Writing emerged among Minoan civilization of the eastern Mediterranean island of Crete around the end of the 3rd millennium BCE. The earliest known inscriptions, dating to c. 2100, is the Arkhanes formula, a sequence of five signs which appears in various inscriptions over the following centuries, generally classified as the first attestation of the Cretan hieroglyphs, an undeciphered syllabary.^[α] These hieroglyphs appear to have developed out of earlier preliterate seals and iconography. As Crete was likely exposed to other writing systems during this period, most notably Egyptian hieroglyphs, this is not generally regarded as an independent invention of writing.^[3][4][5] Cretan hieroglyphs are the earliest of the Aegean scripts, a set of syllabic writing systems which emerged in the ancient eastern Mediterranean. Around 1900–1800 BCE, another undeciphered syllabary known as Linear A emerged. Rather than evolving from Cretan hieroglyphics, it appears to have been a separate innovation, only sharing about 20 of its 97 signs.^[6][7]

The two scripts coexisted on Crete for several centuries, both generally used for economic records, before the disappearance of the hieroglyphs around 1600. The reason two independent scripts were used for the same sort of documents on the island is unknown, as is whether they encoded the same unknown language—Minoan—or separate languages. It is possible that Linear A itself recorded multiple languages. The two scripts have been found together at the same archaeological sites. Linear A is attested across Crete, various Aegean Sea islands, as well as the eastern coast of Anatolia at Miletus. It likely also inspired another Aegean script on Cyprus, the Cypro-Minoan syllabary. Around 1450, the Mycenaean Greeks of the mainland invaded Crete, and Linear A largely fell out of use, beyond sporadic use in private and religious contexts, evidenced by a painted inscription found on a figurine from Poros-Heraklion dating to around 1350.^[6][8][9] Some Linear A inscriptions can be read using apparent glyph correspondences to later Linear B and Cypro-Minoan signs, although it is unknown if such glyphs kept their phonetic value between the systems. As the underlying language or languages are unknown, it may be impossible to read if they belong to an otherwise unknown language family.^[10]

At some point after 1600 and before c. 1370, Linear A was adapted into Linear B in order to encode Mycenaean Greek, the earliest recorded form of the Greek language. Where Linear B was created is a matter of academic dispute, with scholars divided between an origin on Crete or on the Greek mainland. Mycenaeans may have developed the script on Crete and exported it home to the mainland, or it may have been developed in mainland Greece and then brought to Crete with their conquest of the island. An origin in Anatolia at Miletus has also been proposed, although the earliest known Linear B documents may predate the known Myceanean presence in the area.^[11]

All known Linear B inscriptions date from the 14th century BC to the beginning of the 12th. The peak of finds occurs during the Late Helladic IIIB, corresponding to the 13th century BC.^[12][13]

Scribes used sharp styluses to write Linear B texts on the clay tablets.^[14]

Linear B is a logosyllabic script, consisting of syllabic and ideographic signs. Its core inventory consists of at least 59 syllabograms, all representing open syllables (syllables which end in a vowel). These are supplemented by 27 alternate syllabograms and around 170 logograms or ideograms,^[β] which represent specific items which were counted in administrative records. Some signs serve as both syllabograms and logograms, such as 𐀥, which represents both the syllable qi and the logogram for sheep.^[16][17]

The core syllabary consists of at least 59 syllabograms, not including undeciphered signs. This allows for combinations between thirteen possible consonants (including a null consonant, i.e. a syllable which consists only of a vowel) and five vowels.^[18] This is supplemented by fourteen known additional signs, which are not required to write any given word in the language, but which can give a more precise phonetic value than the base signs.^[19]

Core Linear B syllabograms^[20][16]

	a		e		i		o		u
∅	𐀀	a *08	𐀁	e *38	𐀂	i *28	𐀃	o *61	𐀄	u *10
d	𐀅	da *01	𐀆	de *45	𐀇	di *07	𐀈	do *14	𐀉	du *51
j	𐀊	ja *57	𐀋	je *46	—		𐀍	jo *36	—
g/k(h)	𐀏	ka *77	𐀐	ke *44	𐀑	ki *67	𐀒	ko *70	𐀓	ku *81
m	𐀔	ma *80	𐀕	me *13	𐀖	mi *73	𐀗	mo *15	𐀘	mu *23
n	𐀙	na *06	𐀚	ne *24	𐀛	ni *30	𐀜	no *52	𐀝	nu *55
b/p(h)	𐀞	pa *03	𐀟	pe *72	𐀠	pi *39	𐀡	po *11	𐀢	pu *50
kʷ(h)	𐀣	qa *16	𐀤	qe *78	𐀥	qi *21	𐀦	qo *32	—
l/r	𐀨	ra *60	𐀩	re *27	𐀪	ri *53	𐀫	ro *02	𐀬	ru *26
s	𐀭	sa *31	𐀮	se *09	𐀯	si *41	𐀰	so *12	𐀱	su *58
t(h)	𐀲	ta *59	𐀳	te *04	𐀴	ti *37	𐀵	to *05	𐀶	tu *69
w	𐀷	wa *54	𐀸	we *75	𐀹	wi *40	𐀺	wo *42	—|
z	𐀼	za *17	𐀽	ze *74	—|		𐀿	zo *20	—|

Archaeological study of Linear B began in Heraklion in 1895, when the British archaeologist Arthur Evans was shown a ceramic fragment containing the script from the hill of Kephala, the site of the palace of Knossos. This fragment may have been initially excavated by Cretan businessman Minos Kalokairinos during his excavations at Kephala in 1878. Evans continued Kalokairinos's excavations at Knossos, excavating the entirety of the ancient palace and recovering around 3,500 inscribed tablets. He identified the Cretan hieroglyphs as a writing system, distinguishing them from two "highly developed linearised systems", Linear A and Linear B.^[21][22] Evans identified the extensive archives from Knossos as a mixture of inventories, deeds, and public records. He made studies of the Linear B grammar, identifying suffixes, a gender system, and compounding in the manner of the Indo-European languages. He had planned to publish the Knossos Linear B tablets in a sequel to his 1909 Scripta Minoa I, but had not finished his notes for the project by the time of his death in 1941. Evans never published his full studies on the script.^[23][24]

Although a very small corpus of the script had been published, various scholars—both professional and amateur—attempted to decipher it during the early 20th century. These early attempts, following Evans's theories, often conflated the Cretan hieroglyphs, Linear A, Linear B, and the signs of the Phaistos Disc, as writing systems for one underlying language. Many based their decipherment on the already-attested values of the Cypriot syllabary, and attempted to prove a connection to various languages and language families, ranging from ancient languages such as Sumerian and Egyptian to far-flung modern languages such as Basque and Finnish.^[24] Prominent proposals included an early stage of the Greek language, an unknown pre-Greek Indo-European language, or a non-Indo-European language such as the Tyrsenian languages, relatives of Etruscan.^[25]

During the 1920s, philologist Arthur Ernest Cowley identified three words, although wrote that the underlying language could not be Greek.^[23] Twenty years later, the archaeologist Alice Kober demonstrated that the script showed an inflection system, identifying three grammatical cases and a set of signs with shared vowel sounds. By the late 1940s, photographs of the Pylos tablets (which had been put in storage during World War II) had become available to researchers. Philologist Emmett L. Bennett Jr. wrote a PhD thesis which proposed transcriptions for the tablets, later publishing these in The Pylos Tablets (1951) and Minoan Linear B Index (1953). Bennett classified Linear B signs into 78 signs used in sign-groups (likely phonetic), 63 ideograms, and numerals.^[26][27][28] John Myres, a friend of Evans, worked to compile the Knossos tablets in full, receiving help from Kober (until her death in 1950) and Bennett. In 1952, the tablets were published in Scripta Minoa II, allowing for an easily accessible view of a Linear B corpus more than twice the size of the Pylos tablets.^[29]

Michael Ventris, after learning of the Aegean scripts as a schoolboy, published his first article on the topic in 1940. He returned to study of Linear B following military service during World War II and architectural training. Ventris sent a questionnaire to various experts in the field, translated their replies, and circulated these among his mailing list alongside regular updates on his work. He initially interpreted Linear B as encoding a language related to Etruscan, seeing Greek as an unlikely candidate due to archaeological evidence and past decipherment attempts. In his eleventh set of work notes, he noted a common three-word phrase within the Pylos tablets; the first word was always different, the second was always one of four possible words, and the third was the same as the second except with a different endings. He theorized that this phrase could mean "A, the male/female servant of B", and tentatively proposed the readings do-we-lo and do-we-la for the middle word; an apparent etymology for the Greek δούλος ('slave'). However, as the hypothetical Linear B form lacked the expected masculine ending, Ventris initially rejected that the language was actually Greek.^[30]

Ventris employed a syllabic grid in his decipherment attempts: symbols suspected to share a particular vowel were placed in the same column, while those suspected to share a consonant were placed in the same row. This format allowed him to test alternate combinations and make provisional guesses for phonetic readings. He searched the corpus of Cretan tablets for place names; as these generally changed little over time, the Bronze Age toponyms may be recognizable as their later Classical forms. Using tentative values from his grid, Ventris identified the harbor of Amnisos in the form a-mi-ni-so, and from this identified the words for Knossos (ko-no-so) and Tylissos (tu-li-so).^[31] He was able to find various Greek cognates, and compiled a vocabulary list; he announced his findings first in letters to Bennett and Myres in June 1952, and then publicly on a BBC cultural broadcast in July. The classicist John Chadwick connected with Ventris after this broadcast. The two collaborated on further decipherment, and the following year they published an article on their findings entitled "Evidence for Greek Dialect in the Mycenaean Archives". This was followed by their book Documents in Mycenaean Greek in 1956, published shortly after Ventris's death in a motor accident.^[31][32]

The initial decipherment published in Ventris and Chadwick's 1953 article has remained largely accurate in the wake of further research: none of the proposed phonetic volumes have fallen out of favor, with the only major addition to the system being a new class of syllabogram.^[33]

Nearly the entire corpus of Linear B texts is derived from clay media found in major settlements. Administrative documents, the most prevalent, take the form of tablets, labels, and nodules. These were written on unbaked clay, which was then air-dried; they were only incidentally preserved during fires. The study of these documents—termed pinacology^[γ] by Mycenologists— is severely hampered by their distribution and relative rarity. They were intended as ephemeral documents, only to be used for the duration of an administrative year. A smaller corpus includes text (generally painted) on clay vessels before firing; these are most common on stirrup jars, but also found on skyphoi, kylikes, and bowls.^[13][35][36] Solitary examples of Linear B text in other media are attested, such as an ivory seal or a inscribed stone.^[36]

It is unknown if the Mycenaeans produced long-term administrative records on perishable media. Clay tablets were seen as ephemeral documents themselves, as they were left unfired and frequently re-used. Although they were sometimes stored together, there is no evidence that the texts were archived. None reference specific dates, instead referring vaguely to the current year or the previous year; it is likely that they were not kept for more than a year after their creation.^[36][37]

Linear B texts are classified according to the site they were found, their excavation number within the site, and by their content. A two or three letter abbreviation for the site is given at the start of a tablet's designation: for instance, a tablet labeled MY was excavated from Mycenae, while a tablet labeled KN was excavated at Knossos. Following the site designation, capital letters denote the document's class.^[42]

Evans divided the Linear B tablets into page-shaped (roughly rectangular) and leaf-shaped (long and narrow) types, the latter named for a resemblance to South Asian palm-leaf manuscripts.

Linear B corpora are academic publications which include photographs (and/or facsimile drawings) and transcriptions of text from a particular site. Updates are often given through "minor editions", which provide transcriptions of all texts at a site without including photographs or facsimiles. Beginning with Scripta Minoa II in 1952, the Knossos tablets have been published in seven different editions. Successive editions included more fragments discovered in the Heraklion Museum, as well as fragments from the same text which were late rejoined. Such corpora can quickly become obsolete due to new discoveries or research; the first volume of the four-volume Corpus of Mycenaean Inscriptions from Knossos, released in 1986, was obsoleted by The Knossos Tablets, a 1989 minor edition, before another volume of the corpus could be finished.^[43]

Indexes of Linear B list all words attested from recovered texts in Latin alphabetical order, as well as the ideograms in numerical order. Each entry in such publications

Linear B was added to the Unicode Standard in 2003 under three blocks: Linear B Syllabary block, from U+10000–U+1007F, the Linear B Ideograms block from U+10080–U+100FF, and the Aegean Numbers block from U+10100–U+1013F. This enables computers to easily render Linear B characters, and allows for digital fonts to be created for the system.^[44][45]

• Nestor – International bibliography of Aegean studies, including Linear B research
• LiBER (Linear B Electronic Resources) – Online database of Linear B texts