg Knappen, the first 8 bit extension of the original 7 bit Computer Modern fonts. This was digitized in the cm-super fonts (see Latin Modern vs cm-super?).

This new ß was disliked a lot when Jörg created the dc fonts (which later became the ec fonts).

The ß as designed by Knuth (and most likely inherited from Monotype Modern 8A) looks like the one in the answer by Christian Lindig.

The Latin Modern fonts (which you get with \usepackage{lmodern}) provide both glyphs (I believe) but default to Knuth's (after complaints from the Germans).

  • I would like to learn more about this. Sources for Knuth's ß being like the one in lmodern? – Sverre Nov 28 '15 at 13:23
  • 1
    @Sverre: Check the metafont sources. :-) Using all the different fonts (the original metafont sources by DEK, the Type1 by Blue Sky/AMS, the dc and ec fonts, cm-super, both versions of latin modern) is still possible, but I leave that to someone else. – Martin Schröder Nov 28 '15 at 13:36
  • 3
    Not to mention the incredibly ugly euro glyph – egreg Nov 28 '15 at 13:43
  • 1
    to see the shapes of the original computer modern, it's best to find a copy of volume e of *computers & typesetting: computer modern typefaces". unfortunately, it's available only as a hardcover book, not electronic. – barbara beeton Nov 28 '15 at 21:09
  • 1
    @Martin Schröder I had a look at the lmodern-documentation -- but I couldn't figure out how to change the glyp of the eszett. How can I change in a latex document set in lmodern the glyp of the eszett to the glyp favoured by Jörg Knappen? – jlk Nov 29 '15 at 13:43
5

This is a ligature of two 's' characters, one following another.

It was in common usage in English as well, until the end of the 1700s.

A single "s" character was often drawn as an "f" without the crossbar. When two consecutive "s" characters were drawn, they were combined into the form about which you are asking.

In English, the first character morphed into the "s" glyph during the Napoleonic Wars. The ligature then disappears from contemporary English usage, but persisted in German.

In Unicode, while ligatures are provided as convenience characters, the standard strongly recommends that ligatures are decomposed into their component character points, while encoding them as a ligature. Unicode is about encoding things as characters, not as glyphs. The Unicode standard shows the ligature of "f" and "i" as its canonical example of ligatures, in Section 2.2.

Examples of the morph of the long s and the fs ligature in English can be found in "The Art of Defence on Foot", by Taylor and Roworth. There are three revisions, 1798, 1804 and 1824. The revisions between the 2nd and 3rd editions are almost entirely changes to the typesetting of the long 's'. (The 1804 Edition also makes extensive use of the 'ft', 'fh' and 'ct' ligatures, and even the 'fk' and 'ff' ligatures).

[ed. Steven B. Segletes] The fs ligature is shown in the following, well known document (circa 1776), both in the words "necessary" and "Happiness" which appear about mid column in lines 1 and 4. Note: it is also used for words beginning in "s", such as "secure" on line 4.

enter image description here

And as far as printed copies, it is likewise so (though presented as both an "ff" in "diffolve"and an "fs" in "Happinefs"), as seen in the first printings of the Declaration (ref: http://www.ushistory.org/us/10g.asp):

enter image description here

  • 2
    ſ ("long s") did not "morph" into s in English. The long s simply came out of use. It's also news to me that English used to employ a ſs ligature. What's your source for that? – Sverre Nov 30 '15 at 16:11
  • 2
    It would be useful to distinguish historical English typography versus German, since the more important change for German typography is the change from Fraktur or blackletter to roman type in the mid twentieth century. – musarithmia Nov 30 '15 at 17:26
  • 1
    @StevenB.Segletes No. As said in a comment above, a ligature for ſſis well established. The printed document you edited in here also has a ligature for ſſ. I just haven't seen a ligature for ſs in English (that is, anything like the ligature ß). EDIT: Sorry, the answer to your question is "yes, this is a ligature", but it's not the ligature we're looking for here. – Sverre Nov 30 '15 at 17:35
  • 1
    @Sverre See Nils answer at tex.stackexchange.com/questions/117095/…. In particular the word "Essay". It references this page: babelstone.blogspot.ca/2006/06/rules-for-long-s.html – Steven B. Segletes Nov 30 '15 at 17:46
  • 1
    @StevenB.Segletes Great - thank you! I wonder, though, whether this ligature was really "in common usage in English", as claimed in the answer above (but that's a separate question). – Sverre Nov 30 '15 at 17:49
1

Here's what the Unicode Standard has to say on the matter of the existence, history and derivation of this glyph:

ß character point 00DF

from C1 Controls and Latin-1 Supplement codepage.

Latin small letter sharp s
= Eszett
German
Uppercase is "SS"
in origin a ligature of [character point] 017f ſ and [character point] 0073 s
[compare to character point] 03B2 ß greek small letter beta

  • Although, as I type, Windows 7 is displaying an ſs ligature glyph to me rather than an actual beta symbol, and if I use the long handed method of typing, I get a glyph - a smiley face in negative. – Euan M Dec 1 '15 at 18:05
  • 1
    Obviously, different fonts draw each ligature and character in their own ways. I'll leave the discussion of the graphic form in question to others. – Euan M Dec 1 '15 at 18:05

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.