1 Defining the (Un)definable: What Is Linear A?

It is often said that it all starts with the Homeric past.

Does it?

In the Odyssey, Crete makes its appearance as multicultural and multilingual. Different peoples are mentioned in what may well be the longest description of Cretan multiculturalism, enriched by linguistic diversity. From this culturally vibrant and myth-infused island, however, no memory seems to have survived of the existence of writing (or a plurality thereof), which by the era of the Greek epics had already been lost to time. We now know that writing did exist in the Aegean way before the archaic period,Footnote ² but partaking in this knowledge is a modern privilege. Where does it all start then? More explicitly for the purposes of our discussion, when was Linear A rediscovered? Before digging up Linear A’s past, let us first set out to understand what Linear A is.

In mainstream scholarly literature, Linear A is commonly defined as the logo-syllabic, or ideo-syllabic, writing system (or script) used on Crete and the Aegean islands during the Bronze Age (ca. 1800–1450 BCE) to write the still poorly understood Minoan language.Footnote ³ Usually, it is also specified that Linear A remains undeciphered to date, despite the many attempts at breaking the code. However accurate, this definition – if read critically – may need some clarification. To begin with, we encounter three terms of common use, whose precise and contextual meaning may not be straightforward: ‘writing system’, ‘script’ and ‘language’.Footnote ⁴ The first two (‘writing system’ and ‘script’) are often used interchangeably, but they refer to different concepts. For the sake of accuracy, a ‘writing system’ consists of a ‘script’, which is the graphic (visual) representation of speech in the form of written signs (‘graphemes’, from Greek γράφω /gráp^hō/ ‘to write’), plus a set of conventions (orthography) that allow a language to be written down correctly.

Here the term ‘language’ comes into play, making us better understand its role when talking about written communication. Linear A can therefore be referred to as both ‘writing system’ and ‘script’, depending on the aspects being studied, but it cannot be addressed as ‘language’ (too common a mistake in everyday discourse). What Linear A does is to write down, as a code, the ‘language’ spoken by the inhabitants of Bronze Age Crete, the so-called Minoans. Hence we can talk of the ‘Minoan language’ but not of the ‘Linear A language’. This leads us to discuss the term ‘undeciphered’, equally too often misguidedly used. If we bring to mind that, broadly speaking, ‘to decipher’ means to convert a code into understandable language, it is therefore accurate to state that Linear A (which is a written code of communication) still remains ‘undeciphered’, whereas stating that ‘Minoan’ (which is a language) remains ‘undeciphered’ is indeed inaccurate.

Let us now move on to less familiar terms: ‘logo-syllabic’ and ‘ideo-syllabic’. These are both used to define a writing system from a typological standpoint and express a variation on the type ‘syllabary’. A syllabary (or syllabic writing system) denotes a system in which written characters (commonly called signs) stand for syllables of the coded language, namely units of speech containing a vowel, with or without surrounding consonants (e.g. /a/, /do/, /mi/, etc.). Instead, in alphabetic systems (like English) each character stands for a single sound, be it a vowel or a consonant (e.g. /e/, /m/, /t/). Linear A is understood to be a syllabary, consisting of approximately 150 characters standing for syllables, which are standardly called ‘syllabograms’ (from Greek γράμμα /grámma/ ‘letter’). In addition to syllabograms, Linear A also has a good number of signs (almost 200) understood to represent entire words, concepts or ideas.Footnote ⁵

If a sign stands for a word, it is called a ‘logogram’ (from Greek λόγος /lógos/ ‘word’). Believe it or not, logograms are also part of our day-to-day life. Think of numbers, for example: characters like ‘1’, ‘40’, ‘100’ stand for entire words in English, since they read as ‘one’, ‘forty’, ‘one hundred’. Other modern examples are the ampersand (‘&’), expected to be read as ‘and’ (in English, but e.g. ‘og’ in Danish, ‘und’ in German, ‘e’ in Italian and so forth), and the per cent sign ‘%’, read as ‘per cent’ (in English). If a sign stands for a concept or idea, it is called an ‘ideogram’ (from Greek ἰδέα /idéa/ ‘idea, form’). Are there ideograms in English (and/or other modern languages)? Surprisingly, yes. For example, consider wayfinding signage and street/traffic signs, whose meaning can be understood by people regardless of the language they speak (e.g. arrows indicate directionality). This said, also note that the above examples of ‘logograms’ in English (i.e. numerals, ‘&’, ‘%’) can be taken as ‘ideograms’ on a global context, because the ideas (or concepts) they express can be understood regardless of their pronunciation in any given language.

Going back to Linear A, the current consensus is that Linear A may have both logograms and ideograms (with the latter presumably higher in number). Given that Linear A remains undeciphered, a straightforward distinction between these two categories is not easy to make (or may not necessarily need to be made; see especially Salgarella Reference Salgarella2020: 42–149). The more-inclusive term ‘sematogram’ (from Greek σῆμα /sēma/ ‘sign’) has also been proposed (Petrakis Reference Petrakis, Carlier, de Lamberterie, Egetmeyer, Guilleux, Rougemont and Zurbach2012; Reference Petrakis, Nosch and Landenius Enegren2017) to refer to Linear A non-phonetic signs (both ‘logograms’ and ‘ideograms’), which is worth consideration. In this Element, the conventional terminology is retained consistently, since it is widely used in the scholarly literature, with a preference for the more generic term ‘ideogram’ over ‘logogram’ (see also Section 4).Footnote ⁶

Last but not the least, the very name ‘Linear A’ is not unproblematic: is ‘Linear A’ the real name of Linear A? Not in the slightest. Most of the terminology used nowadays to refer to the civilisation that flourished in Bronze Age Crete goes back to Sir Arthur Evans (1851–1941), the British archaeologist renowned for unearthing the palace of Knossos on Crete in the early nineteenth century.Footnote ⁷ Not only did he uncover the remains of a civilization whose memory had been lost to time for millennia, but he is also responsible for constructing the narrative surrounding his finds. Drawing on Greek mythology, he dubbed ‘Minoans’ (after legendary king Minos) the inhabitants of Bronze Age Crete, with Knossos being identified with ‘the Palace of Minos’. To Evans we also owe the names ‘Cretan Hieroglyphic’, ‘Linear A’ and ‘Linear B’ to refer to the three Cretan scripts that came to light during his excavations at Knossos. The script understood as the oldest was dubbed ‘hieroglyphic’ for its pictographic character that reminded Evans of Egyptian hieroglyphs (note that Cretan Hieroglyphic is not derived from Egyptian hieroglyphs, although its creation may have been influenced by Egyptian writing).Footnote ⁸ If compared to this earlier script, the two later scripts seemed to Evans more cursive, more ‘linear’, as it were. Hence the name ‘Linear’ came about, followed by the letters ‘A’ and ‘B’ to distinguish between two slightly different variants as well as chronological stages (Evans’ initial dubbing being ‘Linear script of Class A’ and ‘Linear script of Class B’).

Linear A was understood to be older than Linear B and also more widespread (with some evidence recently dug up during fieldwork at Haghia Triada in central-south Crete). Linear B, instead, had thus far been found at Knossos only. Since Linear B was not only later in date but also graphically different from Linear A to an extent, Evans initially understood it as a ‘calligraphic’ variant of Linear A restricted to Knossos. This latter variant, in Evans’ view, superseded the former as a result of a dynastic revolution. Later on, Evans changed his mind, seeing Linear A and B as ‘parallel’ sister scripts derived from ‘common prototypes’ – these being a sort of proto-linear ancestor whose existence Evans only speculated about but never found evidence of. His views about script relationships, however, changed over time and never quite settled (for a detailed discussion, see Schoep Reference Schoep2018; Salgarella Reference Salgarella2020: 10–30). The separation between Linear A and Linear B as different scripts, as well as writing systems, was triggered by the decipherment of Linear B. In 1952 British architect and self-educated linguist Michael Ventris proved that Linear B encoded an early form of the Greek language, subsequently dubbed ‘Mycenaean’.Footnote ⁹ Since Linear A could not be decoded in the same way as Linear B, its language remained obscure and most unlikely to be Greek. This linguistic difference brought about a starker cultural separation between Greek-speaking ‘Mycenaeans’ writing in Linear B and non-Greek-speaking ‘Minoans’ writing in Linear A. Hence Linear A and Linear B came to be seen as entirely different writing systems and scripts. This view has been influential ever since and has only recently been questioned to promote a more nuanced approach to script development and cultural interpretations (Salgarella Reference Salgarella2020).

In the light of the previous discussion, it now becomes clearer why defining ‘Linear A’ unequivocally is a hard task. For, we have seen that (i) the definition of Linear A as found in print requires further explanation for adequate understanding; (ii) the name given to this writing system is a modern label and (iii) traditional views on the relationship between Linear A and Linear B may be somewhat misleading, preventing us from reaching a less dichotomous appreciation of their (respective and overall) features. Moreover, we do not have any contemporary sources (e.g. grammatical works, meta-linguistic material; see also Section 9) about what we call ‘Linear A’, how it worked, how it was created, how it was taught (and transmitted) and its relationship with both (later) Linear B and (earlier) Cretan Hieroglyphic (not to mention Cypro-Minoan, also thought to be an offspring of Linear A). Therefore, when investigating the nature and characteristics of Bronze Age Cretan writing, we resort to using modern approaches, terminology and theoretical frameworks only. There are also other reasons why Linear A defies a clear-cut definition. To better understand these, we first need to contextualise Linear A among the Aegean (and Cypriot) scripts, a few of which readers may be already familiar with. The most well-known are Cretan Hieroglyphic, Linear A, Linear B, Cypro-Minoan and Cypro-Syllabic, while the Arkhanes Script, the scripts of the Phaistos Disk and the Arkalochori Axe, and Eteocretan are perhaps less well-known.Footnote ¹⁰ Aegean scripts lend themselves to multiple groupings based on the criteria chosen for their classification, as follows.

• Chronological grouping: Bronze Age versus Iron Age scripts

  • Bronze Age: Arkhanes Script, Phaistos Disk, Arkalochori Axe, Cretan Hieroglyphic, Linear A, Linear B, Cypro-Minoan

  • Iron Age: Cypro-Syllabic, Eteocretan

• Geographical grouping: Cretan versus CypriotFootnote ¹¹

  • Cretan: Arkhanes Script, Phaistos Disk, Arkalochori Axe, Cretan Hieroglyphic, Linear A, Linear B, Eteocretan

  • Cypriot: Cypro-Minoan, Cypro-Syllabic

• Typological grouping: ideo-syllabary versus alphabet

  • Ideo-syllabary: Arkhanes Script, Phaistos Disk, Arkalochori Axe, Cretan Hieroglyphic, Linear A, Linear B, Cypro-Minoan, Cypro-Syllabic

  • Alphabet: Eteocretan (Greek alphabet)

• Linguistic grouping: Greek versus non-Greek

  • Greek: Linear B, Cypro-Syllabic

  • Non-Greek: Arkhanes Script, Phaistos Disk, Arkalochori Axe, Cretan Hieroglyphic, Linear A, Cypro-Minoan, Eteocretan

Zooming in on Linear A, it presents itself as a Bronze Age script of Cretan origin, of ideo-syllabic type, encoding a language not yet identified. Having placed Linear A in a broader context, we can now mention another reason why it defies a clear-cut definition. This is the graphic similarity Linear A shows with a number of Aegean scripts, which makes it difficult to draw straight lines between categories. Looking backward in time, the origins of Linear A have blurred boundaries: among the earliest Linear A inscriptions, there are a few labelled dubitanda (i.e. ‘of doubtful reading’), as they could equally be taken as written in Cretan Hieroglyphic (listed in CHIC: 18); likewise, the Arkhanes Script is taken as either an early form (or ancestor) of Linear A or a separate (yet related) script.Footnote ¹² Needless to say, the debate surrounding the relations between Linear A, the Phaistos Disk and Arkalochori Axe is also far from settled.

Looking forward in time, we find Linear A’s offspring Linear B. We have seen that Evans’ understanding of Linear A and Linear B as either one or two scripts fluctuated over time, without ever truly settling. The split between Linear A and Linear B as different scripts (the former associated with the ‘Minoans’, the latter with the ‘Mycenaeans’) became sharper after the decipherment of Linear B, when script designations came to take on cultural meaning.Footnote ¹³ This view has influenced subsequent approaches to script development as well as our appreciation of the population groups we currently dub ‘Minoans’ and ‘Mycenaeans’. Scholars are now distancing themselves from dichotomous classifications, in favour of a more nuanced and fluid approach emphasising the role played by progressive, incremental change in both script development and cultural transformation, as well as problematising implicit associations of material culture (of which ‘script’ is part) with broader, blanket-like cultural narratives.Footnote ¹⁴

Dismissing the role of progressive change may, in fact, have been a contributing factor to a contrastive distinction between Linear A and Linear B. There is a real chronological gap of about fifty years between the latest (extant) inscriptions in Linear A and the earliest (extant) inscriptions in Linear B. It is unclear whether this absence of written evidence from the archaeological record reflects a real historical circumstance or it is simply due to chance and accidents of preservation. Until further evidence is discovered (if ever), there is no way of proving (or disproving) either hypothesis. At the current state of knowledge, this absence implies that we do not know what the Linear script may have looked like in the transitional period from the so-called Minoan to Mycenaean administration on Crete.

To conclude, what we dub ‘Linear A’ is a tricky entity that needs to be handled with intellectual care. General, broadly applied definitions fail to describe what still escapes our full understanding, for which there are more questions than answers. This Element, therefore, aims to problematise what we know about ‘Linear A’ by raising a set of (still hotly debated) questions and by offering to readers the most comprehensive and research-informed array of plausible hypotheses put forward to date in the academic debate.

2 A Tale of Life and Death: What Is the Lifespan of Linear A?

Surely not a short-lived script, Linear A’s life spanned nearly five centuries from ca. 1800 to 1450 BCE.Footnote ¹⁵ In scholarly literature, we also find the alternative nomenclature ‘from Middle Minoan II to Late Minoan IB’, with reference to relative dating. Relative dating may not be as straightforward as absolute dating (which provides specific dates), because relative dating creates chronological sequences that give an estimate of how old an artefact is in comparison to other artefacts and/or archaeological sites, without however specifying its absolute age. Since pottery is often used as a diagnostic tool to determine the relative order of sequences, we often call these ‘pottery phases’. In addition to ‘pottery phases’, we can also find labels referring to geographically-based ‘cultural phases’, following a culture-historical approach to archaeological dating. Hence the absolute dates 1800 to 1450 BCE correspond to ‘Proto-Palatial’ (‘First Palaces’) to ‘Neo-Palatial’ (‘New Palaces’) periods on Crete (Table 1).

Table 1Absolute and relative dating of Linear A¹⁶

Table 1Long description

The table has 4 columns: Absolute dating (high), Pottery phase, Cultural phase, and Scripts in use. It reads as follows: Row 1: 1900 to 1800; Middle Minoan 2; Proto-Palatial; Cretan Hieroglyphic and Linear A. Row 2: 1800 to 1700; Middle Minoan 3; Proto-Palatial; Cretan Hieroglyphic and Linear A. Row 3: 1700 to 1600; Late Minoan 1 A; Neo-Palatial; Linear A. Row 4: 1600 to 1450; Late Minoan 1 B; Neo-Palatial; Linear A. Row 5: 1450 to 1400; Late Minoan 2; Final-Palatial; (unclear). Row 6: 1400 to 1375; Late Minoan 3 A 1; Final-Palatial; Linear B. Row 7: 1375 to 1300; Late Minoan 3 A 2; Post-Palatial; Linear B. Row 8: 1300 to 1200; Late Minoan 3 B; Post-Palatial; Linear B.

Most Linear A evidence comes from Late Minoan IB, corresponding to the end of the Neo-Palatial period (ca. 1450 BCE). This period was characterised by island-wide, large-scale destructive events, during which almost all Minoan palatial centres (‘palaces’) were destroyed. A seismic island, Crete was most likely hit by severe earthquakes followed by fires, which, on the bright side, were responsible for baking Linear A inscriptions on clay and thus preserving them until today. Indeed, most Linear A evidence that has come down to us consists of small clay tablets used as administrative records for the bookkeeping of palatial centres, although other materials (e.g. stone) were also used as writing media (on material supports, see Section 5; on document types, see Section 6). The destruction that took place around 1450 BCE marked the end of the Neo-Palatial period, argued to coincide with the end of the ‘Minoan’ civilisation as well (although softer approaches are being explored). After this event Linear A ceased to be used, at least in administrative contexts, due to the fall of the system for whose purposes Linear A was mostly employed (and perhaps created?). No written documents survive from the following period (Late Minoan II). We need to wait until Late Minoan III (ca. 1400–1375 BCE) to witness the earliest (extant) evidence of Linear B.

Although the most severe, the 1450 BCE destruction was not the only one. The ends of the other period ‘breaks’ (as illustrated in Table 1) are equally characterised by destruction horizons. Thanks to these destructions and changes in material culture, archaeologists theorise successive chronological blocks, characterised by some degree of internal uniformity.Footnote ¹⁷ This is why the chronologies illustrated in Table 1 appear that way. Unfortunate as these events were for causing serious physical damage, they are also the fortunate reason behind the preservation of Linear A clay administrative records throughout time, until today (see Section 6.1). It has to be borne in mind that most of the extant evidence has therefore survived because of accidents of preservation, meaning that we do not have the whole picture of Linear A use (see Section 9 on the likely existence of written documents in perishable materials, such as parchment and/or papyrus, obliterated by fire destructions). Whenever a centre (palatial or not)Footnote ¹⁸ was destroyed, it provided us with a snapshot of contemporary Linear A use. Although the bulk of Linear A is dated to Late Minoan IB (especially evidence from Haghia Triada and Khania), some sites underwent slightly earlier destructions (Phaistos: Middle Minoan IIB–IIIB; Mallia: Middle Minoan IIIB; Knossos, Akrotiri: Late Minoan IA) or later (Zakros: very end of Late Minoan IB; geography discussed in Section 3).

So, when do we find the earliest evidence of Linear A? This is not uncontroversial and is inextricably intertwined with theories on the ‘birth’ of Linear A, calling into question its relation with two earlier Cretan scripts (Cretan Hieroglyphic and the Arkhanes Script).Footnote ¹⁹ The earliest known inscriptions argued to be written in Linear A date to Middle Minoan II (ca. 1900–1800 BCE)Footnote ²⁰ and have blurred boundaries with Cretan Hieroglyphic. The Middle Minoan period, in fact, saw the rise and fall of Cretan Hieroglyphic, with its archival documents spanning Middle Minoan II (Knossos, Petras) to III (Mallia). A few clay inscriptions found at Knossos and Mallia, currently published in the corpus of Cretan Hieroglyphic (CHIC: 18), may however be equally taken as Linear A and are hence labelled dubitanda (‘of doubtful reading’).

Is this circumstance suggestive of a derivation of Linear A from Cretan Hieroglyphic? Maybe – or maybe not. The problem is that evidence for both scripts appear as early as Middle Minoan II, with Cretan Hieroglyphic taking flight first, especially on north Crete (Knossos, Mallia, Petras), whereas Linear A comes second, spreading from central-south Crete (Phaistos). If Cretan Hieroglyphic had the edge over Linear A in the Middle Minoan period, Linear A clearly lorded it over the former in Late Minoan I. But which one was first invented then? Perhaps thinking in terms of chronological precedence is not a productive way of addressing this issue. It has recently been suggested that, instead of theorising script development in monogenetic, unidirectional evolutionary terms (with one system developing straight out of another), the emergence of writing on Crete may be best understood as a sequence of step-like inventions, ‘as rapid bursts followed by subsequent modifications that extended over long periods of time’ (Schoep 2020: 52).Footnote ²¹ The inspiration for the formation of full-fledged writing systems in Middle Minoan may be found in earlier experimentations with writing which took place in the previous period, namely Early Minoan sealstones and the Arkhanes Script.Footnote ²² However, as of now, there is no probative evidence for these to be taken as early forms of either Cretan Hieroglyphic or Linear A. The precise dynamics behind the emergence of writing on Crete, complex as they are, still elude us and are a subject of academic inquiry.

Let us now move to the opposite side of the timeline: what is the latest datable evidence of Linear A? As mentioned, Linear A did not outlast the fall of the (Neo-)Palatial system at the end of Late Minoan IB.Footnote ²³ In the following period (Late Minoan II), we witness what is often called an ‘epigraphic gap’, without evidence of writing. Yet this is not entirely true. In fact, if writing seems not to have survived the 1450 BCE destruction horizon outside of ‘formal’ contexts (i.e. palatial structures or places for cultic activity), there is some evidence suggesting that writing had not been entirely lost. We have a few Linear A spillovers from ‘informal’ contexts, all coming (most notably) from the area around Knossos: two carved inscriptions (Late Minoan II), one on a building block, the other on a pithoid jar; two painted inscriptions (Late Minoan IIIA1), one on a cultic figurine, the other on a miniature cup.Footnote ²⁴

This epigraphic silence was broken around 1400 to 1375 BCE, which produced the earliest Linear B inscriptions known to date.Footnote ²⁵ Linear B appears like a burst, as much a full-fledged writing system as Linear A was. But how so, given the preceding epigraphic gap? It is still unclear whether absence of evidence (from Late Minoan II) is actual evidence of absence or rather a chance event due to accidents of preservation. Yet, this absence prevents us from appreciating how Linear B originated (and when exactly) or, better, from understanding how this ‘Linear Script’ gradually developed over time transitioning from Linear A (accommodating the Minoan language) to Linear B (accommodating the Greek language). This transition, as well as transformation, has been variously explored by scholars to better understand the dynamics at play in the close context of the Linear Script’s transmission.Footnote ²⁶ The traditional view splitting Linear A and Linear B into two different units strictly associated with their respective ‘culture’ (Cretan Minoans versus Mainlander Mycenaeans) is now being revised in favour of a more fluid approach to both script development and societal change. The Linear Script may have survived through time, gradually adapting some of its components to suit changing needs and accommodate new features (e.g. language, purposes, uses). Ultimately, it may well not have been the end of Linear A. Linear A may never have ‘died’, as much as Linear B may never have been ‘born’: the latter may be seen as the gradual transformation of the former, captured by us in a later chronological snapshot, under different conditions of use. Seen this way, the ‘Linear’ line was never truly broken.

3 The Linear A Corpus: Where Is Linear A Found?

A total of fifty-six sites (‘findplaces’) have so far yielded Linear A inscriptions. Most findplaces are located on Crete and the Aegean islands, but a few sites located in Mainland Greece, Western Anatolia and the Levant also yielded some evidence. On Crete, the most evidence comes from Haghia Triada, Khania and Zakros, all dated to the destruction horizon that brought to an end the Neo-Palatial period (Late Minoan IB). Other relevant findplaces are Phaistos, Knossos, Mallia, Arkhanes, Tylissos and Palaikastro. In the Aegean islands, inscriptions have been found in Thera, Kythera, Kea, Melos and Samothrace. The evidence from outside of Crete is sparse, fragmentary and often of doubtful reading. In Mainland Greece finds come from Haghios Stephanos, Argos and Mycenae. In Western Anatolia Miletus and possibly Troy have yielded some evidence, and in Israel inscriptions have been dug out at Tel Haror and Tel Lachish (Figures 1–2).Footnote ²⁷ Most Linear A inscriptions were found in ‘administrative’ contexts, functioning as records of economic transactions for the bookkeeping of palatial administrations, but a smaller number of inscriptions also come from ‘non-administrative’ contexts with a cultic and/or dedicatory function (see Section 6).

Figure 1 Main findplaces of Linear A inscriptions: Crete.Footnote ²⁸

Figure 2 Main findplaces of Linear A inscriptions: outside of Crete.Footnote ²⁹

Now that we have seen where Linear A inscriptions come from, let us have a look at how much is extant: how much Linear A evidence do we have to date? All in all, the corpus of Linear A inscriptions is relatively small. Statistically speaking, only approximately 1,400 inscriptions are preserved well enough to be readable (Schoep Reference Schoep2002: 38), out of a total of around 2,500 finds (including fragmentary evidence). If we had to count every individual sign occurrence in these inscriptions, we would reach an estimated total of 7,400 (Schoep Reference Schoep2002: 38), meaning that ‘if there are 4002 characters (font Times New Roman, pitch 12, no spaces) on a 8 1/2 × 11 inch sheet of paper with 1-inch margins, all extant Linear A would take up 1.84 pages’.Footnote ³⁰ Aside from the small size of the corpus, what complicates the picture is that Linear A inscriptions are also relatively short, and the longest ones are formulaic (i.e. consisting of repeated groups of words). As a result, Linear A inscriptions display very limited syntax, which makes it difficult to investigate the language in which they are written (discussed in Section 7). So, ‘what’ from ‘where’ then? The total amount of Linear A evidence so far unearthed is given in Table 2, where references are made to both (i) each individual site that has produced inscriptions (alphabetical order) and (ii) whether the inscriptions found come from an administrative context (palatial setting) or not (mostly cultic/ritual). This is the traditional classification of Linear A inscriptions, based on the ‘function’ they are understood to have fulfilled (see Sections 5 and 6).

Table 2Findplaces (with standard abbreviations) of Linear A, with the amount of evidence yielded³¹

Table 2Long description

The table has 4 columns: Site, Non-administrative, Administrative, and Total. It reads as follows: Row 1: Apodoulou (A P); 3; blank; 3. Row 2: Argos (A R G); 1; blank; 1. Row 3: Arkalokhori (A R); 2; blank; 2. Row 4: Arkhanes (A R K H); 2; 7; 9. Row 5: Armenoi (A R M); 1; blank; 1. Row 6: Crete (C R); 4; blank; 4. Row 7: Gournia (G O); blank; 3; 3. Row 8: Haghia Triada (H T); 19; 1,250; 1,269. Row 9: Haghios Stephanos (H S); 1; blank; 1. Row 10: Iouktas (I O); 16; blank; 16. Row 11: Kamilari (K A M); 1; blank; 1. Row 12: Kannia (K A N); 1; blank; 1. Row 13: Kardamoutsa (K A); 1; blank; 1. Row 14: Kea (K E); 5; 2; 7. Row 15: Khania (K H); 1; 312; 313. Row 16: Knossos (K N); 24; 8; 32. Row 17: Kophinas (K O); 2; blank; 2. Row 18: Kythera (K Y); 2; blank; 2.

The table consists of four columns: Site, Non-administrative, Administrative, and Total. It reads the following data. Row 1. Larani (L A): 1, blank, 1. Row 2. Mallia (M A): 3, 8, 11. Row 3. Melos (M I): 2, 1, 3. Row 4. Miletus (M I L): 1, blank, 1. Row 5. Mokhlos (M O): 3, blank, 3. Row 7. Mycenae (M Y): 2, blank, 2. Row 8. Nerokourou (N E): 1, blank, 1. Row 9. Palaikastro (P K): 21, 2, 23. Row 10. Papoura (P A): blank, 1, 1. Row 11. Petras (P E): 7, 3, 10. Row 12. Petsofas (PETS): blank, 1, 1. Row 13. Phaistos (P H): 4, 61, 65. Row 14. Platanos: (P L): 1, blank, 1. Row 14. Poros (P O): 1, blank, 1. Row 15. Prassas (P R): 1, blank, 1. Row 16. Pseira (P S I): 2, blank, 2. Row 17. Psychro (P S): 1, blank, 1. Row 18. Pyrgos (P Y R): 1, 3, 4. Row 19. Samothrace (S A M): blank, 2, 2. Row 20. Selakanos (S E): 1, blank, 1. Row 21. Sitia (S I): 1, blank, 1. Row 22. Skinia (S K): 1, blank, 1. Row 23. Skoteino (S K): 1, blank, 1. Row 24. Syme (S Y): 12, blank, 12. Row 25. Tel Haror (T E L): 1, blank, 1. Row 26. Tel Lachish (L A C H): 1, blank, 1. Row 27. Thera (T H E): 8, 6, 14. Row 28. Tiryns (T I): 1, blank, 1. Row 29. Traostalos (T R A): 1, blank, 1. Row 30. Troullos (T L): 1, blank, 1. Row 31. Troy (T R O): 2, blank, 2. Row 32. Trypiti (T R Y): 1, blank, 1. Row 33. Tylissos (T Y): 2, 3, 5. Row 34. Vrysinas (V R Y): 1, blank, 1. Row 35. Zakros (Z A): 4 591, 595. Row 36. Zominthos (Z O): 1, blank, 1. Row 37. Total: 178; 2,265; 2,443.

Last but not least, where is this evidence stored? And, more interesting for the reader, where can we see this material? The answer is an easy one: museums. Most of the Linear A evidence is stored in Greek museums (especially Cretan), although a few inscriptions are also on display elsewhere. Let us start with Greece. The bulk of the material dug out on Crete is on display at the Archaeological Museum of Heraklion, followed by the Archaeological Museums of Khania, Petras/Siteia and Haghios Nikolaos. In the Aegean islands, some Linear A is at the Archaeological Museums of Thera (Santorini), Kythera and Kea. In Mainland Greece, evidence is on display at the National Archaeological Museum of Athens, the Archaeological Museum of Mesara and the Piraeus Museum. Outside Greece, a few inscriptions can be found at the Museo Pigorini (Rome) and at the Ashmolean Museum (Oxford).Footnote ³²

However, it is not necessary to travel long distances to see some Linear A. In the contemporary digital age, effort is being put into making digital collections available online, although we are still at the very outset of this undertaking (a list of currently available resources is given in Section 10). Digital projects aside, the two standard scientific editions of Linear A inscriptions are Raison and Pope’s (Reference Raison and Pope1994) Corpus transnuméré du linéaire A and Godart and Olivier’s (Reference Godart and Olivier1976–85) Recueil des inscriptions en Linéaire A, in five volumes (known as GORILA, with scans available online at http://cefael.efa.gr/result.php?site_id=1&serie_id=EtCret, Études crétoises, 21). Although the latter is now the most widely perused for its photographic set, the former is still valuable in providing detailed archaeological and bibliographical references. A few discrepancies aside, the two editions mostly differ in their respective numbering system of Linear A signs. The five volumes of GORILA contain all Linear A evidence published until 1976. Later finds are published in a Supplement (GORILA Supplement 1, Del Freo and Zurbach 2025; announced in Del Freo and Zurbach 2011), with a few finds already published in specialist journals.

Unlike the aforementioned digital projects, GORILA is aimed at a specialist audience only and is not (strictly speaking) user-friendly. For every inscription, a black-and-white photograph is given, alongside a drawing and a transcription of the Linear A signs it shows in their ‘standardised/normalised’ shape. However, no phonetic transcription is given, meaning that inscriptions cannot be ‘read’ in a phonetic fashion. But we shall not despair: on his Academia.edu collected papers (from the former website Linear A Texts in Phonetic Transcription & Commentary, https://bit.ly/4c4fSLo, University of Kansas), John Younger (Emeritus Professor) provides a phonetic reading of inscriptions by applying to Linear A signs the conventional phonetic values of their corresponding Linear B signs. By moving back and forth between GORILA (visual information) and Younger’s transcriptions (textual information), we can come up with a good overview (and reading) of the extant Linear A evidence (see Section 7). A new digital, interactive and updatable edition of the corpus is a desideratum in the field. It is hoped this wish will soon be granted to benefit specialists and general readership alike.

4 Drawing Lines: What Does Linear A Look Like?

Straight, curved, segmented, dotted – Linear A is composed of various ‘lines’ (Figure 3). When combined together, lines form ‘signs’, the basic graphic units of a script. The standard list of Linear A signs (GORILA V: xxii–xxvii) shows a total of 390 signs. Yet these signs are not all of the same type. These can be classified based on either form or function: if the former, we end up having ‘simple’ versus ‘complex’ signs; if the latter, we have ‘syllabograms’ versus ‘ideograms’ (on terminology, see Section 1). Scripts are, in fact, very much like coins, showing two different yet inextricably linked sides: one side representing their graphic (visual) ‘form’, the other their ‘function’. When talking about ‘form’, we mean what the script (and its signs) looks like on a visual level (sign shapes, ductus, etc.), whereas ‘function’ refers to the function performed by signs to convey a specific type of information (e.g. phonetic, ideographic, numerical, structural, etc.).

Figure 3 What a Linear A inscription looks like (ZA 10a, GORILA III: 168).Footnote ³³

We have seen (in Section 1) that Linear A has both phonetic and non-phonetic components: roughly 150 syllabograms (phonetic signs) and almost 200 ideograms (non-phonetic signs, standing for words or concepts). This way, signs are classified following a functional approach (which is the one adopted for Linear B sign classification). Alongside syllabograms and ideograms, there are just under fifty signs (‘klasmatograms’) representing mathematical values (not speech), more specifically fractions (GORILA V: xxvii). At times, signs occur in isolation (i.e. they are neither part of a word nor ideograms), often placed within dots: these are called ‘isolated signs’, which are standardly subdivided into the categories of ‘transaction signs’ and ‘simple signs’ (more on this in what follows). There are also signs for numerical units (although not included in the standard sign list): a short vertical stroke represents one unit, a short horizontal stroke is a ten, a circle is one hundred. Finally, Linear A seems to have one punctuation mark, which comes in the shape of a small dot (not consistently used in inscriptions) and isolated signs (Figure 4).

Figure 4 Functional classification of Linear A signs (HT 14, GORILA I: 28–9): syllabograms (pink), ideograms (blue), transaction sign (green), numbers (orange), punctuation marks (red).Footnote ³⁴

Figure 4Long description

What follows is a description of the categorisations without using colour. The syllabograms are the first two characters on the first line and the first four on the third line down. The ideograms are the fourth character on the first line, and the character after the horizontal dashes on the first line. The transaction sign is the third character on the first line. The numbers are the horizontal and vertical dashes throughout. The punctuation marks are the dots in the first line around the third character.

This functional approach to Linear A sign classification (much influenced by Linear B sign classification) is coupled with a formal approach. In fact, GORILA’s standard sign list does not group signs based on whether they function as either syllabograms or ideograms (which is at times a controversial choice to make) but rather groups them based on their graphic form: ‘simple’ versus ‘complex’. However useful, this double approach to Linear A sign classification may turn out to be slightly confusing. ‘Form’ and ‘function’ are so deeply intertwined that even in standard scholarship there is no single way of classifying Linear A signs (see the discussion in Salgarella Reference Salgarella2020: 42–149; Salgarella Reference Salgarella2022b). Let us look at these two aspects one by one to get a better understanding of what Linear A looks like and how it functions.

On a graphic level, signs are either ‘simple’, meaning they consist of one single graphic unit, or ‘complex’ (or ‘composite’), meaning they are a combination of two or more simple signs (thus being a complex graphic unit). Based on how simple signs are combined together, we can have different types of complex signs: ‘ligatured’ complex signs result from fusing together two or more simple signs (sharing a number of traits, e.g. A 559 in Figure 5), while ‘juxtaposed’ complex signs are formed by simply juxtaposing two simple signs (e.g. A 608 in Figure 5). Sometimes it is unquestionably challenging (or not possible at all) to assign a complex sign to either of these two categories, as it could equally fit in both. Let us not forget that these classifications are modern conventions: we do not know how Bronze Age writers conceptualised Linear A signs (see also Section 9.3). These classifications are just ways for us to make sense of something whose underlying principles we do not yet fully understand. We shall thus never forget that the applicability and effectiveness of this taxonomy has its own limitations.

Figure 5 Formal classification of Linear A signs: ‘simple’ versus ‘complex/composite’.Footnote ³⁵

Sometimes complex signs represent ‘monograms’, whose individual components spell out a word. Wait, how do we know this if Linear A is undeciphered? We know this thanks to a very fortunate piece of evidence allowing us to read and translate a Linear A monogram. This is complex sign A 559, reading (bottom-up) as MA+RU, if signs are given the phonetic values of their Linear B counterparts (on Linear A ‘readability’, see Section 7). This word is thought to be the term for ‘wool’ in Minoan. In fact, not only was this complex sign retained in Linear B and used as the ideogram for wool (Linear B sign *145/LANA), but also the term maru is likely to have been borrowed into Greek as μαλλός /mallós/ ‘wool/fleece’.Footnote ³⁶ Do we have more such examples? Not quite. Whether more (or most/all?) Linear A ligatured signs may represent words (and therefore be given a phonetic reading) is still unclear.

It is now time to become more familiar with the standard sign list in order to understand how to ‘find’ a Linear A sign and how to cite it in writing. In GORILA’s sign list, signs appear in order of increasing graphic complexity: first come simple signs (‘signes simples’, GORILA V: xxii–xxiii), then complex signs (‘signes complexes’, GORILA V: xxiv–xxvii), followed by all attested combinations of fractional signs (‘fractions simples et complexes’, GORILA V: xxvii).Footnote ³⁷ Each sign is given a number in ascending order (starting from 01), preceded by a prefix: ‘A’ if the sign is only attested in Linear A, while ‘AB’ if the sign is shared with Linear B (note that not all Linear A signs were continued onto Linear B). In this latter case, it is often thought that the shared sign (‘homomorph’) had a comparable phonetic reading (‘homophone’) in Linear A. Given that the phonetic values of Linear B are known, it is quite common to supply an approximate phonetic reading to an ‘AB’ sign in Linear A, alongside its classification number. Hence, here is what a Linear A sign’s ID looks like: AB 01/DA (AB sign, with approximate phonetic reading given, i.e. /da/), A 301 (Linear A–only sign). In case a sign also behaves as an ideogram, whose meaning is known from its Linear B counterpart, the Latin name of the commodity the sign stands for is also given (following Linear B conventional nomenclature):Footnote ³⁸ for example, ‘AB 21 QI/OVIS’, where OVIS is the Latin word for ‘sheep’, meaning that the sign is also used as the ‘sheep’ ideogram. In case of animal ideograms, the gendered version is specified with a superscript: for example, OVIS^f ‘female sheep’ (ewe) versus OVIS^m ‘male sheep’ (ram). The complete list of Linear A signs (alongside the ideograms’ Latin names) is given in the supplementary online appendix (provided in the Contents section at the beginning of this Element).

Moving onto ‘function’, we have seen that a functional classification groups signs based on which function they perform in an inscription (or text, more generically): syllabograms perform a phonetic function, ideograms stand for ideas or concepts, klasmatograms represent mathematical values. On top of these, sometimes Linear A inscriptions show isolated signs, whose function is not straightforwardly identifiable with any of the above. So, how do we determine which function a sign performs? Mostly by context. In fact, it is not possible simply to guess at a sign’s function based on its graphic/formal appearance only: simple signs may behave both as syllabograms and ideograms (despite statistically showing a preference for the former); composite signs prefer an ideographic behaviour, yet we have seen that ‘monograms’ spell out words and therefore can be read phonetically.

To determine which function a sign performs with more certainty, we look at where it appears: if a sign is part of a sign-sequence and is therefore used to write what we assume is a word (this is not always obvious in Linear A), it behaves as a syllabogram; if a sign occurs alone after clearly identifiable sign-sequences (mostly at the end of an entry), it behaves as an ideogram. Isolated signs are identifiable by way of exclusion, as they do not belong in either of the previously mentioned contexts: as their name tells, they are ‘isolated’, associable with neither a precise sign-sequence nor an ideogram.Footnote ³⁹ At times isolated signs are placed between dots (the Linear A ‘punctuation’ mark), as if to indicate to readers, by standing out from the rest of the text, that they refer to ‘something else’ (see the black dots on the first line of Figure 4). The problem is that we do not know what exactly this ‘something else’ is. It has been suggested (Schoep Reference Schoep2002: 37–9, 135–43) that isolated signs may sometimes refer to the type or nature of the transaction recorded in the inscription (as most Linear A texts are administrative records; see Section 6); in such cases, they are called ‘transaction signs’ (see Section 8.2).

In sum, we have seen that neither a fully formal nor a fully functional classification is a productive way of classifying Linear A signs. Both classifications are in fact needed and useful based on which specific aspect of the script we want to focus on and explore. Both ‘form’ and ‘function’ always need to be borne in mind, given their inextricable intertwining. However challenging studying Linear A may be, we shall refrain from ‘flipping a coin’ when undecided: let us always take into account both sides of the ‘Linear A coin’, making them interchange in a productive way.

5 Of Clay and Stone (and Else): Where Does Linear A Appear?

This section focuses on the materiality of the Linear A script (see esp. Flouda Reference Flouda, Piquette and Whitehouse2013, Reference Flouda2015). What do we mean by ‘materiality’ in this context? A wide-ranging term, ‘materiality’ refers to the materials (e.g. clay, stone) used to shape the objects meant to accommodate writing (standardly called writing ‘supports’ or ‘media’), as well as the implements used to write inscriptions (e.g. brush, stylus).Footnote ⁴⁰ Based on the medium and implement chosen, which provide material affordances and constraints (see e.g. Salgarella in press), different techniques can be adopted to write inscriptions. As a modern example, let us take writing on paper (medium): with a pen (implement), the technique is inking; with a pencil, it is drawing; with a brush, it is painting. The same applies to Linear A: on the same medium (clay), an inscription can be incised (with a stylus), painted (with a brush), or carved (with carving tools). In Aegean scholarship, inscriptions are often referred to as ‘written documents’, which can be classified based on function and typology (see Section 6).

In the Linear A context, the material aspects of an inscription and the function it performs very often go hand in hand and are deeply intertwined, although there is no strict one-to-one correspondence between writing medium and function (see Section 6). For instance, clay is the preferred medium for administrative inscriptions (e.g. records of economic transactions), but it can also be used to accommodate inscriptions fulfilling other purposes (e.g. cultic, ritual, dedicatory). So, how much material diversity is there in Linear A? We could say that Linear A inscriptions show reasonably broad material diversity, if compared to the other full-fledged Bronze Age Aegean scripts (i.e. Cretan Hieroglyphic, Linear B, Cypro-Minoan). However, Linear A inscriptions are much less diverse if compared to other ancient contexts (e.g. Egypt, the Near East). Therefore, also in this case (as often when discussing Linear A), answering this question is a matter of perspective.

Now, we may wonder whether the extant material diversity (presented in the following sections) is representative of the overall Linear A production or if it is just a snapshot of what might have existed and is forever lost to us: is this a real or an apparent ‘richness’? In other words, is it that not much has survived to us or that not much was produced? This is hard to evaluate on present evidence. Overall, Linear A does not appear to have been widely employed (as much as we use ‘writing’ nowadays, or if compared to nearby contemporary contexts such as Egypt and the Near East), since the extant Linear A evidence comes from a number of specific contexts only (administrative or cultic). The function of Linear A ‘written documents’ is the subject of Section 6, while Section 9 is dedicated to ‘what is missing’ (and ‘what we may expect’) from the archaeological record.

In what follows, an illustration is given of the material supports/media of Linear A inscriptions, their writing implements and writing techniques, which will set the foundations for discussing document typology and function in Section 6.

5.2 Writing media

5.2.1 Supports

A great variety of writing media (i.e. objects receiving writing) were produced to accommodate Linear A texts. The most common writing media are made out of clay, to such an extent that the Linear A writing practice may be argued to follow a clay+ trend. Clay was shaped into different objects (‘document types’, most notably tablets and sealings; see Section 6) with one or more surfaces to accommodate writing. No wonder why clay was chosen as a preferential material: clay is abundant on Crete, uncostly and reusable until baked. As we will see in Section 6, the majority of inscriptions on clay appear on small rectangular tablets and a variety of sealings, incised with a stylus (or comparable pointed tool).

Both tablets (Section 6.1.1) and sealings (Section 6.1.2) functioned as records of economic transactions for the administration of palatial centres. Lending itself to being reworked at need, clay allowed for flexibility of information recording until the written documents were stored in temporary (likely yearly) deposits (or ‘archives’).Footnote ⁴¹ Never intentionally baked, these documents outlasted their intended lifespan due to accidental fires during conflagrations (see Section 2). However, there are also examples of intentionally baked clay inscriptions (fewer in number): these are clay vessels (either carved or painted), mostly used as long-term storage containers (Section 6.1.3).

Clay aside, less common writing media are (Section 6.2) stone vessels (of varying sizes and shapes), metal jewellery (gold, silver, bronze) such as hairpins and rings, seals in soft stone (e.g. serpentine), wall plaster and architectural blocks. It is worth noting that although clay tablets and sealings were produced solely for written recording (and thus to receive writing or stamped seals), some other classes of objects (e.g. stone vessels, large clay storage jars called pithoi) were also produced (in the majority) without inscription. Writing was therefore applied to these latter classes of objects to make them distinctive in some way.

5.2.2 Classification System

Linear A inscriptions are classified based on support. In the conventional classification system, each writing medium is assigned an abbreviation (corresponding to its ‘class designation’) to ease identification of inscriptions’ supports at first sight (without requiring access to images). Class designation is preceded by an abbreviation of the inscription’s findplace (e.g. HT = Haghia Triada, KH = Khania)Footnote ⁴² and followed by a unique classification number (reference number). The combination of these elements gives us a unique inscription’s ‘ID-card’, used both to retrieve the inscription in edited corpora and reference it in publications.

Table 3Writing media classification system⁴³

Table 3Long description

The table has 3 columns: Designation, Document Type, and Example. Designation is further divided into Class and Subclass. It reads as follows. Row 1: blank; blank; Clay tablets (pageshaped, bars, lames); H T 100.

Under Class W - Clay sealings: Row 2: W a Nodules (single-hole hanging nodules) / noduli; K H W a 1001. Row 3: W b; Sealings/flat-based nodules; K N W b 33. Row 4: W c Roundels; K H W c 2001. Row 5: W d; Two-hole hanging nodules; H T W d 1617. Row 6: W e; Dome noduli; S A M W e 4. Row 7: W g; Miscellaneous ‘roundels’; P H W g 45. Row 8: W y; Miscellaneous nodules; P H W y 42.

Under Class Z – Vessels and other supports: Row 9: Z a; Stone vessels; I O Z a 2. Row 10: Z b; Clay vessels with incised inscriptions; K E Z b 4. Row 11: Z c; Clay vessels with inked/ painted inscriptions; K N Z c 6. Row 12: Z d; Graffiti; H T Z d 155. Row 13: Z e; Architecture; K N Z e 45. Row 14: Z f; Metal objects with incised inscriptions; K N Z f 13. Row 15: Z g; Stone objects; C R Z g 4. Row 16: Z h; Ivory objects; K N Z g 57−8.

Despite this material variety, it should always be borne in mind that, statistically speaking, Linear A inscriptions are not distributed evenly across all writing media. In other words, not all the writing media listed above bear an equal number of inscriptions. The great majority of inscriptions (96 per cent) occur on five media only: single-hole hanging nodules (ca. just under 900), tablets (ca. 450), roundels (ca. 150), clay vessels (ca. 70) and stone vessels (ca. 50) (Decorte Reference Decorte, Ferrara and Valério2018b: 20–2). Last but not the least, in addition to the extant writing media and materials, we may also infer the existence of other writing supports that have not survived to us (but are likely to have existed). In fact, we do have indirect evidence for the use of perishable materials such as parchment and perhaps also papyrus (see Section 9).

5.3 Writing Techniques and Implements

Writing techniques and implements are contingent upon writing media, which thus represent a physical constraint on making inscriptions and influence their visual outcome. Clay can be either ‘incised’ or ‘painted’. Most Linear A clay inscriptions are incised with a stylus (perhaps of metal, ivory, reed or arguably also a thorn); a few are painted with a brush. Neither implement has survived in the archaeological record (there are, however, examples of styluses from the Mycenaean period).Footnote ⁴⁵ Clay can also be ‘stamped’ with seals (although very few seals are also inscribed; see Section 6.2.5). The other durable writing media bearing Linear A inscriptions (stone, metal, ivory) are ‘carved’, most likely with chisels or finer cutting tools (e.g. knives, burins, awls, blades), for which there is no uncontroversial archaeological evidence either.Footnote ⁴⁶ The combination of medium, implement and technique (Table 4) brings forth different Linear A ‘writing styles’, resulting in ‘writing traditions’ (Schoep Reference Schoep2002: 13): a ‘pinacological tradition’ (clay inscriptions), an ‘epigraphic tradition’ (inscriptions on other durable materials), a ‘papyrological tradition’ (inscriptions on perishable materials). The latter, showing a more cursive style of writing, has barely survived (painted clay inscriptions)Footnote ⁴⁷ but is likely to have existed (see Section 9).

Table 4Synoptic overview of Linear A writing media, techniques and implements

Table 4Long description

The table has 2 main columns: Writing media and Inscription. Writing media is divided into: Evidence, Material, and Document type. Inscription is divided into: Technique and Implement. It reads as follows.

Row 1: Direct evidence: Preserved; Clay; Tablets; Incised; Stylus?. Row 2: Direct evidence: Preserved; Clay; Sealings; Incised, stamped; Stylus?. Row 3: Direct evidence: Preserved; Clay; Vessels; Incised, painted; Brush, stylus?. Row 4: Direct evidence: Preserved; Stone; Vessels; Carved; Carving tools. Row 5: Direct evidence: Preserved; Stone; Seals; Carved; Carving tools. Row 6: Direct evidence: Preserved; Metal; Jewellery; Carved; Carving tools. Row 7: Direct evidence: Preserved; Bone; Fragment, disk/ sceptre; Carved; Carving tools. Row 8: Direct evidence: Preserved; Plaster; Mural graffiti; Incised; Stylus?. Row 9: Indirect evidence: Likely existed; Parchment/ leather; Folded sheets; Painted (ink?); Brush?. Row 10: Inferable evidence: May have existed; Papyrus; Rolls; Painted (ink?); Brush?.

6 (Beyond) Accounting: What Was Linear A Used For?

Studying Linear A necessarily means studying ‘inscriptions’, but what exactly is an ‘inscription’? By convention, an inscription is a sequence of at least two or more signs, forming a word. This definition, however, may be too restrictive as it does not encompass the category of ‘single signs’ (i.e. signs occurring in isolation; e.g. potmarks, mason’s marks, transaction signs, etc.; GORILA I: xi clearly excludes these categories). In Aegean scripts, single signs are not unusual, and at times a single sign may function as the abbreviation of a word (e.g. Linear B sign AB 61, with the phonetic value /o/, stands for the word o-pe-ro ‘deficit’). Therefore, single signs (especially when occurring in isolation, like potmarks) may well deserve to be upgraded to ‘inscriptions’. This circumstance has recently made scholars challenge the traditional definition of ‘inscription’ provided previously in favour of a more comprehensive one including both single-sign and multi-sign texts (Donnelly Reference Donnelly, Steele and Boyes2022: 63–5). Most Linear A inscriptions are multi-sign texts. There are, however, instances of single signs occurring in isolation (e.g. potmarks, mason’s marks, transaction signs). Given that Linear A is undeciphered, it is not always easy to understand whether a single sign abbreviates a word (and therefore has a phonetic meaning) or functions as an ideogram (standing for a concept).

In traditional scholarship, Linear A inscriptions are classified as either ‘administrative’ or ‘non-administrative’, thus revealing a (dichotomic) functional approach to taxonomy. For ‘administrative’ inscriptions, we take all those which come from clearly definable administrative contexts (palatial centres, villas, etc.) and were used to record various kinds of economic transactions. By contrast, all those falling outside this category are taken as ‘non-administrative’ inscriptions. Needless to say, defining by negation is never a good practice: if the function of an ‘administrative’ inscription is to record a transaction, then what precisely is the function of a ‘non-administrative’ inscription? There is no univocal answer – and no short answer either. Before discussing what these inscriptions are, let us first look into how (and why) this distinction came about.

First, a quantitative bias most likely played a role: administrative inscriptions significantly outnumber non-administrative ones, and all come from more or less homogeneous contexts, whereas non-administrative inscriptions show considerable heterogeneity in terms of both writing media and contexts. This circumstance had the effect of arranging into a (perhaps unintentional?) hierarchy Linear A inscriptions, prioritising administrative ones over all others: as a result, the latter were squeezed into a less clear-cut category (‘non-administrative’). ‘Little’, however, does not equal ‘belittle’: contemporary users of the Linear A script may not have shared our same perception and hierarchy of uses of the objects bearing Linear A writing. Second comes a qualitative bias. The systematisation of Linear A as we know it (script, signary, document types, functions, etc.) is much indebted to Linear B: once Linear B was deciphered (1952) and systematised, the same interpretative (and taxonomic) framework was retrospectively applied to Linear A (see esp. Salgarella Reference Salgarella2020: 25–32). Linear B inscriptions are almost exclusively administrative to the point that, compared to Linear A, Linear B is often said to show a restricted context of use (i.e. administrative-only). This, as we have seen, does not necessarily hold true for Linear A and is further proof that an ‘administrative’ versus ‘non-administrative’ classification of Linear A evidence is ill-fitting. It has to be borne in mind that in most cases we still see Linear A through the lens of Linear B.

Any classification therefore has to be taken just as a heuristic tool to systematise the extant evidence, and to identify and evaluate patterns. Here the decision has been made to present the rich heterogeneity of Linear A inscriptions based on document type, therefore following a typological classification giving justice to (and accounting for) both the function performed by an inscription and its material aspects (without necessarily implying a functional hierarchy). Inscriptions on clay will be presented first, given that clay is the most common writing support/medium (see Section 5), followed by inscriptions on a variety of other supports.

6.1 Inscriptions on Clay

Inscriptions on clay represent 90 per cent of the extant Linear A corpus, and their function is primarily ‘administrative’.

6.1.1 Tablets

Tablets are the most well-known category of ‘administrative’ documents (a list of findplaces is given in Section 3, Table 2). Tablet deposits (or archives)Footnote ⁴⁸ have been found on Crete at Hagia Triada, Khania, Zakros, Phaistos, Knossos, Malia, Archanes, Tylissos, Palaikastro, Gournia, Papoura, Petras and Pyrgos, on the Aegean islands at Akrotiri (Thera), Hagia Irini (Kea) and Phylakopi (Melos).

Typologically, most Linear A tablets are ‘page-shaped’ (e.g. HT 118, GORILA I: 200–1): small, rectangular documents fitting into the palm of a hand, often inscribed on both sides (hence called ‘opisthographs’: e.g. ZA 10a–b, GORILA III: 168–71).Footnote ⁴⁹ Several tablets are also ‘palimpsests’ (esp. tablets from Haghia Triada, e.g. HT 115, GORILA I: 188–91), meaning that an original inscription was subsequently erased for another one to be superimposed (traces of erased signs are sometimes visible; e.g. HT 115a, lower half, HT 86b, lower half). This suggests that Linear A tablets were often reused (with the older text perhaps copied onto another document for long-term recording, before its being erased to accommodate new information) and thus meant to be temporary records, which survived to us only because of accidental fires (see Section 2). There are also a few examples of ‘oblong tablets’ and ‘bars’ (the latter being a document type typical of Cretan Hieroglyphic administration): these only come from Knossos, Phaistos and Mallia (e.g. MA 1a–c, GORILA I: 268–9), and all predate the Late Minoan I period (Middle Minoan IIB–IIIB), to which the majority of Linear A evidence is dated (see Section 2).

Tablets were used as records of economic transactions (‘recording documents’) to keep track of the inflow and outflow of goods within palatial administrations. Their content could be either miscellaneous (multiple commodities) or specialised (single commodity) (for a classification based on format types, see Schoep Reference Schoep2002: 67–88; for Haghia Triada tablets, see Montecchi Reference Montecchi2019). Among the main basic commodities that we can identify with certainty (through ideograms),Footnote ⁵⁰ there are grains (AB 120/GRAnum: grain, probably barley; A 303: grain, probably wheat;Footnote ⁵¹ AB 65/FAR: spelt), olives (AB 122/OLIVa), olive oil (A 302/OLEum), figs (AB 30/FICus), wine (AB 131/VINum), vessels (A 400^VASum–418^VASum), wool (A 556-563/LANA), textiles (AB 54/TELA, AB 164), animals (livestock: AB 23/BOS; sheep: AB 21/OVIS; goats: AB 22/CAPer; pigs: AB 85/SUS), as well as allocations of foodstuff to personnel (AB 100/VIR; Schoep Reference Schoep2002: 176–89).

Three main types of transactions can be expected to be represented in the Linear A tablets and are at times positively identified (Schoep Reference Schoep2002: 90–1). These are (i) ‘inventories’ (and ‘assessments’) of produce/products (e.g. foodstuff, raw materials, finished products) stored in magazines and workshops located within (or in the vicinity of) the palatial centre, or delivered to the centre from the surrounding areas (e.g. goods produced there and/or information on flocks pastured on them); (ii) ‘contribution or collection records’ booking the inflow of goods and the contributions made to the central administration (e.g. taxes, levies, deliveries, etc.; although these are not often straightforwardly identifiable); (iii) ‘allocation or distribution records’ booking the outflow of produce/products (e.g. agricultural commodities, goods, animals, raw materials) from the central administration to different parties (whose purposes may vary: e.g. food rations, payments, etc.; of equally challenging straightforward identification).

Linear A tablets, recording lists of commodities, show a textual structure characterised by an ‘opening heading’ (usually at tablet-start, but at times headings are placed also elsewhere on a tablet), introducing a section with each entry consisting of one or two words (probably personal or place-names) followed by an ideogram, numbers and fractions. ‘Transaction signs/words’ (e.g. KI-RO ‘deficit’, A-DU ‘assessment’; see Section 8.2) and ‘simple signs’ (of unclear function) can also occur in the text (see Sections 7.1 and 8.2). This basic (theoretical) textual structure may take different (pragmatic) forms based on which elements are present on a contextual basis (see the example in Figure 6; see also Section 7.1) and shows considerable cross-site variation. In other words, despite the overall structural similarities, there is no standard textual structure/format that may be applied to all Linear A tablets.

Figure 6 Example of a Linear A tablet’s textual structure (PE 1).Footnote ⁵²

Figure 6Long description

What follows is a description of the image without using colour. The opening heading refers to the first line of the tablet, plus the first character of the second line. The heading refers to the next three characters on the second line. The next character is the logogram and the dashes are the numbers. On the third line, the logogram is the first character, and the horizontal and vertical dashes, plus the hairpin symbol, are the numbers. The next two characters are the heading. On the fourth line, the first character is the logogram. The horizontal and vertical dashes are the numbers. The next character is the logogram and the dashes on the rest of this line, and on line five, are the numbers.

Tablets’ textual structure (format types) may give us insights into the process of information recording and related administrative procedures. Some format types may in fact reflect consecutive stages in the information processing system, while other types seem to reflect a specific type of information and are thus likely to have been compiled for different purposes (e.g. recording the purpose of the booking or the administrative status of the commodities; see esp. Schoep Reference Schoep2002: 86–7). It is likely that at the end of the administrative cycle, the information contained on tablets was transferred onto documents of perishable material (see Section 9), functioning as long(er)-term records (to which single-hole hanging nodules may have been attached; see Sections 6.1.2 and 9.1).Footnote ⁵³

6.1.2 Sealings

Sealings are small lumps of clay bearing one or more seal impressions and (at times) writing. They come in a variety of shapes, which can be grouped into four main categories: roundels, noduli, flat-based nodules and hanging nodules (further subdivided into single-hole and two-hole; Hallager Reference 66Hallager1996, I: 21–4, figure 2). The first two categories are ‘active’ sealings (free-standing documents) playing a role in the transmission of goods, while the latter two categories are ‘passive’ sealings (auxiliary documents) fastened to goods or written documents (Schoep Reference Schoep2002: 193). Although the precise function(s) of these document types are not yet fully understood, we have a working knowledge of their use (Hallager Reference Hallager and Palaima1990, Reference 66Hallager1996, Reference Hallager and Müller2010, Reference Hallager2021; Schoep Reference Schoep2002: 192–9; Tomas Reference Tomas, Sacconi, Freo, Godart and Negri2008, Reference 74Tsikritsis, Karetsou, Theocharis and Kalokairinos2012; Finlayson Reference Finlayson, Piquette and Whitehouse2013, Reference Finlayson, Μιτσοτάκη and Τζεδάκη-Αποστολάκη2018, Reference Finlayson, Relaki and Driessen2020; Montecchi Reference Montecchi2017, Reference Montecchi, Μιτσοτάκη and Τζεδάκη-Αποστολάκη2018).

Roundels (small clay disks; e.g. GO Wc 1, GORILA II: 2) and noduli (small clay lumps; e.g. HT We/Wc 3020, GORILA II: 78; HT We/Wa 1020–1, GORILA II: 7) were not meant to be fastened to other items (unlike the other three categories of sealings) and are understood to have worked as independent documents meant to authorise transactions. Roundels functioned as receipts for outgoing or incoming goodsFootnote ⁵⁴ (reflecting the ‘internal’ movement of goods within each palatial system): it has been argued that the number of seal impressions on a roundel’s rim may specify the quantity of the commodity recorded in the inscription occurring on one or both of its sides, with each impression representing one unit (Hallager Reference 66Hallager1996: 100–1, 113). The seal user would be the recipient of the goods, certifying with the seal impression the units taken (Hallager Reference 66Hallager1996: 116). Since noduli carry one single seal impression, their purpose seems to have been to authenticate one transaction or certify an activity. It has been suggested that noduli worked as receipts for work done or tokens to be exchanged for goods or services (Weingarten Reference Weingarten1986: 18; Reference Weingarten1990: 19–20).

Flat-based nodules (examples in Section 9) show a seal impression (and rarely an inscription) on the upper side and imprints of thread and fibre on underside, suggesting they were pressed against tightly folded parchment (or papyrus?), arguably in order to secure the integrity of the document and prevent unauthorised viewing. Hanging nodules (e.g. HT Wa 1242; GORILA II: 25) are understood to have been fastened to a string (moulded around its end if showing a single hole, along its length if showing two holes) tied to goods to be delivered to (or collected from) the palatial centre (thus reflecting the ‘external’ movement of goods, between the central administration and external parties). Hanging nodules show one (or two) seal impressions on their sides and one/two Linear A sign/s on another side. They most likely served as labels or tags, supplying details about the delivery’s content and/or parties involved, or production/destination places (Krzyszkowska Reference Krzyszkowska2005: 160; Montecchi Reference Montecchi2017; Hallager Reference 66Hallager1996: 197–9 for the suggestion that one-hole hanging nodules sealed papyrus documents; see also Section 9.1). Finally, in addition to these main categories, we also have ‘direct object sealings’ (e.g. HT Wb 2001, GORILA II: 70; PH Wb 36, GORILA II: 92): these are lumps of clay attached to objects (jars, wooden or wicker baskets, or over pegs securing chests or doors) and impressed with a seal to secure their content or integrity (the seal impression would identify the individual responsible for the action’ Krzyszkowska Reference Krzyszkowska2005: 28).

Regarding the Linear A information processing system, there is no compelling nor probative evidence to suggest that the information held on sealed documents was later transferred onto tablets (which instead happens in the Linear B administrative practice). Thus Linear A sealed documents and tablets are unlikely to reflect consecutive stages of information processing. Rather, they are likely to have acted as complementary, parallel sets of documents within the administrative system and, arguably, to reflect different administrative concerns: tablets appear to relate to obligations concerning agricultural commodities, whereas sealed documents relate to different kinds of transactions involving a different (perhaps wider) geographical scale, or matters of different administrative status (Schoep Reference Schoep2002: 197).

6.1.3 Clay Vessels

Inscriptions on clay vessels can be either incised (e.g. TY Zb 4, GORILA IV: 109) or, more rarely, painted (e.g. Knossos ‘conical cups’: KN Zc 6–7, GORILA IV: 118–25; images in Flouda Reference Flouda, Piquette and Whitehouse2013: 160; Reference Flouda2015: 78; Dimopoulou-Rethemiotaki Reference Dimopoulou-Rethemiotaki2005: 217). It is worth noting that the majority of incised inscriptions on clay vessels were made before firing, meaning that they were part of the vessel’s manufacture. Although most clay vessels were found in cultic contexts and are therefore likely to have fulfilled cultic/ritual purposes, there are a number of inscriptions on large storage jars (pithoi) that clearly indicate their content (e.g. showing the ideogram for ‘wine’).

6.1.4 Clay Figurines

Deposited in cultic contexts (mostly peak sanctuaries or temple repositories), figurines most likely performed a dedicatory function (see esp. Morris Reference Morris and Insoll2017; Peatfield and Morris Reference Peatfield, Morris, Relaki and Driessen2020), but only very few bear inscriptions (e.g. TY Zg 1, GORILA IV: 170; Flouda Reference Flouda2015: 70). Interestingly, the latest attestation of Linear A (ca. 1350 BCE) is a painted (cursive-looking) inscription (retrograde) on a figurine from Poros Herakleios (PO Zc 1; Dimopoulou et al. Reference Dimopoulou, Olivier and Réthémiotakis1993).

6.1.5 Functional Objects

Loom weights (made of clay) occasionally bear inscriptions (e.g. from Kythera: KY Zg 1, GORILA IV: 166).

7 ‘Elementary, My Dear Watson’: What Do We Know from Reading Linear A?

Although Linear A has not yet been deciphered, we can still make sense of the texts (to an extent and with approximation) thanks to several cumulative factors, by (i) analysing Linear A itself and (ii) comparing it with Linear B.

The contexts of use of Linear A documents (‘administrative’ or ‘cultic’ texts; see Sections 5 and 6) allow us to ‘predict’ (to an extent) which kind of information to expect. Internal analysis of Linear A textual structure (see Section 8) can help identify recurrent structures (and words), whose probable meaning may then be deduced based on context (given the short, formulaic nature of Linear A texts). Iconographic analysis (underpinned by comparison with real-world referents, coming from either the archaeological record or the natural environment) can help identify the commodities represented by ideograms.

Linear B offers insightful comparanda in terms of administrative and economic systems (largely continued from the earlier period of ‘Minoan’ administration). Linear A/B shared ideograms (see Schoep Reference Schoep2002: 94–135; Salgarella Reference Salgarella2020: 219–20, 225–6) give us a general understanding of the Linear A commodities recorded on tablets. Linear A/B shared personal names and place-names, along with terms understood to belong in the Minoan substratum, add up to our short ‘Pocket Minoan Dictionary’ (see Section 8).

For these reasons, we are currently in a position to ‘read’ Linear A texts to an extent and with some degree of confidence, despite not knowing the grammatical structures, as well as linguistic affiliation, of the Minoan language (see Section 8). Given that Linear A is ‘readable’ and the contents of its texts are understandable to the extent possible given the present evidence, we propose here a ‘reading⁺ approach’ to Linear A documents, allowing for a reading and tentative, plausible interpretation of the Linear A texts, without however supplying a literal ‘translation’ thereof.

7.1 Reading⁺ a Linear A tablet

As illustrative examples of Linear A tablets, we will look at three tablets from Haghia Triada, a site which has yielded the best-preserved evidence allowing for (relatively) substantial reading and interpretation of the preserved texts. These are HT 28 (Section 7.1.1), HT 120 (Section 7.1.2) and HT 121 (Section 7.1.3).

7.1.1 HT 28

One of the best-preserved documents, HT 28a–b (Figures 7–8), is a page-shaped tablet from Haghia Triada, dated to Late Minoan IB. It records mixed agricultural commodities and is inscribed on both sides with very similar texts. Hence, it is understood to be a ‘balance ledger’ tablet – that is, a transaction document recording contributions made to the centre (side a) and debts/deficits/future payments (side b). The original assessment (i.e. how much produce the centre had originally requested) can therefore be reconstructed by totalling contributions (side a) and outstanding debts (side b).

Figure 7 HT 28a–b.Footnote ⁵⁵

Figure 7Long description

Two Line drawings of the front (side a) and back (side b) of Linear A inscription (clay tablet) HT 28a-b. Line drawings are preceded by details about this document. Type: Page-shaped tablet (5.9 cm times 8.5 cm times 0.9 cm; drawing not to scale); Find-place: Haghia Triada, Villa (Room 72); Period: Late Minoan IB; Classification: Mixed Commodities Tablet (Schoep 2002: 81, Type One A; Montecchi 2010:34, Class Mb); Edited in: GORILA One: 52-3.

Figure 8 Textual structure of HT 28a–b.Footnote ⁵⁶

Figure 8Long description

On the right, it shows a textual structure of tablet HT 28a-b (from Haghia Triada). The tablet starts with an opening heading (of unclear interpretation) occurring in the first line, followed by individual entries (lines 2-6) listing commodities (grains, oil and wine). A number of entries are preceded by an additional word (probably a personal name or a reference to roles or professions). A transaction term (reading sa-ra subscript 2, of unclear interpretation), occurs on line 3, halfway through the text.

Although the language behind Linear A is not deciphered, we can still make sense (approximately) of the contents of the tablet, thanks to its textual structure (layout), comparative analysis with other administrative records and the ideograms it shows (giving us an overview of the commodities recorded; see also Section 6.1.1). To begin with, the same agricultural commodities are listed on both sides (ideograms, plain or ligatured with syllabograms) in almost a fixed order (which is a typical feature of Haghia Triada’s ‘mixed commodities tablets’; see also Section 7.1.3): grain (AB 120/GRA), olive oil (A 302/OLE), figs (AB 30/FIC), wine (AB 131a/VIN) and an unidentified commodity (A 304; Figure 8). We can ‘read’ and ‘interpret’ these ideograms as they were continued in Linear B (except for A 304).

Then, three words (in red, Figure 8) may be taken as transaction terms (see Sections 4 and 8): A-SI-JA-KA and SA-RA₂ (occurring on both sides) and U-MI-NA-SI (on side b). A-SI-JA-KA starts the text on both sides: although (most unluckily) this term is a one-off attestation in the known Linear A vocabulary, it may be argued to be a transaction term because of its marked position and repetition at text-start on both sides. A-SI-JA-KA is therefore likely to function as a ‘heading’ (see Section 6.1.1), referring to the whole record. SA-RA₂ (occurring on both sides: side a, line .3; side b, line .2) is likely to be a transaction term as it functions (mostly) as such on comparative evidence (Schoep Reference Schoep2002: 164–6). U-MI-NA-SI (on side b, split across lines .1 and .2) arguably belongs in this category, given its position after A-SI-JA-KA and between dots (where transaction signs usually occur). In fact, the mixed-commodities textual format is usually characterised (at least on Haghia Triada tablets) by a heading often followed by a transaction sign (or term) and a list of individual entries (sign-sequence + ideogram + numbers). U-MI-NA-SI has been suggested to mean ‘debt’ or ‘[s/he] owes’.Footnote ⁵⁷ But who are the parties involved, giving contributions and owing debts? Most arguably, these are hidden behind the remaining words (in pink, Figure 8): JA-QI^f (on both sides: side a, line .1; side b, line .4); VIR+KA, A-RU-DA-RA and I-TA-JA (side a, lines .4 to .6); PU-RA₂ and WI-DI-NA (side b, lines .3, .5). It has been suggested that these words may be women’s names: they all end in -A, arguably a feminine ending (Godart Reference Godart1984: 125; Schoep Reference Schoep2002: 113, fn. 81; Younger Reference Younger2024, Linear A Texts in Phonetic Transcription: Haghia Triada, under section HT 28). Finally, can we reconstruct the original assessment? If we assume that side a records actual disbursements while side b records deficits, by totalling their respective figures we would have thirty units (+ J E L₂ sub-units) of grain, twenty-five units of oil, ten units of figs, sixteen units of wine and six units of wine associated with VIR+KA (perhaps ‘porters’?; Godart Reference Godart1984: 125).

7.1.2 HT 120

Tablet HT 120 (Figure 9) was unearthed at the site of Haghia Triada and is dated to Late Minoan IB. HT 120 is classified as a ‘specialised commodity tablet’ (Schoep Reference Schoep2002: 82), meaning that it specialises in recording a single commodity (and/or its variants).

Figure 9 HT 120.Footnote ⁵⁸

Figure 9Long description

The line drawing is preceded by details about this document. Type: Page-shaped tablet (5.3 cm times 8.3 cm times 0.9 cm; drawing not to scale); Find-place: Haghia Triada, Casa del Lebete; Period: Late Minoan IB; Classification: Specialised Commodities Tablet (Schoep 2002: 82, Type Two; Montecchi 2010:3 3, Class Eb); Edited in: GORILA One: 204-5.

Tablets belonging in the ‘specialised commodity’ format type mostly focus on agricultural commodities, namely olive oil (A 302/OLE), olives (AB 122/OLIV), a grain (AB 120/GRA)Footnote ⁵⁹ and A 308 (non-identified commodity). Their textual structure is characterised by consecutive entries consisting of sign-sequences (i.e. words) followed by ideograms, numerals and fractional signs (at times words are omitted and only ideograms and numerals/fractions appear). HT 120 specialises in recording a grain (AB 120/GRA) and variants thereof (in the form of complex/composite signs constructed around AB 120). The textual structure of HT 120 is given in Figure 10.

Figure 10 Textual structure of HT 120.Footnote ⁶⁰

Figure 10Long description

What follows is a description of the image without using colour. Line 1 in the list refers to the first three characters on the first line of the tablet. Lines 1 to 2 in the list refer to the last two characters on the first line of the tablet as the statement and the first character on the second line as the commodity. Lines 2 to 3 in the list refer to the final character on line 2 of the tablet. Lines 3 to 4 in the list refer to the final two characters on line 3, plus the first two characters on line 4 of the tablet as the statement, then the third character on line 4 as the commodity. Lines 4 to 5 in the list refer to the final character of line 4 of the tablet, plus the first three characters of line 5 as the statement, then the next character on line 5 as the commodity. The first Line 6 in the list refers to the first character on line 6. The second Line 6 in the list refers to the three characters after the vertical dashes on line 6.

The textual structure of HT 120 shows the pattern ‘word + ideogram + numbers/fractions’ in all entries, which are arguably preceded by an ‘opening heading’ (see Section 6.1.1), namely DA-QE-RA (in bright red in Figure 10, line .1), the meaning of which is unclear (but perhaps a transaction term). The opening heading is graphically delimited by a punctuation mark (in the shape of a dot). The following entries are all introduced by a word (in pink in Figure 10): DA-ME (line .1), DA-U-*49 (split across lines .3-.4), KI-RE-TA-NA (split across lines .4-.5) and PA-I-TO (line .6). Of these, only PA-I-TO can be interpreted with some degree of certainty: this is most likely a place-name, referring to the site of Phaistos in central Crete.Footnote ⁶¹ By inference, we may suppose that the other words are also place-names. In particular, DA-U-*49 is argued to correspond to the place-name da-wo, often found in the Linear B tablets in association with Phaistos.Footnote ⁶²

Each one of these words (perhaps all place-names) is followed by one or more (complex/composite) ideograms featuring AB 120/GRA (grain):Footnote ⁶³ A 584 (line .2) and A 580 (line 5.) consists of AB 120/GRA followed by one or more fractional signs (conventionally transcribed as K, L₂, B; see the list of fractional signs in the online supplementary appendix, the URL for which is provided in the frontmatter), whereas A 574 (lines .2, .4, .6) ligatures A 120/GRA with AB 03/PA (whose two horizontal lines can be recognised in the stalk of AB 120/GRA). It has been suggested (Schoep Reference Schoep2002: 154) that AB 120/GRA may have functioned as a land measure (as it does in Linear B) when modified by fractions (in our case, A 584 and A 580), whereas ligatures of AB 120/GRA with a syllabic sign (e.g. A 574) may refer to specific aspects of the grain (e.g. ‘milled’, ‘raw’, ‘processed’, ‘fodder’, etc.). Ligature GRA+PA (= A 574) is always recorded in small quantities, and elsewhere (PE 1, given in Figure 6, Section 6.1.1) this produce is arguably allocated to personnel (AB 100/VIR) in half units (Schoep Reference Schoep2002: 106, 108). Hence, GRA+PA may perhaps stand for ‘processed’ grain (food ration for human consumption). Finally, the text of HT 120 shows a simple (or transaction) sign: this is AB 28/I (line .4; in dark red in Figure 10), preceded by a punctuation mark (in the shape of a dot). The meaning of AB 28/I in this context is unclear: if it works as a transaction sign, it may be an indication of the type of transaction (e.g. assessment, contribution, disbursement, etc.) being recorded.

7.1.3 HT 121

Tablet HT 121 (Figure 11) was unearthed at the site of Haghia Triada and is dated to Late Minoan IB.

Figure 11 HT 121.Footnote ⁶⁴

Figure 11Long description

Line drawing is preceded by details about this document. Type: Page-shaped tablet (4.1 cm times 6 cm times 1 cm; drawing not to scale; Find-place: Haghia Triada, Casa del Lebete; Period: Late Minoan IB; Classification: Mixed Commodities Tablet; (Schoep 2002: 81, Type One A; Montecchi 2010:34, Class Mc); Edited in: GORILA One: 204-5.

Like HT 28 (see Section 7.1.1), HT 121 is classified as a ‘mixed commodities tablet’ (Schoep Reference Schoep2002: 81), a format type characterised by a wide range of agricultural commodities booked in a (more or less) fixed order. Usually the order is AB 120/GRA and A 303 ‘grains’,Footnote ⁶⁵ A 302/OLE ‘olive oil’, AB 122/OLIV ‘olives’, AB 30/FIC ‘figs’ and AB 131a/VIN ‘wine’ (see also Section 6.1.1). Ideograms aside, the lexical information contained in these tablets is usually minimal (with HT 28a–b being an exception in this respect), as it is mostly limited to a heading, possibly followed by a transaction sign or a transaction term. The textual structure of HT 121 is given in Figure 12.

Figure 12 Textual structure of HT 121.Footnote ⁶⁶

Figure 12Long description

What follows is a description of the image without using colour. Line 1 of the list refers to the first three characters on line 1 of the tablet as the statement, plus the next two characters as the commodity. The first Line 2 of the list refers to the first two characters on line 2 of the tablet as the statement, plus the next character as the commodity. The second Line 2 of the list refers to the character in between the vertical dash symbols as the commodity. The first Line 3 of the list refers to the first character in line 3 of the tablet. The second Line 3 of the list refers to the second character in line 3 of the tablet (ignoring the dash symbols). The third Line 3 of the list refers to the final character on line 3 of the tablet.

The text shows two transaction terms (KI-RI-TA₂ on line .1 and SA-RA₂ on line .2; in red in Figure 12), each one followed by one or more entries of the type ‘ideogram + numbers’. These terms are therefore likely to function as ‘headings’. The word KI-RI-TA₂ is argued to be related to the transaction term KI-RO ‘deficit’ (see also Section 8.2) and to be a verbal form, perhaps meaning something like ‘is/are lacking’.Footnote ⁶⁷ As noted previously (Section 7.1.1), SA-RA₂ is likely to be a transaction term whose precise meaning is unclear, as it functions (mostly) as such on comparative evidence (Schoep Reference Schoep2002: 164–6). Most ideograms on HT 121 are plain: on line .2, AB 120/GRA ‘grain’ and A 302/OLE ‘olive oil’, and on line .3, AB 30/FIC ‘figs’, AB 131a/VIN ‘wine’ and AB 23^m/BOS^m ‘oxen’ (male ‘gendered’ variant of AB 23/BOS; see Section 4). These are all booked under SA-RA₂. The only complex/composite ideogram is A 623 (line .1), a ligature of AB 120/GRA ‘grain’ with syllabic signs AB 78/QE and AB 07/DI. The meaning of this composite sign is unknown, and note that this is the only commodity booked after KI-RI-TA₂. It has been argued that HT 121 (showing a text similar to HT 114) may be related to feasting activities, recording contributions for a communal feastFootnote ⁶⁸ (note that Linear B tablets often record commodities for ‘feasting/banqueting’). All commodities are, in fact, booked in whole units. On the assumption that ‘Minoan’ (Linear A) dry and liquid units are likely to equal ‘Mycenaean’ (Linear B) dry and liquid units,Footnote ⁶⁹ it can be calculated that the three units of wine booked on HT 121 (line .3) correspond to ca. 86.4 litres (1 liquid unit = ca. 28.8 litres). In turn, if we assume a consumption of ca. 0.25 litres per person (corresponding to two average Minoan ‘conical cups’), three units of wine would provide for almost 350 people.Footnote ⁷⁰

7.2 Reading⁺ the ‘Libation Formula’

The so-called libation formula is the longest Linear A text that has survived, which gives us some valuable insights into the morphology and syntax of the Minoan language (see Section 8). This text is only attested on stone (‘libation’) vessels, used in cultic (ritual, dedicatory) contexts and activities, and is called ‘formula’ because of its repetitive structure. The libation formula is therefore standardly classified as a non-administrative text (see Section 6), and is understood to contain a dedicatory sentence or prayer (Davis Reference Davis2013; Karnava Reference Karnava, Alram-Stern, Blakolmer, Deger-Jalkotzy, Laffineur and Weilhartner2016; Younger Reference Younger2024, Linear A Texts: Introduction, under section 12, ‘Libation Formula’). An illustrative example of an inscribed libation stone vessel is a ‘libation table’ (Figure 13), showing cup-shaped hollows into which liquid offerings were poured, and the ‘libation formula’ running along its edges.

Figure 13 Libation table from the Psykhro Cave showing the libation formula (PS Za 2, GORILA IV: 52–5).Footnote ⁷¹

The libation formula comes in two main versions, called ‘principal’ and ‘secondary’ (Karetsou, Godart and Olivier Reference Karetsou, Godart and Olivier1985: 134), showing a number of sign-sequences (words) in strict order (relative to one another). The principal version is the most common form, standardly showing six sign-sequences, whereas the secondary version is a shortened version consisting of three sign-sequences only. The two versions share one sign-sequence only, reading A/JA-SA-SA-RA-ME (with A- alternating with JA- at word-start on a contextual basis), which has a long history of use. It is attested as A-SA-SA-RA-NE in the Arkhanes Script (ca. 2000–1800 BCE) and Cretan Hieroglyphic (ca. 1900–1600 BCE), as A/JA-SA-SA-RA-ME in Linear A (list of shared attestations given in Civitillo Reference Civitillo2016: 165–6) and arguably also in the latest survival of Linear A writing (Poros figurine, ca. 1400 BCE; Dimopoulou et al. Reference Dimopoulou, Olivier and Réthémiotakis1993; see Section 6.1.4). This shared word is argued to be evidence of continuity in the cultural history of Crete throughout the second millennium BCE (Karnava Reference Karnava, Alram-Stern, Blakolmer, Deger-Jalkotzy, Laffineur and Weilhartner2016: 354). The other sign-sequences, instead, show contextual variation. After a comparative study of the sequences’ order, Davis (Reference Davis2013) puts forward compelling arguments as to how to interpret the formula’s syntax (see also the earlier study by Duhoux Reference Duhoux1992), thus suggesting the following readings for each sequence occurring in a fixed position (Table 5).

Table 5Interpretation of sign-sequences of the libation formula

First	Second	Third	Fourth	Fifth	Sixth
Verb (main)	Place name	Dedicant’s name	Object	Verb (subordinate)	Prepositional phrase

The longest version (‘principal’) of the formula shows two clauses starting with a verb (in the first and fifth positions). The first sequence is understood to be the main verb, constructed around a central root (I-*301) variably affixed (e.g. A-TA-I-301, A-TA-I-301-WA-JA, A-TA-I-301-WA-E, TA-NA-I-301-TI, TA-NA-I-301-U-TI-NU), arguably meaning ‘(s/he) gives/dedicates’. The second sequence is identified as a place-name (e.g. DI-KI-TE ‘Mount Dikte’, I-DA ‘Mount Ida’, TU-RI-SA ‘Tylissos’; see also Section 8.2), most likely the location where the dedication was offered. The third sequence is likely to be the dedicant’s name (subject of the main clause), as it always changes. The fourth sequence is A/JA-SA-SA-RA-ME, taken as the direct object of the sentence, probably meaning ‘dedication/offering’. The fifth sequence is argued to be a verb starting a subordinate (dependent) clause and giving a reason for the dedication, thus meaning something like ‘hoping for/requesting’. The sixth element is likely to be a prepositional phrase as it shows the suffix -TE ‘from’ (see Section 8.2). Hence a tentative interpretation of the standard version of the formula may be ‘gives/dedicates, at [place], [dedicant’s name], a dedication/offering, hoping for/requesting, [a favour?] from [a deity?]’.Footnote ⁷³ Not all texts of the libation formula show all the elements listed in Table 5 (which is therefore to be taken as a theoretical model), and a few texts show a couple of extra sign-sequences (eight in total). A good example of a well-preserved standard version of the formula is attested on IO Za 2, a libation table from Iouktas (GORILA V: 18–19), whose text and interpretation are given in Table 6.

Table 6Libation formula on IO Za 2⁷⁴

Table 6Long description

The table reads as follows.

Verb: a-ta-i-*301-wa-ja: ‘gives’. Subject: ja-di-ki-tu: (dedicant). Object: ja-sa-sa-ra-me: ‘(an) offering’. Subordinate clause: u-na-ka-na-si: ‘requesting’. Subordinate clause: i-pi-na-ma si-ru-te: ‘(a) favour’ ‘from (a deity)’.

The first three sequences (main verb, dedicant’s name and direct object) tend always to be present in all versions of the formula. For instance, in Figure 13 (PS Za 2), showing a shorter version of the formula, we can recognise the verb (TA-NAI-I-*301-TI), the dedicant’s name (]-JA-TI) and the object ‘an offering’ (JA-SA-SA-RA-ME). Given that the formula displays a Verb-Subject-Object (VSO) word-order, Davis (Reference Davis2013: 49) argues that there is a substantial chance for Minoan to be a Verb-Subject-Object (VSO) language (see Section 8).

8 Speaking in Riddles: Which Language Does Linear A Encode?

Although Linear B was successfully deciphered as Greek in 1952, its ancestor Linear A still defies us: which language does Linear A encode? Still hard to say. Scholars have long set out to crack the Linear A code by using a number of different approaches, no one of which has so far given uncontroversial results. Why does Linear A still resist decipherment? There are (unfortunately, we daresay) a number of objective obstacles to deciphering Linear A.

First comes the issue of quantity: the extant evidence is scanty (ca. 2,500 inscriptions; see Section 3, Table 2) and in a poor state of preservation (most inscriptions are fragmentary). By contrast, Linear B is much better placed, with more than 4,500 inscriptions, longer and overall better preserved. Second comes the issue of quality: Linear A texts are for the most part short and formulaic, showing very little syntax. As illustrated in Section 6, most Linear A ‘administrative’ documents are terse, brief records of economic transactions (looking like shopping lists); ‘non-administrative’ inscriptions (for cultic, dedicatory, ritual purposes) show slightly longer texts, which are, however, highly formulaic and repetitive. Although these latter texts are the longest we have and play a significant role in our appreciation of the Minoan language, they are neither extensive nor varied enough for a thorough linguistic investigation of the underlying, constitutive features of the Minoan language. Moreover, there appears to be no evidence of Linear A, in our surviving texts, outside of administrative and cultic/ritual contexts: there is no historiography, literature, diplomatic correspondence, monumental inscriptions or private letters written in Linear A (see Section 9). Last but not the least, no bilingual text has so far been found, neither within nor outside of Crete: this means that, on present evidence, there does not exist (perhaps yet?) a ‘Rosetta Stone’ for Linear A. Hence, overall there is not enough cumulative, probative evidence, in terms of both quantity and quality, for a successful, convincing and uncontroversial decipherment of Linear A.

In light of these factors, we may ask ourselves: what does it mean ‘to decipher’ a script? Which conditions are necessary successfully to crack a code? Short answer, ‘to decipher’ a script means to make sense of its written texts. Well, we can ‘make sense’ of Linear A text, to an extent and with an approximation (as seen in Section 7); yet, the language Linear A encodes still remains ‘undeciphered’. Script decipherment is, in fact, a subtler process. Thus, the longer answer is that deciphering a script requires an accurate understanding of the grammatical structures of the language the script encodes – for example, phonological (phonemes), morphological (word formation), syntactical (sentence formation) structures and systems. It also requires an understanding of the language’s position in relation to the already existing linguistic families (i.e. groups of languages of common descent, which share a number of comparable linguistic features).

What has been done in relation to Linear A and the Minoan language? Various methods have been explored to investigate the Minoan language, no one of which (especially if used exclusively) has proven entirely satisfactory nor has it given conclusive and uncontroversial results. The etymological method was the first to be adopted (e.g. Palmer Reference Palmer1958, Reference Palmer1968; Gordon Reference Gordon1966, Reference Gordon1969; Best Reference Best1972, Reference 62Best2001; Finkelberg Reference Finkelberg1990–1; Brown Reference Best1990, 1993). This method consists in a lexical comparison between words written in Linear A and the lexicon of known languages used in other (primarily neighbouring) regions (e.g. Semitic, Indo-European, etc.), on the assumption that, should enough similarities be drawn with one of the target languages, this would prove the linguistic affiliation of the Minoan language (written in Linear A) with said target language (and its linguistic family). When applied to Aegean scripts, however, this approach is fraught with problems.

Given that the etymological method consists in comparing Linear A texts (and vocabulary) against known languages, first comes the issue of how to ‘restore’ (or ‘reconstruct’) the readings of Linear A sign-sequences (i.e. words) for as accurate a comparison as possible with the lexicon of known target languages. The orthographic conventions of the Linear A syllabary are not yet fully understood, although scholars often work on the assumption that Linear B orthography can also be retrospectively applied (with an approximation) to Linear A: this is, however, not to be taken for granted. By way of example, Linear B sign-sequence pa-te can be restored as both πάντες /pántes/ ‘all’ and πατήρ /patēr/ ‘father’, based on Linear B orthographic conventions.Footnote ⁷⁵ This shows that one and the very same sign-sequence lends itself to being ‘restored’ (hence read and interpreted) in different ways (based on context). If this equally holds true for Linear A, a number of different ‘readings’ can be given to Linear A sign-sequences: a circumstance which has often produced the misguided outcome of deeming Linear A as ‘a good fit’ for several (even unrelated) target languages based on preferential readings (and restorations) of its sign-sequences. Moreover, another problem with the etymological method is that vocabulary alone is not probative enough for language identification: words can easily be borrowed (as loanwords) from one language to another through language contact.

To avoid the pitfalls of misguided, unfounded or misleading interpretations, in a pioneering article Duhoux (Reference Duhoux1998: 34–5) outlined a theoretical eleven-step methodology for a sound investigation of an undeciphered script as challenging (both contextually and linguistically) as Linear A, with a view to leading to convincing decipherment. This is also known as ‘Duhoux’s methodology for decipherment’ (see e.g. Davis Reference Davis2014: 10–13) and is centred around the following guidelines (here summarised): using only correctly edited Linear A texts (e.g. GORILA) as primary sources, which clearly distinguish between phonetic and ideographic signs; giving sound methodological justifications for Linear A signs’ phonetic readings; reconstructing the phonological, morphological and syntactical systems of the Minoan language based on text-internal patterns; reconstructing Linear A orthographic conventions as accurately as possible; ensuring that the identified patterns (grammatical, orthographic) do occur systematically and regularly across texts; and ensuring that interpretations of Minoan words are compatible with the archaeological and historical contexts (and not anachronistic). A convincing decipherment shall satisfy all these requirements, and as a result it should explain the majority of the Linear A lexicon in a consistent fashion, allowing for an accurate interpretation of all texts (even the most complex). Duhoux’s principles are objective and still observed as theoretical guidelines for decipherment.

Another method to recover meaning is comparative palaeographic analysis between related scripts, which has been done for Linear A and Linear B, given their historical development (see Sections 1 and 4). A good number of Linear A signs’ phonetic values could be surmised by comparison with their Linear B counterparts, or ‘descendants’, as it were. There are contextual, historical reasons legitimising the validity of adopting this approach for the Aegean Linear scripts (see lastly Steele and Meissner Reference Steele, Meissner and Steele2017). However, the proviso needs adding that this is not to be taken as a blanket approach universally applicable to script decipherment in any context.

Textual and internal analysis (textual structure) of Linear A texts, also known as the context-based ‘combinatorial’ method, has proved more promising than the etymological method to shed light on the characteristics of the Minoan language. This contextual approach consists in studying the textual structure of Linear A documents (see also Section 6.1.1) to infer the potential meaning of signs (especially ideograms and transaction terms) and sign-sequences (words) by identifying systematic patterns and assessing contextual cues. This method has been systematically and convincingly adopted by John Younger (Reference Younger2024, Linear A Texts: Introduction), Ilse Schoep (esp. Schoep Reference Schoep2002) and Brent Davis (Davis Reference Davis2014). This method allows us to deduce the (at least approximate) meaning of words without necessarily knowing the underlying language. For instance, given that the word KU-RO tends to occur at the end of a list and is followed by the total number of the commodities listed in the preceding entries, it is highly likely it means something like ‘total/sum’ (see also Section 8.2).

Recently, a novel methodology for investigating Linear A (and also, more broadly, all undeciphered Aegean scripts) has been explored by Brent Davis. This is called ‘syllabotactic’ analysis:Footnote ⁷⁶ a linguistics-based and statistics-based Linear A system-internal and syllable-specific (i.e. ‘syllabotactic’) analysis of sounds’ constraints and words’ positions. It consists in using phonotactic and syllabotactic constraints (which govern the ways in which phones and syllables are arranged into words in a given language and are thus language-specific) as investigative tools to examine the linguistic features of the Minoan language, and also whether Bronze Age Aegean scripts encode the same or different languages. Davis’ research sits at the forefront of linguistic analyses of Linear A and the Minoan language, and is expected to produce intriguing and exciting results (some of which have already been appreciated in the academic community).

Lastly, digital and statistical analyses may also help in deciphering Linear A and/or shedding light on the nature of the Minoan language (one limitation being the reasonably small size and repetitive nature of the Linear A corpus) by detecting meaningful recurrent structures and clusterings that may escape the human eye (as already pinpointed by Packard Reference Packard1968, Reference Packard1971). This is a pathway of research which is the next desideratum in the field, and will necessarily require significant interdisciplinary and multidisciplinary collaborations.

By combining the methodologies and approaches illustrated above, and integrating the ensuing results, we have thus arrived at ‘reading’ Linear A to a great extent and making sense of a good number of texts (see Section 7). Although Linear A cannot be purported to be fully ‘deciphered’ yet, we can say (with good reason) that it is at least ‘partially deciphered’: on present knowledge, it is a matter of ‘degree of decipherment’. So, let us now briefly review what is known of the Minoan language.

9 More Unresolved Mysteries: What Do We Not Have in Linear A?

Another challenge we face is ‘managing absence’. There are no surviving literary texts of any genre written in Linear A (historiography, poetry, treatises, etc.), nor diplomatic correspondence, monumental inscriptions, private letters or the like: anything that does not fall (more or less neatly) within the remits of either administration or cultic activity is missing. A legitimate question springing to mind is: would we expect this evidence? Also in this case, there is no short answer (nor uncontroversial).

9.1 Likely Yes

There is indirect evidence for the use of perishable materials as writing supports, leading us to assume (speculatively and cautiously) that perhaps more extensive writings falling outside the administrative sphere stricto sensu did exist (e.g. historiography, literature, diplomatic correspondence, private documents, etc.). But these writings did not survive: the very same conflagrations responsible for baking the Linear A clay documents (and thus preserving them until today) incinerated writings on perishable material (e.g. parchment, papyrus).

Our best (indirect) evidence for perishable materials used as writing supports are two typologies of clay sealings: flat-based nodules and single-hole hanging nodules, both used to authenticate records, secure their integrity and prevent unauthorised viewing (see Sections 5.2, 5.3 and 6.1.2). Flat-based nodules show string imprints on their undersides (while seal impressions on the front),Footnote ⁸⁰ suggesting they were pressed against tightly folded sheets of parchment (or, more loosely, leather) tied with a string (Figure 14; see also CMS II, 6 pp. 350–7; Hallager Reference 66Hallager1996: 140; Montecchi Reference Montecchi2019: 247–70). Flat-based nodules are attested as early as the Proto-Palatial period and are common to both Cretan Hieroglyphic and Linear A administrative systems (Hallager Reference 66Hallager1996: 39–74, 135–6, 230–1), but are not continued into Linear B administration. Single-hole hanging nodules likely (or possibly) hung from strings tied to (papyrus?) rolls: imprints of the string passing through their top-hole are in fact still visible at times (Hallager Reference 66Hallager1996: 135–45, 197–9; Tomas Reference Tomas, Boehm and Müller-Celka2010: 124; Montecchi Reference Montecchi2019: 270–97). This category of sealings is a new introduction in the Linear A administrative practice, but did not continue into Linear B (as there is no indirect evidence for the use of perishable materials in Linear B; see esp. Palaima Reference Th., Duhoux and Davies2011: 124).

Figure 14 Flat-based nodule showing string imprints on the underside.Footnote ⁸¹

Likely to be placed in archives to be consulted at need, this evidence may have contained information we do not find on Linear A tablets (which we may expect, especially by comparison with Near-Eastern administrative practices) – for example, laws, decrees, balances from previous years, landholdings, diplomatic correspondence and so on. It is worth noting that, in the more extensive Linear B clay documents, these categories of texts are not evident either (except for landholdings and a reference to a possible legal ‘dispute’). Writing supports like parchment/leather sheets (and papyrus rolls), which were obliterated by fire destructions (see Section 2), may have contained longer texts (esp. if written on both sides) compared to short and concise clay tablets. The existence of perishable materials brings up the question of the extent of literacy in Minoan Crete, which is still (partially) terra incognita.

10 Current and Future Pathways of Research: What’s Next?

Although most (and most reliable) material on Linear A is only available in institutional and university libraries and academic databases, in recent years there have been a number of projects disseminating (textual and photographic) material to a wider, educated (yet not necessarily specialist) audience. This section gives an outline of the resources currently available online (often under ongoing development), hoping that their number will steadily increase over time, thanks to international interdisciplinary collaborations between scholars, museums and institutions, and the use of novel digital technologies.