31 January, 2008

Esszet or ß

The German character ß includes the only surviving form of the long s in modern typographical usage. Since its sound is ss, it might be thought to be a ligature made up of ſ and s, to be used at the end of a word or a syllable. But the matter is not nearly so simple. What ever it may look and sound like, it is essentially ſ and z, and the rules for its use have as much to do with the quality of the associated vowel as with its position. It is still required in Germany and Austria for printing in German with roman types (and if the printing is in gothic types, ſ must also be used). The roman form appears to have been first introduced in the 1660s by a printer in Sulzbach, a place that is commemorated in the name by which its design was later known, and it was reintroduced in the 1820s by the philologist Jacob Grimm, who subsequently rejected it. German printers using roman type did not normally employ ß at all until its use became one of the topics raised in the 1870s in the context of spelling reform, and a movement began that led to a requirement in 1903 for its universal use in German schools. As might be guessed from these dates, its modern history has been bound up with the unity and identity of the German state.

Esszet or Eszett is used after a long vowel, as in Straße, street, but not after a short one, as in Gasse, alley. New and wide-ranging spelling rules in Germany, introduced in 1996 and legally required in schools since 2006, have reduced the number of occasions that require the use of ß. The rules have not met with universal assent. The use of ß died out during the 20th century in German-speaking Switzerland, but its retention has been strongly defended in Germany. Latin Small Letter Sharp S, as ß is known, not only has its place among the glyphs of the ASCII character set (since 1986) and Unicode (U+00DF), but it is accessible on mobile phones. Young Germans (and Swiss and Hungarians) currently use it in their texting as a substitute for ss or sz, regardless of spelling rules, in order to reduce the overall character count of their messages. Moves are currently under way, not for the first time, to introduce a majuscule version.

Since ß is a ligature that is made up of ſ and z it is called by the names of these letters, Esszet; perhaps more widely and correctly spelt Eszett; or Eszet: the spelling is inconsistent because it was not generally recognized in the past as a real word, but merely the phonetic rendering of the two letters: Grimm used the term eszet – so spelt – in the foreword to his own dictionary, but there is no headword for it in the body of the work. The word, however spelt, does not appear in some other dictionaries. It is sometimes also known as scharfes S, or unvoiced letter s. The expression ‘sharp s’ in the title for the ASCII and Unicode glyph quoted above is a literal rendering of the German term. In English it makes no sense at all.

In the historical form of the language known as Old High German, and the subsequent Middle High German, until about 1500, there were two closely related sounds for unvoiced s for which the letters s and z were used. Since z was also used (as it still is) for the sound ‘tz’ (as in Zug, train), the second s-sound was spelt sz. And because the s component did not come at the end of the word or the syllable, it appeared as ſ.

The ſz combination can be seen above in separate letters in the word boſzheit in the Weltgericht, a surviving fragment that was presumably printed by Gutenberg in about 1450, using an early state of the so-called DK type. The ligature ß is used in the name of the printer Peter Schöffer, Petrus Schoiffer de Gernßheim, in this colophon of 1477.

The early history of this combination in both manuscript and type is complex and in need of further investigation, as an essay by Max Bollwage of 1999 indicates. The form of z used with long s does not always correspond with the normal z in writing or type, and may have a different derivation. (See the bibliography at the end of this post.) However, ß is visible in the Fraktur of the early 16th century which became the dominant German national type. Below is a detail from Dürer’s Underweysung der Messung, printed in Nürnberg in 1525 by Hieronymus Andreae, whose Fraktur type was made following patterns drawn by the calligrapher Johann Neudörffer. The Esszet can be seen in lines 1, weyß, 3, auß, and 6, reyß. There are two other points of typographical interest worth noting in this example. The round, or ‘ragged’ r, as it was known in English, that follows letters that are rounded behind like b and o, as in breiten in line 2 and vor in line 3. By the 18th century its use had largely ceased in Germany. Note too that lower case is used to begin nouns, such as strich. In Germany by the end of the 16th century they nearly all began with a capital letter, and by and large they still do.

The widespread confusion that still exists about the nature of ß derives from the fact that a shift in the pronunciation of modern High German had led by the 18th century to the same pronunciation of ss and sz, so that the ß character was now used purely as a visual sign to distinguish the use of the long and short vowel. In the late 18th century the dictionary-maker Johann Christoph Adelung noted the complexity of keeping so many letters for the single s-sound: initial and medial ſ, final s, and ß after the long vowel; and also for double s: medial ſſ, final ſs, but ß following the long form of the vowel. He added that this had led to discontent among teachers: ‘It is certainly true that that having four different ways of writing the sound s has its difficulties and irritations, especially when ß is used not only for the unvoiced sound but also for double s. But nothing better has been suggested in its place.’

Roman type was rarely used for setting the German language before the 19th century, but it does appear in a work printed in Sulzbach in East Bavaria by Abraham Lichtenthaler, who attracted attention to his printing-office by printing Knorr von Rosenroth’s Kabbala denudata (1677, etc) there. The lines above, in a tiny type, apparently of the 16th-century and French in origin, with the addition of W and w and vowels with Umlaut as well as ß, are from the German translation by F. M. van Helmont of Boethius, The consolation of philosophy, printed by Lichtenthaler in 1667. (The preliminary pages and the text are set in Fraktur, but the roman type is used for additional verses which appear throughout the translation.) He left his mark on German typography by introducing in the Boethius a design of ß that would be adapted for use in roman type by reformers in the 19th century and called the Sulzbacher form. His use of this ß is referred to by Uta Stötzner in an article in the journal Signa, no. 9 (2006).

Printers who in Germany who began to flirt with the use of roman type for German texts at the end of the 18th century had no easy answer to the rendering of ß with the new founts. When Goethe’s Römische Carneval was printed by Johann Friedrich Unger in Berlin in 1789 in a roman type by Firmin Didot, long s was not used in the text, following the current practice in France – except, paradoxically, for the words in which ß would have been used, which he printed ſs, using a Didot-style long s which must have been cut for this purpose. The contrast in contemporary attitudes to the use of roman types for works in German is nicely shown in two well-known letters that were written in 1794. The poet Wieland expressed to his publisher Göschen his feelings about the roman types made ‘after Didot’ by Justus Erich Walbaum that were used for his own collected works: ‘I cannot say how much I enjoy the pure beauty of those letters. Each is in its way a Medici Venus.’ Goethe’s mother, Elizabeth, wrote to her son, ‘I am more pleased than I can say that your own works have never seen the light of day in those terrible Latin types. Well, they may have been tolerable in your Roman Carnival, but otherwise do stay German even in your type, I beg you. I was going to subscribe to the works of the good Wieland – but I was put off by the new Fashion, and let it go.’ The wars with Revolutionary and Imperial France helped to damp enthusiasm for the alien types. Mrs Goethe’s dislike of roman and strong attachment to ‘German’ type is evidence of a feeling that was widespread and lasted for a long time.

Nonetheless there was some use of roman types for learned publications during the early years of the 19th century. In many German type specimens at this period the showings of roman and italic types, where they occur, are set in Latin or French. Where German is used, ß is generally set as ſs, following the style used by Göschen. Here is a setting of the ‘Didot’ style of roman made by the punchcutter and typefounder Prillwitz in Jena, from his specimen of 1790, with ‘daß’ in the fourth line.

However in the specimen of Carl Tauchnitz, Leipzig, 1825, the German texts set in the ‘French’ roman and italic types use simply ss, which would become the common practice of German printers.

The second edition of the Deutsche Grammatik of Jacob Grimm (Göttingen, 1822–6), is set in a French-style (but probably German-cut) roman, using not only ſ but also an ß that was presumably made for him, reminiscent of the character used some 160 years earlier by Lichtenthaler. The illustration above is from the second volume, dated 1826. The italic version, shown below, from a headline in the first volume of 1822, seems closer to the cursive ſs character that is discussed later. (In the same work Grimm reverted to the early 16th-century practice of using small letters for all nouns, the beginning of the so-called Kleinschreibung that would have echoes in the 20th century in the poetry of Stefan George and the work of modernist designers. It seems worth noting that Lichtenthaler’s Boethius of 1667 had used small letters to begin most nouns in both the Fraktur and the roman parts of the text, unusually for its date.) Grimm had changed his mind when, with his brother Wilhelm, he began publication of his great German dictionary in 1854. In this work, which is set throughout in roman types, although he repeated his use of small initial letters for nouns he proposed the total rejection of Fraktur types. To distinguish ß, he proposed to go back to the spelling sz, a combination which, as he noted, was philologically correct, currently used in Hungary and Lithuania, and had the advantage that it worked equally well in capitals.

By and large the German printing trade took little notice of Grimm’s proposals. Although most books and newspapers were set in Fraktur type throughout the 19th century, roman type was increasingly being used for scientific and technical books and journals. Capital letters were used for all nouns, and words with ß in Fraktur were spelt in roman with ss, and neither ſs nor sz. The Journal für Buchdruckerkunst, a major printing trade monthly, was set in Fraktur until 1874, when it changed to roman, remarking simply that, ‘we have put on new clothes’. It printed notes of congratulation from readers in Germany and abroad. For some years, advertisers in the journal had used roman type, and ss for ß. Similarly, another printing trade journal, the Archiv für Buchgewerbe, changed to roman in 1886, and again used ss for ß. So far as I can tell this usage reflects the general practice of German printers and typefounders during the larger part of the 19th century, as the use of roman type spread. The typefounders did not make ſ and ß for roman and italic founts, nor for new types like sanserifs, and the printers did not ask for them.

And so things might have remained. But after the making of the new unified German state that began with a customs union and led in 1871, after the defeat of France, to the creation of the German Empire, with its capital city in Berlin, the first of several successive moves was made towards the reform of German spelling and its consistent teaching in German schools. At a conference in Berlin in 1876 a move was made towards the introduction of ß in roman type for the first time. A trade society in Leipzig, the Typographische Gesellschaft, reviewed a series of designs that were submitted for discussion and these were published in 1879, in a report in the Journal für Buchdruckerkunst.

It is was noted that some of them were based on ſz and others on ſs. The pattern that met the greatest support from the typefounders, notably Ferdinand Theinhardt of Berlin, were 1 and 3 at the bottom left, which it was noted were based on the ß introduced by Grimm. Later, presumably from an awareness that the original model was the character used by Lichtenthaler, this became generally known as the ‘Sulzbacher Form’. Here is an early use, in a work on the invention of printing, J. van der Linde, Geschichte der Erfindung der Buchdruckerkunst, Berlin, 1886. The setting is in one of the ‘old style’ romans known in Germany as ‘Medieval’, of which the design originated in Britain, and exceptionally – or was the practice commoner than we think? – it uses Grimm’s Kleinschreibung or small initial letters for nouns. In this instance the design of the ß, which seems to have been intended for a relatively narrow modern-face type, does not match that of the old style.

In 1903 a pronouncement was made in the name of the association of printers of Germany, Austria and Switzerland that a ß character of the Sulzbacher form, consisting of a ligature of ſ and z, would henceforth be used in roman type. In his essay of 2006, Mark Jamra illustrates the document that was published in a trade journal dated 9 July 1903 (Zeitschrift für Deutschlands Buchdrucker, Steindrucker und gewandte Gewerbe) in which the approved forms of the ‘Sulzbacher’ form of ß were described, with examples, as comprising ‘the so-called lower case long ſ, attached at the head to a z, of which the lower bow ends in a thin or medium-thin line, or a full-point’. The Sulzbacher ß has often been criticised for being ugly and too easily confused with roman capital B and Greek Beta. But it is essentially the model for the ß that has been used in many types ever since, including the Georgia screen font in which some readers will see this text.

Since the use of roman type made a majuscule form at least possible (as it was not in Fraktur, of which the majuscules are rarely used to make words), it was suggested that, since no agreement had been reached concerning its design (and SS was incorrect), SZ would provisionally be used in its place, as in PREUSZEN for Preußen (Prussia). But it was also decided that the typefounders should give their attention to it, perhaps by offering a prize for a design. There was some discussion of possible models in the printing journals, and capital forms do appear in the Kleukens Antiqua of the Bauer foundry (1910), and the Schelter Antiqua of Schelter & Giesecke (1912), shown below. They seem rarely to have been used, and their existence was largely forgotten until recently.

A new type, with a Sulzbacher ß that inclines towards the appearance of a combination of ſ and s, was introduced by the Archiv für Buchgewerbe from the date of its first issue under a new publisher in 1900. It is the Römische Antiqua (1888) of Genzsch & Heyse, designed by H. König:

During the next few years the journal published a number of articles on the subject of ß. A selection is listed at the end of this text. Other designers headed in the opposite direction, making a more explicit combination of s and z. The example below, using an archaic form of z, is the Ingeborg Antiqua (1910) of F. W. Kleukens, cast by the Stempel foundry:

And in Koloss (Ludwig & Mayer 1923) by Jakob Erbar, the z (once again in the archaic form) comes loose from the s altogether:

Many founders kept closely to the basic Sulzbacher form, which even found its way into the new sanserifs of the late 1920s, some of which were also provided with long s, like Renner’s Futura (Bauer 1927) and Koch’s Kabel (Klingspor 1927), below:

It remains to note how designers eventually subverted what many perceived as the ugliness of the Sulzbacher form, and the oddity of some of the versions that emphasized the z, by changing ß more clearly to a ligature of ſs. To this end a note is needed of the ligatured form of long and short s that had in fact existed in the Italian humanistic cursive since the end of the 15th century, and was included in the chancery cursive.

In his Operina of 1522, above, Arrighi shows it as one of two permissable ligatured forms for ss in the chancery cursive: the other is formed from two long esses.

In his own writing book of about the same date, above, the Venetian writing master Tagliente shows this ligature not only in his cursive but also in his Lettera antiqua tonda, or roman script (neceſsario in line 3).

Two alternative ligatures for ss, ſſ and ſs, are seen in the italic type above, the Parangon italic of the French punchcutter Robert Granjon, a type of which the use was widespread in many countries during the 16th and 17th centuries. The combination of types for long and short s, ſs, is also used together within words in many texts in both roman and italic types in Italy and France during the 16th century, apparently as an alternative to the normal ligature of two long esses which is seen in the example above.

There is a technical matter that needs explaining here, since commentators are often unaware of its relevance. In metal printing types, long s, like f, kerns or overhangs the letter that follows, and if the letter that follows is a tall one, there is likely to be a collision and the kerned part may bend or break. It was for this reason that single ligatured types were eventually made for most of the awkward combinations, like f and i, or ſ and i. Long s followed by short s is a safer combination in type, physically, than two separate long esses.

Nonetheless, the example above in the Granjon Parangon, showing a ligature of two long esses, followed by ligatures for ſi and ſs, all in the same line, is a puzzle. Perhaps there were not enough ſſ ligatures in the case, or perhaps the compositor used the alternative form for the sake of variety. But the lack of ligatures hardly seems to explain the frequent use of the medial and final form together in the middle of words in several 16th-century Italian and French texts set in roman types, such as the example below, from a work printed in Venice in 1571. Note ſteſsi in line 2 and cariſsimi in line 8, but miſſero, with a ligatured form of ſſ, in the first line. (Is the intention possibly to distinguish the voiced ſſ of miſſero from the unvoiced consonants of the other words, in the manner suggested earlier in the century by the spelling reformer Trissino?)

Whatever the explanation, it should be noted that the combination of ſs as a single sort is seen only in italic type, presumably on the model of the calligrapher’s ligature, and has not been found in roman type – unless, as seems not unlikely, Lichtenthaler’s ß was indeed an exercise in making an upright form of the existing italic ligature.

In 1940 Jan Tschichold published an essay on ß in which he asserted with his characteristic force and dogmatism that its basis was ſ and s, and that all the suggestions that it was ſ and z had simply been wrong. He repeated the assertion in his Meisterbuch der Schrift (1952), issued later in English as Treasury of alphabets and lettering. Later writers agree that in this instance it was Tschichold himself who was mistaken, but perhaps it was his authority that encouraged more type designers to explore this new model for the roman ß. Hermann Zapf’s Palatino, drawn for Stempel in 1948, is an early example. It has now become familiar in such fonts as Adobe Garamond, in which the italic form is borrowed from the original by Robert Granjon, which had also been a special character made for the Monotype Garamond of 1923. The Sulzbacher form of ß had been made by Monotype for Times New Roman and its other types before the Second World War. Here are examples of setting in the 17th-century ‘Luther Fraktur’ of the Stempel foundry (from a modern adaptation), Times, Linotype’s Palatino, and Adobe Garamond:

It is currently claimed in Germany, among reasons why ß should be preserved, that it distinguishes some different words with different sounds and meanings that would otherwise look the same. It is also said that since some personal names are traditionally spelt with ß, and since legal forms must be filled in using capital letters, there is a need for the majuscule version that once existed but has now been forgotten. Cases are cited in which the discrepancy between the look of a name in minuscules in a birth certificate and in capitals on a form has caused real confusion and bureaucratic obstruction. Here is a problem to which the simple, logical suggestion made by the brothers Grimm in 1854 (use sz) would have had provided an answer. There is an active movement reviving the idea of the addition of majuscule Esszet to the character set, and the latest news indicates that it should shortly have an identity in Unicode. The subject is reviewed in detail in the journal Signa.

In Hungarian the letters sz, as the brothers Grimm noted, are treated as a single consonant with the sound of s. When he returned home and printed in his native Kolozsvár, Nicholas Kis made ligatures for sz, with initial long s, for his own use in both roman and italic. The design illustrates the problems in a roman type of uniting ſ with a broad letter like z, and this ſz ligature never seems to have caught on generally. The example above is from György Haimann, Nicholas Kis (Budapest, 1983). But, as noted on the website Signographie, a ß not unlike that of Lichtenthaler was used for sz in a roman lower case in a Latin–Hungarian–German dictionary in 1782.


There is still work to do on the origins of the ß. The work of Uta Stötzner on this subject is especially valuable. The essay published in 2006 by Mark Jamra, to which reference is made above, is also a useful and well-informed summary, and so is the entry in Wikipedia, which has become more wide-ranging. (Search under Eszett, or the character ß. The German-language version is the most extensive.) As this link to Typophile Forum shows, the revival of the debate concerning the majuscule version of ß, which originally arose over century ago in the context of the creation of a German national orthography, still continues. My initial suspicion is confirmed: the use and retention of ß, and consequently its belated (and in my view hardly necessary) insertion into the majuscule character set, is felt in Germany to be bound up with national identity. The feeling is undoubtedly strong, but how widely it is shared is difficult to gauge. In German-speaking Switzerland ß was gradually abandoned during the 20th century, and its loss seems hardly to be noticed there, let alone regretted.

Here are some printed sources:

Uta Stötzner, ‘Die Geschichte des versalen Eszetts’, Signa, Nr. 9 (2006), pp. 21–38.

Herbert Brekle, ‘Zur handschriftlichen und typographischen Geschichte der Buchstabenligatur ß aus gotisch-deutschen und humanistisch-italienischen Kontexten’, Gutenberg-Jahrbuch 2001, pp. 67–76.

Max Bollwage, ‘Ist das Eszett ein lateinischer Gastarbeiter? Mutmaßungen eines Typografen’, Gutenberg-Jahrbuch 1999, pp. 35–41.

John Flood, ‘Jacob Grimm’s advocacy of roman type’, in ‘Das unsichtbare Band der Sprache’: studies in German language and linguistic history in memory of Leslie Sieffert (Stuttgart: Heinz, 1993), pp. 279–312.

‘Das Antiqua-sz: Referat der Typographischen Gesellschaft zu Leipzig’, Journal für Buchdruckerkunst, 42 (1879), col. 852–6, 43 (1879), col. 873–7.

Heinz König, ‘Vom Antiqua-Eszet’, Archiv für Buchgewerbe, vol. 40 (1903), pp. 7–9.

H. S. [= Hermann Smalian?], Leipzig, ‘Nochmals das Antiqua-Eszet’, Archiv für Buchgewerbe, vol. 40 (1903), pp. 103–4.

Heinrich Schwarz, ‘Das Antiqua-Versal-SZ’, Archiv für Buchgewerbe, vol. 43 (1906), pp. 93–6. ‘Nochmals das Antiqua-Versal-SZ’, Archiv für Buchgewerbe, vol. 44 (1907), pp. 111-12.

Wilhelm Hellwig, ‘Das lange s in der Antiqua und die Rechtschreibung’, Archiv für Buchgewerbe, vol. 49 (1912), pp. 306–9.

The original version of this post, which was intended simply as a brief supplement to the earlier one on long s, was an ill-prepared venture into the topic. The present considerably revised and corrected version owes much to the kind intervention of others. More correction, if it is needed, will be welcome. It can be claimed to include some typographical illustrations that are not easy to find elsewhere.