T with Accent: Pick ț, ť or ṭ by Where Your Text Comes From

Which t with accent does each language or transliteration use?

Ask where the text comes from before you look at the mark. A t with accent is either a letter of a living alphabet or a tool for writing another script in Latin letters, and the two groups barely overlap. The jobs below come from Unicode's names list (opens in a new tab) and from Wiktionary's entries for ť (opens in a new tab), ț (opens in a new tab), ţ (opens in a new tab), ṭ (opens in a new tab), ṯ (opens in a new tab), ẗ (opens in a new tab), ṫ (opens in a new tab) and ṱ (opens in a new tab). The examples are names and places from AccentPad's Wikidata and GeoNames facts.

Your text	Letter	Mark	What it writes	Example from the facts
Romanian	ț	comma below	/t͡s/, the ts of cats	Rădăuți, Ionuț
Czech or Slovak	ť	caron	/c/, a soft t	Piešťany, Taťána
Gagauz or Kabyle	ţ	cedilla	/t͡s/
An old Romanian file or data set	ţ	cedilla	Romanian ț, in a nonstandard form	Constanţa, Galaţi
Arabic or Sanskrit in Latin letters	ṭ	dot below	Arabic ط, Sanskrit ट	ʿAbd al-Laṭīf, Fāṭima
Arabic or Hebrew in Latin letters	ṯ	line below	Arabic ث, Hebrew ת
Arabic in Latin letters	ẗ	diaeresis	Arabic ة
Irish in Gaelic type	ṫ	dot above	th in roman type
Venda or Herero	ṱ	circumflex below	/t̪/, a dental t

The Unicode CLDR data behind the languages card on this page holds ț in Romanian, ť in Czech and Slovak and ṭ in Kuvi, and no alphabet for the rest. Turkish is not in the table at all: it puts a cedilla on ç and ş, but its CLDR alphabet has no t with a mark.

Two neighbours tell the stories this page leaves out. The S with accent page explains how Romanian's comma letters split from the cedilla ones. The caron page explains why ť prints like a t followed by an apostrophe.

Why do Romanian place names in data still show ţ instead of ț?

Because for years the comma t had no code point. Wiktionary's ț entry says the letter was unified with ţ until Unicode 3.0 formally told them apart in 1999. AccentPad's facts agree: ţ has been in Unicode since 1.1, ț only since 3.0. Text typed before then could only use the cedilla.

AccentPad's own sources show the split today. We read it from the facts on 2026-10-10.

Source in AccentPad's facts	ţ, with cedilla	ț, with comma below
CLDR Romanian alphabet	auxiliary letters only	main alphabet
GeoNames cities in Romania and Moldova	Constanţa, Galaţi, Piatra Neamţ, Reşiţa, Bistriţa, Bălţi	Rădăuți
Wikidata given names	none	Ionuț, Laurențiu, Luminița
Unicode names list	says U+021B should be used instead for Romanian	points back to ţ

So the places card on this page files Constanţa and Galaţi under the cedilla t. That is the data as GeoNames spells it, not the Romanian spelling. Wiktionary marks ţ as a nonstandard Romanian form of ț, and the Romanian alphabet page lists only ț.

Before you reuse a Romanian place name from such data, search it for U+0163 and U+0162. Then check the language of the line. In a transliterated Arabic name, ţ is a legitimate spelling of ط, which Wiktionary gives as a synonym of ṭ. A blind replace would turn it into a Romanian letter it never was.

What does a t with accent look like after a code-page mix-up?

A t with accent in a mixed-up file often turns into a different letter, and sometimes into nothing at all. We encoded each word below with Unicode's mapping file for the code page it was saved in, then decoded it with the WHATWG TextDecoder in Node.js v26.0.0 (ICU 78.3) on 2026-10-10. This is our own measurement. The UTF-8 rows match the mojibake values stored in AccentPad's facts.

You see	It was	Saved as	Opened as
Constanþa	Constanţa or Constanța	cp1250, ISO 8859-2 or ISO 8859-16	Windows-1252
þarã	ţară	cp1250	Windows-1252
Piešany, with a hidden control character	Piešťany	cp1250	Windows-1252
Pie¹»any	Piešťany	ISO 8859-2	Windows-1252
Rădăuţi	Rădăuți	ISO 8859-16	Windows-1250
ConstanÅ£a	Constanţa	UTF-8	Windows-1252
ConstanÈ›a	Constanța	UTF-8	Windows-1252
PieÅ¡Å¥any	Piešťany	UTF-8	Windows-1252
Laá¹­Ä«f	Laṭīf	UTF-8	Windows-1252

The thorn rows have one cause. Byte FE is ţ in cp1250 (opens in a new tab) and ISO 8859-2 (opens in a new tab), ț in ISO 8859-16 (opens in a new tab), and þ, the Icelandic thorn, in Windows-1252 (opens in a new tab). So both Romanian t letters come out as þ. The word țară, country in Wiktionary (opens in a new tab), saved with the old cedilla t, reads þarã.

The vanishing ť is worse, because nothing marks the spot. Cp1250 stores ť at byte 9D. The Windows-1252 mapping file leaves 9D undefined, and the WHATWG index (opens in a new tab) decodes it as the control U+009D, which draws nothing. A reader sees Piešany and has no reason to suspect a missing letter.

The Rădăuţi row is the quiet one. ISO 8859-16 read as Windows-1250 keeps every Romanian letter readable, but the comma t turns into the cedilla t, and ș into ş the same way. That is one way a ț can become ţ without anyone typing it. In each case, reopen the file with the code page it was saved in; do not retype the letters.

Which Arabic, Sanskrit and Hebrew letters turn into ṭ, ṯ or ẗ?

In transliteration, a t with accent belongs to no alphabet. It stands for a letter of another script, and the scheme decides which mark it gets. Wiktionary's entries give these pairs.

Letter in its script	Script	Latin form	Other spellings Wiktionary gives
ط	Arabic	ṭ	ţ
ث	Arabic	ṯ	t͟h
ة	Arabic	ẗ
ጠ	Amharic	ṭ
ट	Sanskrit	ṭ, in IAST and ISO
ת	Hebrew	ṯ, in ISO 259	plain t, in the common Israeli and Hebrew Academy schemes

The dot below means "not the plain t", and the language sets the sound. Wiktionary calls the Arabic and Amharic one an emphatic t. For Sanskrit and in phonetic notation it is a retroflex t, the IPA ʈ, said with the tongue tip curled back. Unicode's names list sums up ṭ as Indic transliteration and ṯ as Semitic transliteration.

Names show it in use. Wikidata, in this page's names card, spells ʿAbd al-Laṭīf and Fāṭima with ṭ for the Arabic ط. A search for Latif in a plain-text index will not match the exact string Laṭīf, so store both spellings or compare with accents ignored.

Why does ẗ have no capital letter of its own?

Unicode never encoded one, so ẗ is the only t with accent in AccentPad's catalogue that exists in lowercase alone. SpecialCasing.txt (opens in a new tab) maps ẗ to an uppercase made of two code points: T followed by U+0308, the combining diaeresis. It lists ǰ, the j with caron, nearby for the same reason. That is why the copy tiles at the top have no capital for ẗ when you switch them to Uppercase, and the list of all t letters ends its capitals without one.

We checked the effect in Node.js v26.0.0 on 2026-10-10. Uppercasing ẗ gave two code points, and NFC normalization left them as two, because no single capital exists to compose into. Lowercasing that pair and applying NFC returned the one-character ẗ. So a title in capitals grows by one code point for every ẗ in it. Code that expects a word to keep its length when its case changes, or a field limited by characters, can break there.