The Undeciphered Signs of Linear B: Interpretation and Scribal Practice

Anna Judson, The Undeciphered Signs of Linear B: Interpretation and Scribal Practice (Cambridge: Cambridge University Press, 2020). 9781108494724.

Reviewed by Theodore Nash, University of Michigan,

This review has been reverted to its unedited version per the request of the reviewer (1/4/2022)

In the story best known, Linear B was deciphered by Michael Ventris in 1952.[1] Other tellings add more heroes—John Chadwick, Alice Kober, Emmett Bennett—but do not generally challenge the notion that, as a result of their work, the script passed from a state in which it could not be read (“undeciphered”) to one in which it could (“deciphered”). But though it has been known since decipherment that Linear B was used to write the dialect of Greek we call Mycenaean, there remain many words which defy interpretation, and many signs without certain phonetic values. It is that corpus of fourteen “undeciphered” signs which Judson has undertaken to investigate. Far from limiting herself to discussion of abstract phonemes, Judson’s interdisciplinary approach shows how richly the tablets repay holistic study. From the firm foundation of linguistic and epigraphic data, she enters the historical and archaeological realms of social context, conceiving of Linear B signs not simply as markers of this sound or that, but as reflections of the lives and relationships of those who wrote them.

Judson begins by reviewing the structure of Linear B and its relation to other scripts used in the Bronze Age Aegean. Three are of special importance: Cretan Hieroglyphic (formally unrelated to proper Egyptian hieroglyphs), Linear A, and Linear B. Shared signs betoken relationships: Cretan Hieroglyphic and Linear A are probably descended from a common model, and Linear B was modelled on Linear A, with little change.[2]

Linear B is a syllabary: the “core” signs record not individual consonants (C) or vowels (V), as letters in our alphabet, but combinations of consonants and vowels (CV): pa, pe, pi, po, pu, and so on. In most cases neither aspiration nor voicing are marked; vowel length is never distinguished; and diphthongs need not be spelled in full: pa can stand for πα, πᾱ, or παι; βα, βᾱ, or βαι; φα, φᾱ or φαι. Conversely, the “extra” signs mark explicitly features such as aspiration (ChV), diphthongs (CVV), or consonantal clusters (CCV): pu2 stands only for φυ or φῡ; a3 only for the diphthong /ai/; ro2 only for the cluster /ryo/. As some “core” values have not yet been identified, the undeciphered signs could be of either type.

Ventris deciphered the script by identifying the relationships between “core” signs before the language was known. The decipherment of “extra” signs followed, often enabled by context. For example, *71 is read as dwe based on variation between te-mi-*71-te and te-mi-de-we-te (τερμίδϝεντες, “having a τέρμις”), both used to describe chariot wheels. Here and elsewhere knowledge of the underlying Greek language helped: *85 was deciphered as /au/ based on the fact that a-u- never appeared at the beginning of a word and because the sequence *85-to-jo yields good sense if read as αὐτοῖο, the genitive singular of αὐτός familiar from Homer. The signs which remained undeciphered generally lack such clear contextual or linguistic evidence.

In the second chapter, Judson identifies the values we may expect to find among the undeciphered signs. Most of the missing “core” values (ji, ju, qu, wu, zi, and zu) are either impossible within the Greek language or excluded by Linear B orthography: Judson argues that only ju and zu are particularly probable.[3] The fourteen undeciphered signs should therefore represent primarily “extra” values. If the sign was inherited from Linear A, Judson argues that it is likely to represent a phoneme similar to one already known. For example, nwa, with its obvious model in Linear A, may be the best represented of an inherited CCV series in nw-, and undeciphered signs may well have values such as nwe. By the same logic, missing “core” values should only be attributed to inherited signs.

Inventions within Linear B, meanwhile, are unlikely to belong to these inherited series. These signs should rather represent ad hoc inventions, morphologically helpful clusters, or even puns. So dwo is constructed of two facing wo signs: duo wo = dwo (δύο), a derivation possible only within the Greek language. Meanwhile dwe, mentioned above, seems to have been motivated by the utility of representing the cluster /dw/, which results from the addition of the suffix /-went/ (Attic -εις, -εντος) to dental-stem nouns (as in te-mi-dwe-te, /termid-went-). Judson plausibly argues that this served as the model for twe, uniquely attested in o-da-twe-ta, “having teeth” (from the zero-grade of the “tooth” stem ὀδόντ- with the suffix -went- again). Similarly, two, used only by a single scribe, may have been a personal invention on the model of dwo. The prevalence of CwV clusters may also owe something to inherited nwa.

Two especially interesting cases are a2 and ra3, which Judson argues were invented to make plurals more distinctive. The former appears primarily in the nominative plural of neuter s-stems (pa-we-a2 = *φάρϝεα, Att. φάρη), the latter only in nominative plurals of the first declension (ku-te-ra3 = Κυθέρραι, “women of Kythera”), never the orthographically equivalent dative singular. Judson plausibly links this with the demands of the administrative system, where syntax was minimal and number more important.[4] This judicious assessment of the extra signs, though formally preliminary to discussion of the undeciphered signs, represents a valuable and original assessment of how Mycenaean scribes made Linear B work for them.[5]

The following chapters constitute a catalogue of the fourteen undeciphered signs, continuing the division between inherited (Chapter 3) and new (Chapter 4). Though Judson offers no novel solutions, she has produced a holistic and critical study which improves our understanding of these signs and their place in the LB syllabary. Judson’s approach is cautious, and she is hesitant to deal with values proposed based on a single word: such ad hoc solutions, though they may be right, remain unfalsifiable without new evidence. Undeciphered scripts are well-served by such diligence.

The discussion of *65 well illustrates her careful method. This inherited sign appears frequently among the Odos Pelopidou tablets from Thebes, discovered in the early 1990s and plagued by interpretative issues. Judson’s careful handling of the evidence (quite literally: uncertain readings were checked in person) is a model of how to work with this disputed corpus. The problem, in brief, is that *65 can be used either syllabically or as the logogram FAR (farina, for its traditional identification with flour), and interpreting its value requires identifying its use in each case. The answer is not always obvious, but Judson argues convincingly that a syllabic reading is much more widespread than initially allowed. In this context she defends the interpretation /ju/, which is derived primarily from the reading of i-*65 as i-ju, /hiyus/, “son” (<*suyus, whence also Attic υἱός). At the end of a word, as often at Thebes, -ju could be taken as an alternative form, perhaps /hyus/, so ra-ke-da-mo-ni-jo-ju would be equivalent to Λακεδαιμόνιος, υἱός, “Lacedaemonius Jr.” As an inherited sign, this “core” value is consistent with Judson’s principles. But though this is all coherent enough, the unparalleled construction and linguistic difficulties leave room for doubt. Moreover, none of the other words in which *65 appears clearly requires the value ju, so Judson commits only to “extremely probable” (28) for this reading.

The only other inherited sign with a single likely value is *56. Based on its distribution and alternate spellings, Judson argues that a reading pa3, /pha/, is most likely (cf. pu2, /phu/). Other signs offer even less for discussion. The distribution of *22 suggests its vowel should be -i, but there is no clear evidence for its consonant. Judson, after considering previous arguments, suggests only that it most probably belongs to an inherited series (zi, pi2, or nwi). Meanwhile, *49 is attested six times at most. In such a small corpus, even distribution is unlikely to be meaningful, and nothing can be said with conviction.

In Chapter 4, Judson moves to the signs without Linear A equivalents. She argues that two (*18 and *19) have equivalents in Cretan Hieroglyphic, and therefore posits Linear A equivalents too rare for attestation in our sparse corpus. Only three undeciphered signs are therefore likely to be Linear B inventions. Of these, *64 alone offers much hope for decipherment. It is found exclusively in the sequence a-*64-jo/-ja, which must be an ethnic adjective. An equation with a-si-wi-jo/-ja (*Ἄσϝιος; cf. Ἄσία, “Asia,” and Hittite Aššuwa-) has been suggested, giving a value swi. This works with the sign’s limited distribution: inherited /sw/ had already become /hw/ or /u/ in Mycenaean Greek, but was still possible in foreign toponyms and derived terms. However, without further evidence the equation of the two terms is never more than possible, and *64 never alternates with si-wi in other contexts. In the absence of more certain evidence, Judson concludes that swi is only a “good possibility” (119).

The other two invented signs, *63 and *83, appear only in personal names and toponyms. This distribution, alongside the attestation of *64 exclusively in an ethnic adjective, suggests that inventions were not limited to sound values useful for Greek word-formation or morphology. It may not be an accident, then, that such transparently useful signs are already deciphered: it is much easier to identify the values of signs in words with convincing Greek interpretations. It is also the case that longer words can be interpreted with greater confidence than shorter ones, and none of these three invented signs appears in a word more than three signs long.[6] The better model here may be two, a sign apparently invented to speed the writing of a repeated cluster, and not susceptible of decipherment without a clear alternate spelling or etymological interpretation.

In the fifth and final chapter, Judson turns to questions of scribal practice, which she approaches through palaeography. This analysis of sign forms is the basis for Linear B scribal attributions in much the same way Beazley’s method of connoisseurship is used to identify the painters of unsigned Attic vases. Though the value and accuracy of connoisseurship are disputed, there has been no equivalent movement within Linear B studies, and many arguments rest uncomfortably on a few pillars of received wisdom. Judson, using her fourteen undeciphered signs as a case study, investigates their foundations for the first time. She acknowledges that such a narrow approach can only be a starting point, and the broader study promised (177) is eagerly awaited. After establishing the palaeographic variants attested at each site, she considers two areas where palaeography has been given special importance: the reconstruction of administrative relationships between scribes, and the dating of tablets.[7]

Scribal relationships present special difficulties. As scribes have left no accounts of their daily lives, their interactions can be reconstructed only from internal (if, e.g., one edits another’s tablets) or contextual evidence (archaeological findspots).[8] Administrative structures, moreover, differ between sites, and each offers different evidence and difficulties: at Pylos the corpus is almost entirely contemporary and most tablets were found in a central archive; at Knossos deposits generally linked by subject matter were scattered across the palace (so we do not speak of archives but departments), and primitive excavation methods and recording have left their chronological relationship largely obscure. Smaller corpora from Thebes and Mycenae allow for less certainty in reconstruction.

Judson argues for a minimal link between administrative links and palaeographic similarity. At Pylos, both Hand 1 and Hand 21 oversaw grain allocations to the same person, and Hand 1 added information to multiple of Hand 21’s tablets.[9] Despite this demonstrable working relationship, the two scribes use the same variant of only two undeciphered signs and write three differently. Judson argues, therefore, that Hand 1’s supervisory role did not extend to a strong effect on Hand 21’s writing practices. A similar situation obtains at Knossos. Both Hand 103 and Hand 115 worked in the textile “department” and share an idiosyncratic feature in their writing of *83; but beyond this Hand 103’s version of the sign looks much more like Hand 117’s, with whom he had no demonstrable relation. Hand 223, who worked with Hand 103 in aromatics, writes *83 without this feature.

Because she is working with such a small corpus, the there are limits to the force of Judson’s conclusions. In the case of Pylos Hands 1 and 21, two shared variants against three different shows that nothing can rest on a single sign, but I am less comfortable that it proves that administrative relations had no impact on palaeography. The same is true at Knossos. That *83 has an extra feature which appears only in one Knossian “department” may suggest that administrative proximity was meaningful in that context. The interpretative caution which serves her so well in dealing with phonetic values is perhaps too stringent here. Judson is, however, open about the limitations of dealing only with so few signs, and her approach is a welcome methodological advance in the study of scribal relations.

The same is true for her study of palaeography as a dating criterion. It has generally been accepted that signs became less “elaborate” over time, though this has never in fact been tested. Especially at Knossos, where the archaeological data are wanting, palaeography has taken on an outsized role in chronological debates: Judson offers an important corrective. After establishing the relative chronology of Linear B deposits, she demonstrates that the earliest attestations of the undeciphered signs are only rarely the most elaborate, the latest rarely the simplest (*19, e.g., has an added central vertical only at Pylos, our final deposit chronologically). She also cites the important work of Ester Salgarella, who demonstrated that Linear B inherited palaeographic variants from Linear A, meaning more and less elaborate forms can be equally old.[10] Overall, many deposits contain multiple variants of the same sign, and even when more complex forms are limited to earlier deposits (such as *64 and *82), the numbers are so small (only five attestations combined) that this may be due simply to chance.

Given the importance of palaeography to our understanding of Linear B, even the demonstration that this small selection of signs does not follow our “rules” is important. As Judson notes, a broader study might well demonstrate that there are certain diachronic trends, but the current focus on individual signs must be reconsidered. This is especially important for tablets such as IK X1 from Iklaina, whose early date has been defended on palaeographic grounds; but it has only eight signs preserved.[11] While palaeography remains a powerful tool, it cannot be used in isolation.

This study represents a thorough and skillful handling of disparate evidence. The linguistic discussion of sound values is rich, but not myopic, and writing is understood as a human practice, rather than the disembodied representation of phonemes. The careful study of palaeographic variants, with an eye to testing old hypotheses, represents a valuable model for future studies of Linear B. As Judson acknowledges, the scope of her investigation was necessarily limited, but she has demonstrated both the value and possibility of pushing these questions further.[12]

Table of Contents

1. The (Ongoing) Decipherment of Linear B (1–35)
2. Identifying “Missing” Values in the Linear B Syllabary (36–95)
3. The Undeciphered Signs Inherited from Linear A (96–150)
4. The Undeciphered Signs with No Certain Linear A Correspondences (151–74))
5. Exploring the Potential of Palaeography with the Undeciphered Signs (175–236)
Conclusions (237–40)
Appendix: Corpus of Attestations of the Undeciphered Signs (241–89)


[1] I am grateful to Dimitri Nakassis for his expertise and to Richard Bott for his critical eye.

[2] These broad strokes are fairly clear, the details less so. The date of publication did not allow Judson fully to incorporate Ester Salgarella’s Aegean Linear Script(s) (Cambridge: Cambridge University Press, 2020) (which I reviewed here), though she frequently cites the dissertation on which it was based.

[3] Although signs with these values could have existed in Linear A, they were presumably not inherited or else given new values in Linear B

[4] Often the so-called nominative of rubric is found where we would strictly expect a dative, such as on lists of rations, in which cases the nominative and dative singular were both functionally and orthographically identical.

[5] The only area where I disagree with Judson is in her treatment of a3̄, which I do not believe has anything to do with the sign on the Mycenaen kessel: see my “Cultures of Writing: Rethinking the ‘Spread’ and ‘Development’ of Writing Systems in the Bronze Age Mediterranean,” in The Social and Cultural Contexts of Historic Writing Practices, ed. P. Boyes, P. Steele, and N. Elvira Astoreca (Oxford: Oxbow, 2021), 209–29, at 219–21. The date Judson gives for the kessel, Late Helladic IB, does not exist, and must be an error for Late Helladic I.

[6] Full attestations for each are helpfully compiled in an appendix.

[7] The book was written before the explosion of new editions and scribal studies of the Pylos tablets in the past year, and so does not assess any of their new scribal identifications. Judson keeps a very useful concordance of Pylian scribal attributions here.

[8] For more on scribal editing, see Judson’s most recent article, “Scribes as Editors: Tracking Changes in Linear B Documents,” American Journal of Archaeology 124 (2020), 523–49.

[9] The two tablets are PY Fg 368 and 828, for which 868 in the text is a very easy error.

[10] See n. 2.

[11] See C. Shelmerdine, “Iklaina Tablet IK X” in Études Mycéniennes 2010: Actes du XIIIe colloque international sur les texts égéens, ed. P. Carlier, C. de Lamberterie, M. Egetmeyer, et al. (Pisa: Bibliotheca di Pasiphae, 2012), 75–77.

[12] The book is produced to the standards that one would expect from Cambridge University Press, and the cover, with its high-quality image of the tablet KN An 1910, is especially handsome. The copy editing of Linear B, alphabetic Greek, and phonetic representations is exceptional. Though generally well-illustrated, there are some areas where extra figures would have been helpful; I note, e.g., the discussion of *79 (145), where Linear A and Cretan Hieroglyphic parallels are discussed but not pictured. It may be that the constraints of print books are not amenable to in-depth palaeographic discussion; online appendices or companions are a good, but not infallible, option. For example, Salgarella’s Aegean Linear Script(s) has an online appendix of palaeographic variants, but the PDF is so buggy that it cannot be used. Its companion website SigLA is, conversely, invaluable.

Leave a Comment

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s