The TrueType OS/2 table

The OS/2 table, named after IBM’s operating system where it began, holds the metrics and classifications Windows uses: weight and width, embedding permissions, line spacing, which Unicode ranges and code pages the font covers, and style bits. OpenType requires it; Apple’s specification makes it optional. There are six versions, 0 to 5, and each adds fields at the end, so a reader must check the version and the table’s length before reading the later fields. Version 4 or later is strongly recommended.

Fields

Offset	Type	Name	From version	Description
`0`	`uint16`	`version`	0	0 to 5.
`2`	`FWORD`	`xAvgCharWidth`	0	The average advance width of all glyphs with non-zero width. Versions 0 to 2 used a weighted average of the lower-case letters and space instead (a 64, b 14, c 27, d 35, e 100, f 20, g 14, h 42, i 63, j 3, k 6, l 35, m 20, n 56, o 56, p 17, q 4, r 49, s 56, t 71, u 31, v 10, w 18, x 3, y 18, z 2, space 166, divided by 1,000), and many fonts still carry that value. Not reliable for layout.
`4`	`uint16`	`usWeightClass`	0	Visual weight, 1 to 1,000: see weight classes.
`6`	`uint16`	`usWidthClass`	0	1 to 9: see width classes.
`8`	`uint16`	`fsType`	0	Embedding permissions: see fsType.
`10`	`FWORD`	`ySubscriptXSize`	0	Recommended horizontal size of subscripts, in design units: 205 in a 2,048-unit em means a tenth of the em.
`12`	`FWORD`	`ySubscriptYSize`	0	Recommended vertical size of subscripts.
`14`	`FWORD`	`ySubscriptXOffset`	0	Horizontal offset of subscripts; usually 0 except in slanted fonts.
`16`	`FWORD`	`ySubscriptYOffset`	0	Vertical offset of subscripts, positive downwards from the baseline.
`18`	`FWORD`	`ySuperscriptXSize`	0	Recommended horizontal size of superscripts.
`20`	`FWORD`	`ySuperscriptYSize`	0	Recommended vertical size of superscripts.
`22`	`FWORD`	`ySuperscriptXOffset`	0	Horizontal offset of superscripts.
`24`	`FWORD`	`ySuperscriptYOffset`	0	Vertical offset of superscripts, positive upwards.
`26`	`FWORD`	`yStrikeoutSize`	0	Thickness of the strikeout stroke, normally that of the em dash and of the underline in `post`.
`28`	`FWORD`	`yStrikeoutPosition`	0	Position of the top of the strikeout stroke above the baseline.
`30`	`int16`	`sFamilyClass`	0	High byte: IBM family class; low byte: subclass.
`32`	`uint8` × 10	`panose`	0	The PANOSE classification.
`42`	`uint32` × 4	`ulUnicodeRange1` to `4`	0 (bits defined from 1)	128 bits, one per group of Unicode blocks the font supports.
`58`	`Tag`	`achVendID`	0	The vendor’s four-character ID, from Microsoft’s registry, or blank (zeros or four spaces). By convention only registered IDs are all upper case.
`62`	`uint16`	`fsSelection`	0	Style bits: see fsSelection.
`64`	`uint16`	`usFirstCharIndex`	0	The lowest code in the Windows `cmap` subtable (3, 0 or 1), or `0xFFFF` if it is above U+FFFF.
`66`	`uint16`	`usLastCharIndex`	0	The highest code, or `0xFFFF` if the font has characters above U+FFFF.
`68`	`FWORD`	`sTypoAscender`	0	Typographic ascender, for portable line spacing. In CJK fonts for vertical text, the top of the ideographic em box.
`70`	`FWORD`	`sTypoDescender`	0	Typographic descender, usually negative.
`72`	`FWORD`	`sTypoLineGap`	0	Typographic line gap. Default line spacing is ascender − descender + line gap.
`74`	`UFWORD`	`usWinAscent`	0	Height above the baseline of the clipping region: Windows GDI clips anything above it, so it should be at least `yMax`.
`76`	`UFWORD`	`usWinDescent`	0	Depth below the baseline of the clipping region, as a positive number; at least −`yMin`. A `VDMX` table overrides both at the sizes it covers.
`78`	`uint32` × 2	`ulCodePageRange1`, `2`	1	64 bits, one per code page the font supports.
`86`	`FWORD`	`sxHeight`	2	The x-height: by default the top of the glyph for U+0078 x, or 0 if there is none.
`88`	`FWORD`	`sCapHeight`	2	The capital height: by default the top of U+0048 H, or 0.
`90`	`uint16`	`usDefaultChar`	2	The character to show for missing ones; 0 means glyph 0. Cannot name characters above U+FFFF; discouraged.
`92`	`uint16`	`usBreakChar`	2	The word break character, usually U+0020. Discouraged for the same reason.
`94`	`uint16`	`usMaxContext`	2	The longest glyph context any layout feature uses: 2 for pair kerning, 3 for an “ffi” ligature.
`96`	`uint16`	`usLowerOpticalPointSize`	5	The smallest size, in twips (twentieths of a point), this optical-size design is for, inclusive; 0 if the font has no optical sizes. Superseded by `STAT`.
`98`	`uint16`	`usUpperOpticalPointSize`	5	The size it is for up to, exclusive; `0xFFFF` means no limit, and is the value for fonts without optical sizes.

Version	Defined in	Length	Notes
0	TrueType 1.5	78 bytes	Apple’s documentation stops at `usLastCharIndex`, so some old fonts have a 68-byte version 0 table.
1	TrueType 1.66	86 bytes	Adds the code page ranges.
2	OpenType 1.1	96 bytes	Adds x-height, cap height, default and break characters, maximum context.
3	OpenType 1.4	96 bytes	Same fields; Unicode range bits for Unicode 3.2; embedding bits made exclusive.
4	OpenType 1.5	96 bytes	Same fields; `fsSelection` bits 7 to 9; Unicode range bits for Unicode 5.1.
5	OpenType 1.7	100 bytes	Adds the optical point size range.

Weight classes

Any value from 1 to 1,000 is valid, on the same scale as the wght variation axis. The usual values, with their windows.h names:

Value	Weight	C name
100	Thin	`FW_THIN`
200	Extra-light (Ultra-light)	`FW_EXTRALIGHT`
300	Light	`FW_LIGHT`
400	Normal (Regular)	`FW_NORMAL`
500	Medium	`FW_MEDIUM`
600	Semi-bold (Demi-bold)	`FW_SEMIBOLD`
700	Bold	`FW_BOLD`
800	Extra-bold (Ultra-bold)	`FW_EXTRABOLD`
900	Black (Heavy)	`FW_BLACK`

Width classes

Value	Width	C name	% of normal (`wdth`)
1	Ultra-condensed	`FWIDTH_ULTRA_CONDENSED`	50
2	Extra-condensed	`FWIDTH_EXTRA_CONDENSED`	62.5
3	Condensed	`FWIDTH_CONDENSED`	75
4	Semi-condensed	`FWIDTH_SEMI_CONDENSED`	87.5
5	Medium (normal)	`FWIDTH_NORMAL`	100
6	Semi-expanded	`FWIDTH_SEMI_EXPANDED`	112.5
7	Expanded	`FWIDTH_EXPANDED`	125
8	Extra-expanded	`FWIDTH_EXTRA_EXPANDED`	150
9	Ultra-expanded	`FWIDTH_ULTRA_EXPANDED`	200

fsType: embedding permissions

Bits	Mask	Meaning
0 to 3	`0x000F`	Usage permissions, one of the values below. Bit 0 is permanently reserved and must be 0.
4 to 7	`0x00F0`	Reserved; 0.
8	`0x0100`	No subsetting: the font must not be subset before it is embedded.
9	`0x0200`	Bitmap embedding only: only the font’s bitmaps may be embedded, no outlines; a font with no bitmaps cannot be embedded.
10 to 15	`0xFC00`	Reserved; 0.

Value of bits 0 to 3	Permission
0	Installable embedding: the font may be embedded and installed permanently on the remote system, whose user gets the same rights as the purchaser.
2	Restricted licence embedding: the font must not be modified, embedded or exchanged without the owner’s explicit permission.
4	Preview and print embedding: the font may be embedded and loaded temporarily to view or print the document, which must be opened read-only.
8	Editable embedding: as preview and print, but the document may be edited and saved.

From version 3 the four values are exclusive. Versions 0 to 2 allowed several bits at once, the least restrictive winning, and some fonts set both 4 and 8. Versions 0 and 1 define only bits 0 to 3. Apple’s documentation gives a meaning to bit 1 alone, from a draft of the table, and some early fonts wrongly used the value 1. Software that loads a font embedded for preview or editing must delete it when the document is closed.

fsSelection

Bit	Mask	Name	Meaning
0	`0x0001`	`ITALIC`	Italic or oblique glyphs. Must match `macStyle` bit 1 in `head`.
1	`0x0002`	`UNDERSCORE`	Glyphs are underscored.
2	`0x0004`	`NEGATIVE`	Foreground and background reversed.
3	`0x0008`	`OUTLINED`	Hollow outline glyphs.
4	`0x0010`	`STRIKEOUT`	Glyphs are struck through.
5	`0x0020`	`BOLD`	Bold. Must match `macStyle` bit 0.
6	`0x0040`	`REGULAR`	The regular style of the family; bits 0 and 5 must then be clear.
7	`0x0080`	`USE_TYPO_METRICS`	Use `sTypoAscender` − `sTypoDescender` + `sTypoLineGap` for line spacing, not the `usWin` values. Version 4 and later.
8	`0x0100`	`WWS`	The family and subfamily names already fit the weight, width and slope model, so name IDs 21 and 22 are not needed. Version 4 and later.
9	`0x0200`	`OBLIQUE`	An oblique rather than a true italic design, for software such as CSS that tells them apart. Version 4 and later.
10 to 15	`0xFC00`	reserved	0. Bit 15 has been used by some legacy software; its use is deprecated.

In versions 0 to 3 only bits 0 to 6 are defined, and readers should ignore the rest. Bits 1 to 4 are rarely used and mark decorative or special-purpose fonts.

IBM family classes

The high byte of sFamilyClass is the class and the low byte the subclass, as IBM defined them. In every class, subclass 0 is “no classification” and 15 “miscellaneous”; the subclasses not listed are reserved. Classes 6, 11, 13 and 14 are reserved.

Class	Name	Subclasses
0	No classification	–
1	Oldstyle serifs	1 IBM rounded legibility, 2 Garalde, 3 Venetian, 4 modified Venetian, 5 Dutch modern, 6 Dutch traditional, 7 contemporary, 8 calligraphic
2	Transitional serifs	1 direct line, 2 script
3	Modern serifs	1 Italian, 2 script
4	Clarendon serifs	1 Clarendon, 2 modern, 3 traditional, 4 newspaper, 5 stub serif, 6 monotone, 7 typewriter
5	Slab serifs	1 monotone, 2 humanist, 3 geometric, 4 Swiss, 5 typewriter
7	Freeform serifs	1 modern
8	Sans serif	1 IBM neo-grotesque gothic, 2 humanist, 3 low-x round geometric, 4 high-x round geometric, 5 neo-grotesque gothic, 6 modified neo-grotesque gothic, 9 typewriter gothic, 10 matrix
9	Ornamentals	1 engraver, 2 black letter, 3 decorative, 4 three-dimensional
10	Scripts	1 uncial, 2 brush joined, 3 formal joined, 4 monotone joined, 5 calligraphic, 6 brush unjoined, 7 formal unjoined, 8 monotone unjoined
12	Symbolic	3 mixed serif, 6 oldstyle serif, 7 neo-grotesque sans serif

For example, 0x0805 is a neo-grotesque gothic sans serif, the class of Helvetica.

PANOSE

Ten bytes that describe how the typeface looks, so that software can find a similar font. The first byte, the family kind, decides what the other nine mean. For Latin text faces (family kind 2) they are, in order:

Byte	Name	Byte	Name
0	`bFamilyType`	5	`bStrokeVariation`
1	`bSerifStyle`	6	`bArmStyle`
2	`bWeight`	7	`bLetterform`
3	`bProportion`	8	`bMidline`
4	`bContrast`	9	`bXHeight`

Family kind 3, Latin hand written, makes byte 1 the tool kind instead, and so on; some software uses family kind 5, Latin symbol, to spot symbol fonts, and the proportion byte to spot monospaced ones. The values of each byte are defined in the PANOSE Classification Metrics Guide, which OpenType names as the normative source and no longer repeats; this site does not list them yet. A variable font can carry only one PANOSE value, normally that of its default instance. All zeros means “any”, no classification given.

Unicode range bits

Bit n is in ulUnicodeRange1 for 0 to 31, ulUnicodeRange2 for 32 to 63 and so on, counting from the least significant bit of each uint32. A set bit says the font is functional for those ranges, as judged by its designer. The ranges ran out with Unicode 5.1; the dlng and slng entries of the meta table are the newer way to declare coverage.

Bit	Ranges
0	Basic Latin 0000–007F
1	Latin-1 Supplement 0080–00FF
2	Latin Extended-A 0100–017F
3	Latin Extended-B 0180–024F
4	IPA Extensions 0250–02AF; Phonetic Extensions 1D00–1D7F; Phonetic Extensions Supplement 1D80–1DBF
5	Spacing Modifier Letters 02B0–02FF; Modifier Tone Letters A700–A71F
6	Combining Diacritical Marks 0300–036F; Combining Diacritical Marks Supplement 1DC0–1DFF
7	Greek and Coptic 0370–03FF
8	Coptic 2C80–2CFF (versions 1 and 2: “Greek Symbols and Coptic”)
9	Cyrillic 0400–04FF; Cyrillic Supplement 0500–052F; Cyrillic Extended-A 2DE0–2DFF; Cyrillic Extended-B A640–A69F
10	Armenian 0530–058F
11	Hebrew 0590–05FF
12	Vai A500–A63F (versions 1 and 2: “Hebrew Extended”)
13	Arabic 0600–06FF; Arabic Supplement 0750–077F
14	NKo 07C0–07FF (versions 1 and 2: “Arabic Extended”)
15	Devanagari 0900–097F
16	Bangla 0980–09FF
17	Gurmukhi 0A00–0A7F
18	Gujarati 0A80–0AFF
19	Odia 0B00–0B7F
20	Tamil 0B80–0BFF
21	Telugu 0C00–0C7F
22	Kannada 0C80–0CFF
23	Malayalam 0D00–0D7F
24	Thai 0E00–0E7F
25	Lao 0E80–0EFF
26	Georgian 10A0–10FF; Georgian Supplement 2D00–2D2F
27	Balinese 1B00–1B7F (versions 1 and 2: “Georgian Extended”)
28	Hangul Jamo 1100–11FF
29	Latin Extended Additional 1E00–1EFF; Latin Extended-C 2C60–2C7F; Latin Extended-D A720–A7FF
30	Greek Extended 1F00–1FFF
31	General Punctuation 2000–206F; Supplemental Punctuation 2E00–2E7F
32	Superscripts And Subscripts 2070–209F
33	Currency Symbols 20A0–20CF
34	Combining Diacritical Marks For Symbols 20D0–20FF
35	Letterlike Symbols 2100–214F
36	Number Forms 2150–218F
37	Arrows 2190–21FF; Supplemental Arrows-A 27F0–27FF; Supplemental Arrows-B 2900–297F; Miscellaneous Symbols and Arrows 2B00–2BFF
38	Mathematical Operators 2200–22FF; Supplemental Mathematical Operators 2A00–2AFF; Miscellaneous Mathematical Symbols-A 27C0–27EF; Miscellaneous Mathematical Symbols-B 2980–29FF
39	Miscellaneous Technical 2300–23FF
40	Control Pictures 2400–243F
41	Optical Character Recognition 2440–245F
42	Enclosed Alphanumerics 2460–24FF
43	Box Drawing 2500–257F
44	Block Elements 2580–259F
45	Geometric Shapes 25A0–25FF
46	Miscellaneous Symbols 2600–26FF
47	Dingbats 2700–27BF
48	CJK Symbols And Punctuation 3000–303F
49	Hiragana 3040–309F
50	Katakana 30A0–30FF; Katakana Phonetic Extensions 31F0–31FF
51	Bopomofo 3100–312F; Bopomofo Extended 31A0–31BF
52	Hangul Compatibility Jamo 3130–318F
53	Phags-pa A840–A87F (versions 1 and 2: “CJK Miscellaneous”)
54	Enclosed CJK Letters And Months 3200–32FF
55	CJK Compatibility 3300–33FF
56	Hangul Syllables AC00–D7AF
57	Non-Plane 0: at least one character above U+FFFF
58	Phoenician 10900–1091F
59	CJK Unified Ideographs 4E00–9FFF; CJK Radicals Supplement 2E80–2EFF; Kangxi Radicals 2F00–2FDF; Ideographic Description Characters 2FF0–2FFF; CJK Unified Ideographs Extension A 3400–4DBF; Extension B 20000–2A6DF; Kanbun 3190–319F
60	Private Use Area (plane 0) E000–F8FF
61	CJK Strokes 31C0–31EF; CJK Compatibility Ideographs F900–FAFF; CJK Compatibility Ideographs Supplement 2F800–2FA1F
62	Alphabetic Presentation Forms FB00–FB4F
63	Arabic Presentation Forms-A FB50–FDFF
64	Combining Half Marks FE20–FE2F
65	Vertical Forms FE10–FE1F; CJK Compatibility Forms FE30–FE4F
66	Small Form Variants FE50–FE6F
67	Arabic Presentation Forms-B FE70–FEFF
68	Halfwidth And Fullwidth Forms FF00–FFEF
69	Specials FFF0–FFFF
70	Tibetan 0F00–0FFF
71	Syriac 0700–074F
72	Thaana 0780–07BF
73	Sinhala 0D80–0DFF
74	Myanmar 1000–109F
75	Ethiopic 1200–137F; Ethiopic Supplement 1380–139F; Ethiopic Extended 2D80–2DDF
76	Cherokee 13A0–13FF
77	Unified Canadian Aboriginal Syllabics 1400–167F
78	Ogham 1680–169F
79	Runic 16A0–16FF
80	Khmer 1780–17FF; Khmer Symbols 19E0–19FF
81	Mongolian 1800–18AF
82	Braille Patterns 2800–28FF
83	Yi Syllables A000–A48F; Yi Radicals A490–A4CF
84	Tagalog 1700–171F; Hanunoo 1720–173F; Buhid 1740–175F; Tagbanwa 1760–177F
85	Old Italic 10300–1032F
86	Gothic 10330–1034F
87	Deseret 10400–1044F
88	Byzantine Musical Symbols 1D000–1D0FF; Musical Symbols 1D100–1D1FF; Ancient Greek Musical Notation 1D200–1D24F
89	Mathematical Alphanumeric Symbols 1D400–1D7FF
90	Private Use (plane 15) F0000–FFFFD; Private Use (plane 16) 100000–10FFFD
91	Variation Selectors FE00–FE0F; Variation Selectors Supplement E0100–E01EF
92	Tags E0000–E007F
93	Limbu 1900–194F
94	Tai Le 1950–197F
95	New Tai Lue 1980–19DF
96	Buginese 1A00–1A1F
97	Glagolitic 2C00–2C5F
98	Tifinagh 2D30–2D7F
99	Yijing Hexagram Symbols 4DC0–4DFF
100	Syloti Nagri A800–A82F
101	Linear B Syllabary 10000–1007F; Linear B Ideograms 10080–100FF; Aegean Numbers 10100–1013F
102	Ancient Greek Numbers 10140–1018F
103	Ugaritic 10380–1039F
104	Old Persian 103A0–103DF
105	Shavian 10450–1047F
106	Osmanya 10480–104AF
107	Cypriot Syllabary 10800–1083F
108	Kharoshthi 10A00–10A5F
109	Tai Xuan Jing Symbols 1D300–1D35F
110	Cuneiform 12000–123FF; Cuneiform Numbers and Punctuation 12400–1247F
111	Counting Rod Numerals 1D360–1D37F
112	Sundanese 1B80–1BBF
113	Lepcha 1C00–1C4F
114	Ol Chiki 1C50–1C7F
115	Saurashtra A880–A8DF
116	Kayah Li A900–A92F
117	Rejang A930–A95F
118	Cham AA00–AA5F
119	Ancient Symbols 10190–101CF
120	Phaistos Disc 101D0–101FF
121	Carian 102A0–102DF; Lycian 10280–1029F; Lydian 10920–1093F
122	Domino Tiles 1F030–1F09F; Mahjong Tiles 1F000–1F02F
123 to 127	Reserved; 0

Version 1 defined bits 0 to 69 only, version 2 (from OpenType 1.3) up to 83, version 3 up to 92, and version 4 bit 58 and 93 to 122. The old meanings given in brackets were dropped in version 2 (bit 53 in version 3) and the bits reassigned in version 4; in a version 1 or 2 table the old meaning may be the one intended.

Code page bits

Bits 0 to 31 are in ulCodePageRange1, 32 to 63 in ulCodePageRange2. If the symbol bit (31) is set and the font has a Windows Unicode cmap subtable, the characters U+F000 to U+F0FF form the symbol character set.

Bit	Code page	Description	Bit	Code page	Description
0	1252	Latin 1	32 to 47	–	Reserved for OEM
1	1250	Latin 2: Eastern Europe	48	869	IBM Greek
2	1251	Cyrillic	49	866	MS-DOS Russian
3	1253	Greek	50	865	MS-DOS Nordic
4	1254	Turkish	51	864	Arabic
5	1255	Hebrew	52	863	MS-DOS Canadian French
6	1256	Arabic	53	862	Hebrew
7	1257	Windows Baltic	54	861	MS-DOS Icelandic
8	1258	Vietnamese (from version 2)	55	860	MS-DOS Portuguese
9 to 15	–	Reserved for alternate ANSI	56	857	IBM Turkish
16	874	Thai	57	855	IBM Cyrillic; mainly Russian
17	932	JIS/Japan	58	852	Latin 2
18	936	Chinese: Simplified (PRC and Singapore)	59	775	MS-DOS Baltic
19	949	Korean Wansung	60	737	Greek; former 437 G
20	950	Chinese: Traditional (Taiwan and Hong Kong SAR)	61	708	Arabic; ASMO 708
21	1361	Korean Johab	62	850	WE/Latin 1
22 to 28	–	Reserved for alternate ANSI or OEM	63	437	US
29	–	Macintosh character set (US Roman)
30	–	OEM character set
31	–	Symbol character set

In variable fonts

A variable font should set USE_TYPO_METRICS and make the hhea line metrics equal to the sTypo ones. Metrics that change with the design are varied through the MVAR table, keyed by these tags: cpht sCapHeight, hasc sTypoAscender, hdsc sTypoDescender, hlgp sTypoLineGap, xhgt sxHeight, hcla usWinAscent, hcld usWinDescent, stro yStrikeoutPosition, strs yStrikeoutSize, sbxo ySubscriptXOffset, sbxs ySubscriptXSize, sbyo ySubscriptYOffset, sbys ySubscriptYSize, spxo ySuperscriptXOffset, spxs ySuperscriptXSize, spyo ySuperscriptYOffset, spys ySuperscriptYSize. Weight, width and optical size come from the wght, wdth and opsz axes instead; a wdth above 200 gives width class 9.

Example

The 96-byte version 2 OS/2 table of KaTeX’s Size2 Regular font:

00000000  00 02 03 73 01 90 00 05  00 00 02 8A 02 BC 00 00
00000010  00 8C 02 8A 02 BC 00 00  01 E0 00 31 01 02 00 00
00000020  00 00 00 00 00 00 00 00  00 00 80 00 00 EF 10 00
00000030  EC ED 00 00 00 00 00 00  00 00 50 66 45 64 00 40
00000040  00 20 2A 06 03 20 FF 38  00 5A 05 50 03 5E 20 00
00000050  00 8F 5E 03 00 00 00 00  00 00 00 00 00 20 00 01

Version 2; average width 883; weight 400, normal; width 5, normal; fsType 0, installable embedding.
Subscripts 650 × 700, 140 below the baseline; superscripts 650 × 700, 480 above; strikeout 49 thick at 258.
No family class and PANOSE all zeros: not classified.
Unicode ranges 800000EF 1000ECED 0 0: bits 0 to 3, 5 to 7 and 31 (Latin, modifier letters, combining marks, Greek, punctuation), and 32, 34, 35, 37 to 39, 42, 43, 45 to 47 and 60 (sub- and superscripts, symbols, arrows, mathematical operators, technical, box drawing, geometric shapes, dingbats, private use).
Vendor PfEd, the ID FontForge writes by default; fsSelection 0x0040, regular.
Characters U+0020 to U+2A06. Typographic ascender 800, descender −200, line gap 90; Windows ascent 1,360 and descent 862, matching yMax and −yMin in head.
Code pages 2000008F 5E030000: 1252, 1250, 1251, 1253, 1257 and Macintosh (bits 0 to 3, 7, 29), and 869, 866, 855, 852, 775, 737 and 850 (bits 48, 49, 57 to 60, 62).
x-height and cap height 0, default character 0, break character U+0020, maximum context 1.

.TTF File Format – OS/2