Transliteration and Scripts

4.1 Scope of this chapter

This chapter covers words and passages from languages written in scripts other than the Latin alphabet, and characters outside the ordinary keyboard. It sets out how to give such a word in English prose; how to romanize Hebrew, Greek, Coptic, Egyptian, and the other languages the field draws on; when to set the original script; how to treat Enochian and the magical alphabets; how to transcribe manuscripts; and how to enter and display the special characters the field uses constantly. Italics for non-English words are treated in 3.10.

4.2 Three forms of a non-English word

A word from a language in another script reaches English prose in one of three forms. A received form is the spelling that English, or the field’s own literature, has made its own: Kether, Thelema, Osiris, yoga. A romanization represents the word in Latin letters by a stated system: Ḥokhmah, thelēma, nṯr. The original script gives the word in its own writing system: חכמה, θέλημα. Use the received form wherever one exists for the referent (see 4.3). Romanize any other word by the system this chapter gives for its language. Set the original script where the letters themselves are the argument (see 4.8), in quoted passages of original text, and where a title contains it (see 4.17).

4.3 Received forms

Use a received form for its own referent, in scholarly and devotional writing alike. Do not correct it toward a romanization; an account of the Tree of Life as the Golden Dawn and Crowley taught it writes Chokmah throughout. Do not extend it to a referent it does not name: the sefirot of the Zohar take the romanization that Kabbalah scholarship uses (see 4.11). Persons, places, and works with an established English form keep it (Maimonides, Avicenna, Thebes, Sefer Yetzirah). The received forms of the field’s terms are fixed in this chapter and in the Word List; for theonyms, see 3.36.

4.4 One system for each language

Romanize each language in a work by one system, and name it on the style sheet: the system this chapter gives for that language, another established system where this chapter allows one (see 4.14), or, for a language the chapter does not cover, the system 4.34 directs. Record on the style sheet, for each language, whether the work uses full or simplified romanization (see 4.5). A book by several authors, or an issue of a journal, uses one system and one form throughout. A writer who does not read a language does not romanize it from the script. Such a writer takes the romanization from a reliable source, cites the source, and has it checked (see 4.10).

4.5 Full and simplified romanization

Full romanization writes every mark the system provides: macrons, dots and other diacritics, and the half rings for alef, hamza, and ayin. Simplified romanization omits the diacritics and keeps the letters and digraphs; it may keep alef, hamza, and ayin as 4.52 allows. Use full romanization in linguistic studies, in editions and transcriptions, and in other specialized work. Elsewhere prefer simplified romanization. Whichever form a work uses for a language, no word in it keeps some diacritics and drops others.

4.6 Forms in quoted matter

Quote a romanized word, name, or title exactly as the source has it, even where the source’s system differs from the one the work uses. A different system is not an error and is never marked [sic]. Where a reader might not recognize a quoted form, give the house form in the writer’s own sentence or in brackets after the quotation. Quoted script is reproduced with the source’s spelling, pointing, and accents. A published translation is quoted with the translator’s forms.

4.7 Script beside a romanization

At its first mention, a romanized word may be followed by its original script in parentheses, with a gloss after the script where one is needed; thereafter the romanization stands alone. Do not move the script to a note. A received form may take the same parenthesis where the script helps the reader. In a glossary or a list of terms, every entry may carry its script. A word whose letters are the argument takes its script wherever the argument is made (see 4.8).

4.8 Script in arguments from letters

Wherever an argument turns on the letters of a word, give the word in its original script. Such arguments include a numerical value, a spelling, a word formed from initial letters (notariqon), a substitution of letters (temurah), and any letter-by-letter comparison. A romanization or English form may accompany the script, and with a numerical value one always does (see 6.59); neither replaces the script. Give Hebrew without vowel points and Greek in capitals without accents or breathings, unless the argument concerns the vowels or accents themselves. The presentation of values and calculations follows 6.58–6.67.

4.9 Setting script in running text

Set a word or passage in a non-Latin script upright, never in italics, even where its romanization would be italic and even inside an italic title. Do not enclose it in quotation marks. Run a word, phrase, or short passage into the line of text, and set a longer one as a block quotation by the measure of 3.68, with its translation following. Every passage in script is accompanied by a translation, a romanization, or both. For setting Hebrew and Arabic within English, see 4.58.

4.10 Verifying non-Latin script

Every passage in a non-Latin script is checked before publication by someone who reads that script. An editor who does not read it compares the passage character by character with the author’s source, queries every difference (see 2.25), and never retypes script from a printed page by eye. The commonest errors confuse letters that resemble one another: in Hebrew, ד with ר, ה with ח, ב with כ, and ו with ן; in Greek, ν with υ and ς with ϛ. A mechanical check of the characters replaces neither the reader of the script nor the comparison. Where no reader of the script can be found, the author answers for it, and the romanization appears beside it.

4.11 Sephiroth and Qabalistic terms

The Sephiroth and the other terms of Hermetic Qabalah take the received forms that the Golden Dawn and Crowley established. Use them wherever the subject is the Hermetic or Thelemic system, including scholarly writing about it. Where the subject is Jewish Kabbalah or its scholarship, use the romanization of 4.14. Table 4.1 gives both, with the Hebrew. Daath, which is not counted among the ten, is listed last.

Table 4.1. The Sephiroth

4.12 Sephirah, Qliphoth, and Ain Soph

The received forms are Sephirah, plural Sephiroth, and Qliphoth, capitalized; the romanized forms are sefirah, plural sefirot, and qelipot, lowercase. Keep each ending with its own stem: Sephiroth and sefirot, never Sephirot or sefiroth. The three veils are Ain, Ain Soph, and Ain Soph Aur (אין, אין סוף, אין סוף אור); in writing on Jewish sources, the infinite is Ein Sof. The spellings Qabalah, Kabbalah, and Cabala are distinguished in 3.5.

4.13 The Hebrew letters

Name the Hebrew letters by the house names in table 4.2, capitalized and roman: the letter Aleph, the path of Tau. Use these names in all writing on the Hermetic and Thelemic systems. A work on Jewish tradition or on Hebrew as a language may instead use that scholarship’s names (alef, bet, gimel), lowercase; use one set throughout a work. The table also gives each letter’s romanization. In simplified romanization, ʾ and ʿ are omitted, ḥ is written h, ṭ is written t, and ś is written s.

Table 4.2. The Hebrew letters

4.14 Romanizing Hebrew

Romanize Hebrew by table 4.2, or by another established system, such as the ALA-LC table for Hebrew, named on the style sheet and used throughout (see 4.4). Table 4.2 follows the traditional pronunciation used in Jewish studies, writing b or v, k or kh, and p or f according to whether the letter takes a dagesh; it writes Qoph as q, and in its full form it marks Aleph, Cheth, Teth, Ayin, and the sin form of Shin. In that system, supply vowels as a, e, i, o, u, unmarked for length; write a vocal shewa as e and a silent shewa as nothing. Do not double a consonant for dagesh forte. Final letters romanize as their ordinary forms. Join the article and the other one-letter prefixes to the following word with a hyphen.

Titles of Hebrew works are romanized by the system the work uses. A work with an established form keeps it (see 4.3), and an edition or translation published under another form keeps its own title.

4.15 Hebrew script

Set Hebrew without vowel points or cantillation marks unless the vowels are part of the argument or the passage is quoted from a pointed text. Use the final forms (ך ם ן ף ץ) where the source uses them. In gematria, the value given to a final letter is a matter of counting convention (see 6.61) and is never signaled by changing the letter. Hebrew pasted from a PDF often arrives as presentation forms (the characters at U+FB1D–U+FB4F) rather than as letters and points; replace them before submission (see 4.57).

4.16 Divine names

The Hebrew divine names in common use take their received forms: Eheieh, Yah, El, Elohim, Adonai, Shaddai El Chai. Write the Tetragrammaton as YHVH where the subject is the Hermetic or Thelemic tradition, and as YHWH where it is the Hebrew Bible or Jewish tradition; its letters are spelled out by their house names, Yod Heh Vau Heh. Jehovah, IHVH, and other forms appear only in quotation. Yahweh is used only for the god of ancient Israel as historians describe him.

In script, give a divine name unpointed (see 4.15); never print יהוה with the vowels of Adonai except in quoting a pointed text that has them. The Golden Dawn and Crowley often spell Hebrew names in Latin capitals letter by letter (AHIH, ShDI AL ChI); reproduce these in quotation, and in the writer’s own prose use them only beside the Hebrew, as a key to its spelling.

4.17 Hebrew in titles

A title in Latin letters that contains a Hebrew or Greek word, as some Liber titles do, keeps the word in its own script wherever the title is printed; a title wholly in a non-Latin script is romanized (see 5.20). Do not substitute a romanization (ADNI, Adonai) for it. The script stays upright within the italic title (see 4.9), and in web output Hebrew is isolated as a right-to-left run (see 4.58).

4.18 Hebrew numerals

Mark a number written in Hebrew letters with the geresh (׳, U+05F3) after a single letter, and with the gershayim (״, U+05F4) before the last letter of a number of two or more letters. Never use the typewriter apostrophe or quotation mark, or the curly quotation marks, in their place.

4.19 Romanizing Greek

Romanize Greek by the ALA-LC romanization table for Greek. Table 4.3 gives its letters. For diphthongs, the iota subscript, and other details, follow the ALA-LC table itself; macrons, accents, and breathings are treated in 4.20. Greek written after the fall of Constantinople in 1453 is modern Greek and follows the modern Greek provisions of the same table.

Table 4.3. Romanization of Greek

4.20 Macrons, accents, and breathings

The macron marks eta (ē) and omega (ō) and nothing else; it does not mark stress. A romanized Greek word never carries an acute, grave, or circumflex accent. Write the rough breathing as h before the vowel or diphthong it stands on, and as rh on rho; write ρρ as rrh. Smooth breathing and accents are not represented. Upsilon is y standing alone and u in a diphthong (au, eu, ēu, ou, ui, ōu). Gamma before γ, κ, ξ, or χ is written n. Simplified romanization drops the macrons and keeps the h.

4.21 Received Greek forms

Greek words the field has made its own take their received form, roman, without macrons: Thelema, Therion, Agape, Aeon, Logos. Greek words entered in the dictionary take the dictionary’s form (gnosis, psyche, eros). A Greek word discussed as a word is romanized and italic, whatever its English form (see 3.9): thelēma, psychē. A motto or title keeps the form its bearer or author gave it (see 3.25).

4.22 Greek in script

Set ancient and medieval Greek in polytonic script and modern Greek in monotonic script. Quote an edition’s Greek exactly, including its accentuation, punctuation, and use of capitals. In an argument from letters or values, give the word in capitals without accents or breathings (see 4.8). For entering accented vowels, see 4.56.

4.23 Greek numerals

Write the numeral letters with their own characters: stigma ϛ (U+03DB) for 6, koppa ϟ (U+03DF) for 90, and sampi ϡ (U+03E1) for 900. Digamma ϝ (U+03DD) is also a correct 6 where the source uses it. Never write final sigma ς (U+03C2), which counts 200, for stigma. Mark a number written in letters with the keraia (U+0374 GREEK NUMERAL SIGN) after it, and a thousand with the lower keraia (U+0375) before it. Normalization stores the keraia as U+02B9 MODIFIER LETTER PRIME (see 4.57).

4.24 Coptic names and romanization

Romanize Coptic by the ALA-LC romanization table for Coptic. Names and titles that reached the field through Greek or Latin keep their received forms: Pistis Sophia, the Books of Jeu. Quote Coptic text from a named edition, with that edition’s divisions and conventions, and give a translation. A writer who works from a translation cites the translation. For citing the Nag Hammadi Codices, see 8.37.

4.25 Coptic script and its encoding

Encode Coptic with the letters of the Coptic block (U+2C80–U+2CFF), except the seven letters Coptic took from Demotic (shei, fei, khei, hori, gangia, shima, dei), which are encoded at U+03E2–U+03EF. Older digital texts encode all of Coptic with Greek letters; re-encode such text. Encode the supralinear stroke by its use. Over a single consonant, marking it as syllabic, it is a combining macron (U+0304). Over a numeral or an abbreviation, such as a nomen sacrum, it is a combining overline (U+0305) after each letter it covers. Over a name or word that it sets apart, it is a combining macron left half (U+FE24) after the first letter, a combining conjoining macron (U+FE26) after each letter between, and a combining macron right half (U+FE25) after the last. Never encode it as a spacing macron or a line drawn by the word processor.

4.26 Egyptian names in English

Give Egyptian gods, kings, and places their conventional English forms: Osiris, Isis, Horus, Nut, Thoth, Anubis, Khonsu, Ra; Ramesses II, Hatshepsut; Thebes. Where a Thelemic deity bears a received name, use it for the Thelemic deity and the conventional form for the Egyptian one, choosing by referent (see 3.36). The same holds for persons: the historical priest of the stele is Ankhefenkhonsu, and the figure of the Thelemic texts is Ankh-af-na-khonsu. Do not reshape a conventional form to resemble a Thelemic one, or the reverse, and do not supply an Egyptian transliteration for a Thelemic name unless an Egyptological source gives one.

4.27 Egyptian transliteration

A writer who transliterates Egyptian uses the Leiden Unified Transliteration, adopted by the International Association of Egyptologists in 2023, and takes its characters from the association’s published table. A writer who does not read Egyptian and takes transliterations from an Egyptological source gives them in that source’s convention, cites the source, and does not convert them. Transliterations quoted from older conventions, including Budge’s, stay as printed (see 4.6). Set transliteration in italics. It uses characters that ordinary text fonts may lack, among them ꜣ (U+A723), ꜥ (U+A725), ḥ (U+1E25), ḫ (U+1E2B), ẖ (U+1E96), š (U+0161), ṯ (U+1E6F), and ḏ (U+1E0F). Never write yogh (ȝ, U+021D) or the digit 3 for ꜣ.

4.28 Sign, transliteration, and reading

Keep three things apart. A hieroglyphic sign is cited by its number in a named sign list. A word is given in transliteration. The conventional reading by which Egyptologists pronounce a transliterated word, inserting e between consonants, is a reading only and says nothing about how the word was spoken. Never present a sign number as a transliteration, or a reading as either.

4.29 Hieroglyphs, hieratic, and demotic

Hieroglyphs are encoded in Unicode (U+13000–U+1342F, with later extensions), and an isolated sign or a short horizontal group may be set as text in a font that carries them. Set anything that requires signs to be arranged in blocks, and any passage longer than a few signs, as an image made with hieroglyphic typesetting software, with the transliteration beside it. Hieratic and demotic are reproduced only in facsimile or photograph, and are always transliterated.

4.30 Arabic

Romanize Arabic by the ALA-LC romanization table for Arabic, except that hamza is written ʾ (U+02BE) and ʿayn ʿ (U+02BF), not the table’s ʼ and ʻ. Write the article as al-, joined by a hyphen and never assimilated, lowercase except at the beginning of a sentence. Write tāʾ marbūṭah as -ah, and as -at when the word is followed by a noun it governs. Names and words with an established English form keep it: Avicenna, Sufi. A work known to the field through a Latin translation is called by the title of the version consulted, with the Arabic title given once.

4.31 Sanskrit and Pali

Romanize Sanskrit and Pali by the International Alphabet of Sanskrit Transliteration (IAST). Words entered in the dictionary keep its form, roman and without diacritics: yoga, karma, mantra, guru, chakra, nirvana, and among the terms of Crowley’s yoga, asana, pranayama, dharana, dhyana, and samadhi. Other words take IAST and are italicized at their first use and set in roman thereafter (see 3.10), Crowley’s yama, niyama, and pratyahara among them. In a work that draws on both languages, name the language of any term whose Sanskrit and Pali forms differ. Quote Crowley’s own spellings of Sanskrit as he wrote them (see 4.6).

4.32 Chinese

Romanize Chinese by Hanyu Pinyin, without tone marks unless the argument concerns tones. In the writer’s own prose, give titles and names in Pinyin: the Daodejing, the Yijing, Laozi, Zhuangzi. Keep older romanizations in quotations and in the titles of works that use them. English words of Chinese origin take the dictionary’s form: Tao, Taoism, yin, yang.

4.33 Akkadian and Sumerian

Mesopotamian gods, heroes, and texts take their conventional English forms: Tiamat, Marduk, Ishtar, Inanna, Gilgamesh, the Enuma Elish. Quote Akkadian or Sumerian in the transliteration of the edition cited, with its conventions, and name the edition. Set names that occur only in modern occult texts as those texts print them, without the apparatus of Assyriological transliteration (sign indices, superscript determinatives, hyphenated syllables) that the texts themselves do not carry.

4.34 Languages not covered here

For a language this chapter does not cover, follow CMS chapter 11, or, where CMS gives no system, the convention established in current scholarship on that language. Name the system on the style sheet (see 2.2). Latin needs no romanization; for the orthography of early printed and manuscript Latin, see 4.47.

4.35 Enochian

Enochian is the language, script, and magical system recorded in the spirit diaries of John Dee and Edward Kelley in the 1580s, together with its later development by the Golden Dawn and Crowley. Dee called the language Angelical; a writer may use that name for Dee’s own material where the distinction matters, saying so at first use. Spellings differ among Dee’s manuscripts, Casaubon’s A True & Faithful Relation (1659), the Golden Dawn’s papers, and Crowley’s Vision and the Voice, and none is standard. Give a word as the source used spells it, name that source at first use, and keep to it. Do not harmonize forms from different sources; where they differ and the difference matters, give both and name both. For the sources themselves, see 6.52.

4.36 Enochian words in running text

Set an Enochian word cited in running prose in italics, like other non-English words, and gloss it at its first use. The names of angels, Governors, and Aethyrs are proper nouns, set in roman in the capitalization of the source edition. Where an Enochian letter represents more than one Latin letter (see table 4.4), keep the Latin letter the source uses. Barbarous names that are not Enochian follow 3.11.

4.37 Quoting the Calls

Quote an Enochian Call, or any passage of Enochian text, exactly as the named edition prints it, with its spelling, capitalization, punctuation, accents, and word division. Give the English from the same source, or name the translation used; do not pair one source’s Enochian with another’s English without saying so. A pronunciation is attributed to its source (Dee’s notes, Golden Dawn practice, or a modern author) and presented as a pronunciation, never as the spelling.

4.38 The Enochian letters

The Enochian script has twenty-one letters. Name a letter by its name, capitalized and roman, and give its Latin value at first mention: Gisg (T). Table 4.4 gives the names in their usual form, with the Latin letters each represents.

Table 4.4. The Enochian letters

4.39 Alphabets not in Unicode

Enochian, Theban, Malachim, Celestial, Passing the River, and the other magical alphabets that Unicode does not encode are never set in type. In the text, give the romanized form. Where the letterforms themselves are the subject, reproduce them as an image (a vector image where possible), name the source of the letterforms (a manuscript, a printed edition, or a modern drawing), and give the romanization beside the image or in its caption. For alt text, see 2.11.

4.40 Theban and other magical alphabets

Theban is a cipher: each of its letters stands for a letter of the Latin alphabet. Give a Theban inscription’s plaintext, in the language it encodes, and treat it as that text. Malachim, Celestial, and Passing the River are three distinct alphabets, printed together with Theban in book 3 of Agrippa’s De Occulta Philosophia (1533); name the one meant. A modern redrawing or regularization of any of these alphabets is credited to its maker and is not presented as the historical form.

4.41 Sigils, seals, and other figures

A sigil, seal, lamen, magic square, character of a planetary intelligence or spirit, or other magical figure is an image, never text. Reproduce it as an image, with a caption naming the figure and the edition or manuscript it is taken from. Alt text says what the figure is and where it comes from; it does not trace the drawing line by line. Never let a Unicode character that resembles a figure stand in for it. Where a figure is described rather than reproduced, say so and cite a reproduction the reader can consult. Reproducing a seal or an item of insignia may require permission (see 2.19).

4.42 Transcription conventions

Transcribe a manuscript, typescript, or damaged document by the Leiden conventions, using the marks in table 4.5, and name the conventions at the head of the transcription or in a note. Within a transcription these marks take the place of ellipses and the 2-em dash for lost or illegible text. Use each mark only for its own purpose and only where the document warrants it, and invent no others. An editorial comment goes in a note, never in square brackets. In running text, mark the end of a line of the original with a vertical bar (|).

Table 4.5. Transcription marks

4.43 Diplomatic and normalized texts

A diplomatic transcription reproduces the document as it stands: its spelling, capitalization, punctuation, line division, abbreviations, deletions, and insertions. A normalized transcription regularizes these for reading. Use a diplomatic transcription wherever the argument depends on the document’s form, such as a deletion, a hand, a spelling, or the place of a word on the page. Use a normalized transcription where a passage is quoted for its content. State which at the head of the transcription, and for a normalized transcription name the kinds of change made. Do not mix the two in one transcription.

4.44 Deletions and insertions

Say what kind of deletion the document shows: struck through, overwritten, erased, or blotted out. Say where an insertion stands (above or below the line, in the margin, or written over other letters) and, where it can be told, whose hand made it and when. In a diplomatic transcription, record an insertion where it stands.

4.45 Hands and later annotations

Where more than one hand appears in a document, identify each, and state the basis for the identification: comparison with signed or dated documents, the ink, the writer’s own statement. Where hands cannot be told apart, say so. Treat a writer’s later annotations on their own manuscript as a separate layer, dated where possible, and never merge them silently with the original text.

4.46 Folios, pages, and lines

Cite a foliated manuscript by leaf and side (see 8.15) and a paginated notebook or typescript by page. Never write r or v standing alone, or immediately after a verse number, where it could be read as part of a reference to a received text. Cite the manuscript leaves of a received text by the numbering of the facsimile used (see 6.21, 6.25). The full citation of an archival document follows 5.43.

4.47 Long s and early orthography

A diplomatic transcription keeps the source’s u and v, i and j, long s (ſ, U+017F), ligatures, and abbreviations. A normalized transcription, or a quotation from an early printed book, writes long s as s and keeps the source’s u and v and i and j unless the work states once that it modernizes them. Keep the letters æ and œ; do not reproduce typographic ligatures such as fi and fl. In a normalized transcription, expand abbreviations within parentheses (see table 4.5); in a quotation, keep them and explain any that would puzzle a reader (see 8.12). Never write f for long s. Titles keep their letterforms (see 3.47).

4.48 Received texts in manuscript

Transcribe the manuscript of Liber AL (ALM), and the manuscripts of the other received texts, diplomatically wherever their form is in question. A received text is never normalized (see 3.70). A drawing or mark on a manuscript page is reproduced as an image and cited by its leaf (see 4.41); it is not described as though it were part of the text. The relation between a received text’s manuscript and its printed text, and their variants, are treated in 6.21–6.26.

4.49 Claims about a manuscript

A claim about a document’s physical form (its hand, a deletion or insertion, its layout, its damage, or the reading of an unclear letter) cites a facsimile or photograph in which a reader can check it, by page or plate. Alternatively, it states that the writer examined the original and gives the repository and shelfmark. A transcription alone does not support such a claim. Where a reading depends on machine recognition of the handwriting, say so. Do not call a transcription an edition: a transcription records what one document says, while an edition establishes a text from its witnesses and records their variants.

4.50 Citing another’s transcription

When a writer knows a document only through someone else’s transcription, cite the transcription, name the transcriber, and do not cite the manuscript as though it had been examined. Give the manuscript’s own identification as well where it is known. Where two transcriptions of a document disagree, report both readings and name both transcribers. Where the writer has checked a transcription against a facsimile, say so. Much of Dee’s record reaches the field through Casaubon’s printed text of 1659; cite it as Casaubon’s text.

4.51 Common special characters

Table 4.6 lists the special characters the field relies on, with their code points and the characters most often typed in their place. Enter each from its code point or a reliable character map, never by approximating it with a look-alike, and search for the wrong forms before a manuscript goes to production (see 2.5).

Table 4.6. Special characters

4.52 Apostrophes and half rings

Use the right single quotation mark (’, U+2019) as the apostrophe. In full romanization, write alef and hamza with the modifier letter ʾ (U+02BE) and ayin with ʿ (U+02BF). Simplified romanization may keep them instead as the apostrophe for alef and hamza and the opening single quotation mark (‘, U+2018) for ayin; record the choice on the style sheet and keep it throughout. Never use the typewriter apostrophe (') for any of them. Automatic quotation marks choose between ‘ and ’ by a mark’s position, not its letter: turn them off while typing romanized Hebrew or Arabic, or check every such mark afterward.

4.53 Astrological and planetary glyphs

Table 4.7 gives the planetary and zodiacal glyphs. Use them in date lines, tables, charts, and quoted matter; in running prose, write the names (see 3.65). For the Moon, use ☽ (U+263D); use ☾ (U+263E) only where the waning Moon is meant. Never use an emoji in place of a glyph, in particular ☀︎ (U+2600 BLACK SUN WITH RAYS) for the Sun or 🌙 (U+1F319 CRESCENT MOON) for the Moon.

Table 4.7. Astrological glyphs

4.54 Emoji forms and text equivalents

Follow each astrological glyph that has an emoji form, the twelve zodiacal glyphs and ♀︎ and ♂︎, with VARIATION SELECTOR-15 (U+FE0E), which requests the text form in place of color emoji. In web output, give every astrological glyph a text equivalent that a screen reader can announce, such as the planet’s or sign’s name supplied as an accessible label.

4.55 Grade notation

Proof grade notation in a font known to carry the white square (□, U+25A1), never by eye alone: a correct square looks like the empty box a font shows for a missing character (see 4.60). Raise the square by styling it as a superscript (see 3.31), never by substituting U+2E0B RAISED SQUARE, a text-critical sign that few fonts carry. The form and use of the notation are treated in 3.31–3.34.

4.56 Polytonic Greek: tonos and oxia

Unicode has two sets of precomposed Greek vowels with the acute accent: the tonos forms (U+03AC, U+03AD, U+03AE, U+03AF, U+03CC, U+03CD, U+03CE) and the oxia forms of the Greek Extended block (U+1F71, U+1F73, U+1F75, U+1F77, U+1F79, U+1F7B, U+1F7D). The two sets are canonically equivalent, and normalization (see 4.57) replaces each oxia form with the corresponding tonos form. Enter the tonos forms. Choose a Greek font that draws the tonos vowels with the slanted acute of polytonic Greek.

4.57 Unicode normalization

Normalize every text to Unicode Normalization Form C (NFC) once, at a fixed point in production. NFC joins a base letter and a combining mark into the single precomposed character where one exists, so that h followed by a combining dot below becomes ḥ (U+1E25), and the same word is stored the same way wherever it appears.

Several of its changes look like errors and are not: oxia vowels become tonos vowels (see 4.56), the keraia (U+0374) becomes U+02B9, the ohm sign (U+2126) becomes the Greek capital omega (U+03A9), and Hebrew presentation forms with points, such as U+FB2A, become the letter and its point. The deprecated angle brackets U+2329 and U+232A become the CJK brackets U+3008 and U+3009; use U+27E8 and U+27E9 for editorial angle brackets (see table 4.5). NFC does not separate typographic ligatures pasted from a PDF (U+FB01, U+FB02); replace them with separate letters. Do not use Form KC (NFKC), which also changes characters a transcription needs, among them the long s.

4.58 Right-to-left text

Hebrew and Arabic run right to left, and in a line of English a run of either can carry neighboring punctuation and numbers with it, so that a comma or a number appears on the wrong side. In print, proof every line that mixes directions. In web output, put each right-to-left run in its own element marked dir="rtl", or in a bdi element, and leave surrounding punctuation outside it; in a table, set the direction of each cell that holds Hebrew or Arabic. In plain text without markup, enclose the run between FIRST STRONG ISOLATE (U+2068) and POP DIRECTIONAL ISOLATE (U+2069); use the older marks U+200E and U+200F only where isolates are unsupported. For non-Latin entries in an index, see 7.23.

4.59 Language tagging in web output

In web output, mark every passage in another language with its language tag. The tags this field needs most are he (Hebrew), grc (ancient Greek), el (modern Greek), cop (Coptic), egy (Egyptian), ar (Arabic), sa (Sanskrit), pi (Pali), zh (Chinese), akk (Akkadian), sux (Sumerian), and la (Latin). A romanized passage may carry its language with the script subtag Latn.

4.60 Fonts

Choose fonts that carry every character the work uses, and test them before production by setting the work’s list of special characters (see 2.2) in each face; a missing character appears as an empty box or in a substituted face. Name a primary face and fallbacks for Hebrew, polytonic Greek, Coptic, Egyptian transliteration, hieroglyphs, and symbols, and in web output serve the fonts with the page rather than relying on the reader’s system. Faces with wide coverage include the Noto families, which have separate faces for Hebrew, Coptic, hieroglyphs, and symbols; Gentium Plus for Latin and Greek; New Athena Unicode for polytonic Greek and transliteration; and Ezra SIL for pointed Hebrew. Confirm that each face’s license permits commercial publication and web embedding. Never use a symbol font to produce a character that Unicode encodes. For fonts in a submitted manuscript, see 2.5.