The TrueType name table

The name table holds the font’s text: its family and style names, full name, PostScript name, version, copyright, licence and so on. Each string is stored once per platform, encoding and language, so a font can carry the same name for Mac and Windows, and in several languages. Software finds a string by its name ID and the platform, encoding and language it wants.

Header

Type	Name	Description
`uint16`	`version`	0 or 1.
`uint16`	`count`	The number of name records.
`Offset16`	`storageOffset`	Where the string storage starts, from the start of the table.
`NameRecord` × `count`	`nameRecord`	The name records.
`uint16`	`langTagCount`	Version 1 only: the number of language-tag records.
`LangTagRecord` × `langTagCount`	`langTagRecord`	Version 1 only: the language-tag records.
bytes	storage	The strings themselves.

Name record

Offset	Type	Name	Description
`0`	`uint16`	`platformID`	The platform.
`2`	`uint16`	`encodingID`	The platform’s encoding of the string.
`4`	`uint16`	`languageID`	The language: a platform’s own ID below `0x8000`, or a language tag from `0x8000` up.
`6`	`uint16`	`nameID`	What the string is: see name IDs.
`8`	`uint16`	`length`	The string’s length in bytes.
`10`	`Offset16`	`stringOffset`	Where the string starts, from the start of the storage.

Records are sorted by platform ID, then encoding ID, then language ID, then name ID. Strings are not terminated, and two records may share the same bytes.

Language-tag records (version 1)

Version 1 lets a font name a language with a BCP 47 tag, such as en, fr-CA or zh-Hant, instead of a platform’s number. Each record is a uint16 length in bytes and an Offset16 langTagOffset from the start of the storage; the tag is in UTF-16BE. The first record is language ID 0x8000, the second 0x8001, and so on. A name record whose language ID is beyond the last tag has an unknown language and should not be used. Version 0 tables may use only IDs below 0x8000, except on user-defined platforms 240 to 255.

Name IDs

ID	Meaning
0	Copyright notice.
1	Font family name. Shared by at most four fonts that differ only in weight or slope (regular, italic, bold, bold italic), as many applications expect. Other members of a large family put the difference in this name, such as “Arial Black”, and use IDs 16 and 17 for the full family.
2	Font subfamily name: Regular, Italic, Bold or Bold Italic, matching bits 0 and 5 of `OS/2` `fsSelection`.
3	Unique font identifier.
4	Full font name, usually IDs 1 and 2 together, or 16 and 17; Windows uses it.
5	Version string. Should start “Version ” followed by a number with a full stop, such as “Version 1.002”; each part is digits worth less than 65,535. Windows reads the font’s version from here, not from `head`.
6	PostScript name. At most 63 printable ASCII characters (33 to 126) except `[ ] ( ) { } < > / %`.
7	Trademark notice.
8	Manufacturer name.
9	Designer.
10	Description: history, features, usage notes.
11	URL of the vendor, with its protocol.
12	URL of the designer.
13	Licence description.
14	Licence information URL.
15	Reserved.
16	Typographic family name, with no limit on the number of members (formerly “preferred family”). When absent, ID 1 is the typographic family.
17	Typographic subfamily name, unique within the typographic family (formerly “preferred subfamily”). When absent, ID 2 serves.
18	Compatible full name, Macintosh only: the name for menus built from the FOND resource, when it should differ from ID 4.
19	Sample text.
20	PostScript CID findfont name. When present, ID 6 is the name to use with `composefont` and this one is the name for `findfont`. Same character rules as ID 6.
21	WWS family name: a family name that fits the weight, width and slope model, for fonts whose IDs 16 and 17 include other attributes such as optical size (“Minion Pro Caption”). Not needed when bit 8 of `fsSelection` is set.
22	WWS subfamily name, used with ID 21: only weight, width and slope (“Semibold Italic”).
23	Light background palette: used in `CPAL` palette labels to mark a palette for light backgrounds.
24	Dark background palette: marks a palette for dark backgrounds.
25	Variations PostScript name prefix: in a variable font, the family prefix for generating instance PostScript names; ASCII letters and digits only, the same in every record.
26 to 255	Reserved for future standard names.
256 to 32,767	Font-specific names, referred to from other tables: layout feature names, variation axes and instances, and so on.

Platforms and encodings

Platform ID	Platform	Encoding of the strings	Language IDs
0	Unicode	UTF-16BE. Encoding ID 3 (BMP) or 4 (full repertoire); 0, 1 and 2 are deprecated.	None defined; 0 means no particular language. Language tags allowed.
1	Macintosh	The Mac OS single- or double-byte encoding of the script in the encoding ID.	Macintosh language IDs.
3	Windows	UTF-16BE, except with encoding IDs 3, 4 and 5, which use code pages 936, 950 and 949. Unicode fonts use encoding 1, symbol fonts encoding 0; the IDs should match the font’s Windows `cmap` subtables (see Windows encodings).	Windows LCIDs.

Platforms 2 and 4 are defined only for cmap. Some old Traditional Chinese fonts, such as one in the Traditional Chinese edition of Windows 95, have Big5 records whose subfamily string is in UTF-16BE; Windows GDI accepts that.

Macintosh encoding IDs (Script Manager codes)

Used in both name and cmap.

ID	Script	ID	Script
0	Roman	17	Malayalam
1	Japanese	18	Sinhalese
2	Chinese (Traditional)	19	Burmese
3	Korean	20	Khmer
4	Arabic	21	Thai
5	Hebrew	22	Laotian
6	Greek	23	Georgian
7	Russian	24	Armenian
8	RSymbol	25	Chinese (Simplified)
9	Devanagari	26	Tibetan
10	Gurmukhi	27	Mongolian
11	Gujarati	28	Geez
12	Odia	29	Slavic
13	Bangla	30	Vietnamese
14	Tamil	31	Sindhi
15	Telugu	32	Uninterpreted
16	Kannada

Macintosh language IDs

From Apple’s TrueType Reference Manual. IDs 95 to 127 are not assigned. In a cmap subtable the language field is the ID plus 1.

ID	Language	ID	Language	ID	Language
0	English	51	Armenian	89	Swahili
1	French	52	Georgian	90	Kinyarwanda/Ruanda
2	German	53	Moldavian	91	Rundi
3	Italian	54	Kirghiz	92	Nyanja/Chewa
4	Dutch	55	Tajiki	93	Malagasy
5	Swedish	56	Turkmen	94	Esperanto
6	Spanish	57	Mongolian (Mongolian script)	128	Welsh
7	Danish	58	Mongolian (Cyrillic script)	129	Basque
8	Portuguese	59	Pashto	130	Catalan
9	Norwegian	60	Kurdish	131	Latin
10	Hebrew	61	Kashmiri	132	Quechua
11	Japanese	62	Sindhi	133	Guarani
12	Arabic	63	Tibetan	134	Aymara
13	Finnish	64	Nepali	135	Tatar
14	Greek	65	Sanskrit	136	Uighur
15	Icelandic	66	Marathi	137	Dzongkha
16	Maltese	67	Bengali	138	Javanese (Roman script)
17	Turkish	68	Assamese	139	Sundanese (Roman script)
18	Croatian	69	Gujarati	140	Galician
19	Chinese (Traditional)	70	Punjabi	141	Afrikaans
20	Urdu	71	Oriya	142	Breton
21	Hindi	72	Malayalam	143	Inuktitut
22	Thai	73	Kannada	144	Scottish Gaelic
23	Korean	74	Tamil	145	Manx Gaelic
24	Lithuanian	75	Telugu	146	Irish Gaelic (with dot above)
25	Polish	76	Sinhalese	147	Tongan
26	Hungarian	77	Burmese	148	Greek (polytonic)
27	Estonian	78	Khmer	149	Greenlandic
28	Latvian	79	Lao	150	Azerbaijani (Roman script)
29	Sami	80	Vietnamese
30	Faroese	81	Indonesian
31	Farsi/Persian	82	Tagalog
32	Russian	83	Malay (Roman script)
33	Chinese (Simplified)	84	Malay (Arabic script)
34	Flemish	85	Amharic
35	Irish Gaelic	86	Tigrinya
36	Albanian	87	Galla
37	Romanian	88	Somali
38	Czech
39	Slovak
40	Slovenian
41	Yiddish
42	Serbian
43	Macedonian
44	Bulgarian
45	Ukrainian
46	Byelorussian
47	Uzbek
48	Kazakh
49	Azerbaijani (Cyrillic script)
50	Azerbaijani (Arabic script)

Windows language IDs

Windows records use Windows language code identifiers (LCIDs). OpenType no longer lists them itself but points to Microsoft’s [MS-LCID] reference, which gives several hundred, each with its BCP 47 tag; this site does not repeat that list. In the 16-bit value, the low 10 bits are the primary language and the top 6 bits the sublanguage, usually a country. Common ones in fonts:

LCID	Language	LCID	Language
`0x0409`	English (United States)	`0x0411`	Japanese
`0x0809`	English (United Kingdom)	`0x0412`	Korean
`0x0407`	German (Germany)	`0x0804`	Chinese (PRC)
`0x040C`	French (France)	`0x0404`	Chinese (Taiwan)
`0x0C0A`	Spanish (Spain, modern sort)	`0x0419`	Russian
`0x0410`	Italian (Italy)	`0x0408`	Greek

Example

The name table of KaTeX’s Size2 Regular font: version 0, 9 records, storage at offset 114 (6 + 9 × 12). Every record is Windows, Unicode BMP, English (United States): platform 3, encoding 1, language 0x0409.

00000000  00 00 00 09 00 72                                header
00000006  00 03 00 01 04 09 00 00  00 A2 00 00             ID 0, 162 bytes at 0
00000012  00 03 00 01 04 09 00 01  00 16 00 A2             ID 1, 22 bytes at 162

Name ID	Length	Offset	String (UTF-16BE)
0	162	0	Copyright (c) 2009-2010 Design Science, Inc. …
1	22	162	KaTeX_Size2
2	14	184	Regular
3	70	198	(unique identifier)
4	38	268	KaTeX_Size2-Regular
5	66	306	Version 1.1; ttfautohint (v1.8.3)
6	38	268	KaTeX_Size2-Regular: the same bytes as ID 4
13	614	372	(licence)
14	52	986	(licence URL)

The records for IDs 4 and 6 point at the same 38 bytes, since the full name and PostScript name are the same here. “KaTeX_Size2” is stored as 00 4B 00 61 00 54 00 65 00 58 00 5F …; a Macintosh Roman record would hold it in single bytes, 4B 61 54 65 58 5F ….

.TTF File Format – name