Looting and Faking

This post is rather a departure from the Mycenological material typically presented here, and was born out of recent Twitter conversations that required responses of greater depth than 280 characters allow.  I am not, of course, a papyrologist, though being at Michigan I have been lucky to meet and learn from many; nor am I an expert in the antiquities market, though again I have been lucky to learn from those who are.  If I am therefore wrong in any of the points I present, I will be happy to learn it.  

Papyrology, as the study of written documents, exists at the juncture of two academic fields — the study of ancient languages, and the study of the material remains of the past: which is to say archaeology.  It has,  however, proven to ‘belong’ rather more to the former.  Some of this depends on the fact that the physical artefact itself is rarely as informative as the text written on it, which requires a great deal of expertise to decipher even in the best preserved cases.  Moreover, most papyri were excavated (if not looted) in the late 19th and early 20th century, when scientific archaeology was still very much embryonic, and many were excavated at any rate from secondary contexts (a midden, notably, at Oxyrhynchus); not the most promising material for archaeologists.

Because, in part, the excavation of the material in major western study collections had largely finished by the 1930s, experience in excavation has not been (if it ever was) a pre-requisite for the professional papyrologist, as it remains for the study of other excavated material.  In recent years, however, the publication of certain high-profile papyri (the Gospel of Jesus’ Wife, P.Sapph.Obbink), have foregrounded relating to the texts as artefacts.  How these ancient documents, found (when authentic) invariably in Egypt, made their way to major western research institutions, raised eyebrows, and archaeological questions of provenance and provenience became primary avenues of investigation. These related directly to two other, higher profile, questions: looting and forgery.

The different coverage of the Gospel of Jesus’ Wife and P.Sapph.Obbink have been interesting.  The former, far more sensational, raised immediate controversy over its textual material, and on this basis its authenticity was challenged.  This led to scientific testing and further uncertainty.  In the case of P.Sapph.Obbink, however, presented no major problems in terms of its content; it was the fact that it had appeared, suddenly and without sufficient explanation, in Oxford, that created the controversy.

Provenance and Provenience

Much depends on an understanding of the significance of provenance and provenience.  The former properly indicates an object’s post-excavation history, especially as relates to owners or collections of which it was part.  The latter is simply where an object was found — this can be as vague as the report that it was looted from a specific site, or as precise as a specific context within an archaeologist’s trench.  Neither can nor should be taken for granted; both can be and are misrepresented by the unscrupulous.  

A secure provenience is the sole guarantor of authenticity — unless you can be sure an artefact was responsibly excavated (i.e. not planted) in the course of properly documented archaeological work, you cannot be sure it is what it purports to be.  For this reason, the art market has a complicated relationship with it.  On the one hand, anyone selling an antiquity is quite concerned to guarantee its authenticity and show that it came from the ground and not a forger’s workshop.  But they are generally more reticent about revealing when it came out of the ground: artefacts that left their source country after 1970 — the year a major UNESCO convention was passed — are no longer considered ethical for sale or purchase.  While this is essentially arbitrary, as different countries ratified it at different times, and it was at any rate illegal to remove antiquities from most source countries well before 1970, it serves as a useful line in the sand.  And so provenance becomes important.  On the art market today, a secure provenance is the only way to demonstrate that an antiquity left its source country before 1970 and so aligns with this accepted standard.

In all this the word ‘secure’ is doing a lot of work.  Especially when considering objects on the market, it cannot be assumed that the given provenance and provenience are true.  The fundamental problem is that, 50 years after the UNESCO Convention was passed, the number of artefacts that demonstrably meet its standards is not nearly enough to satiate collectors.  So: people lie.  Of course this came from Greece well before 1970; my Swiss uncle has been keeping it in his attic since the 40s.  For many, this lie (near enough to those often used) is enough.  The simple fact is that, if provenance were subject to even basic scrutiny, the market would not be able to function the way it does.  And, since those implicated in the market are interested in its continuing function, tighter controls will never be willingly applied.  As it is, those with more money than integrity are quite able to buy antiquities, illegal though they may be, and get away with it.  Arthur Houghton, a curator at the Getty in the 1980s, developed the Museum’s policy of ‘optical due diligence’, and this approach — asking for but accepting uncritically any given provenance — remains a standard modus operandi.

The importance of provenance, and the flimsy scrutiny it tends to receive, means that a great number of antiquities now on the market are bought and sold with only the most feeble accounts of their history.  In the case of looted antiquities, it is not in the buyer’s best interests to ask too many questions.  But it is important to remember that not all ‘licit’ antiquities have a secure provenance either: if nothing more than oral family history serves to remove the object from its source country before 1970, then the provenance is no more secure than those constructed for recently looted artefacts.  Fake artefacts must also (necessarily) be sold with falsified histories.  All three categories of artefact — looted, faked, and poorly documented — will have either unverified or fabricated provenances, and will appear for this reason functionally identical on the market.  Because they are shared across categories of artefacts, problems with provenance are not in and of themselves evidence of anything other than the need for further investigation.

Publication Ethics

It should, but unfortunately does not, go without saying that academics should not be interacting with artefacts offered for sale on the market.  If presented with one, a secure provenance placing it outside its source country before 1970 is the bare minimum they should acquire before discussing it in print.  Indeed, the Archaeological Institute of Archaeology has increased the language in its code of ethics from simply prohibiting members from ‘participat[ing] in the illegal trade in antiquities’ (1990) to ‘refrain[ing] from activities that enhance the commercial value of such objects’ (1997).  The 2016 (and current) version makes the latter more explicit, and is worth quoting in full:

[Society members of the AIA should:] Refuse to participate in the trade in undocumented antiquities and refrain from activities that give sanction, directly or indirectly, to that trade, and to the valuation of such artifacts through authentication, acquisition, publication, or exhibition. Undocumented antiquities are those that are not documented as belonging to a public or private collection before December 30, 1970, when the AIA Council endorsed the UNESCO Convention on the Means of Prohibiting and Preventing the Illicit Import, Export, and Transfer of Ownership of Cultural Property.

In principle, therefore, no American archaeologist should ever be fooled into publishing a fake artefact as though it were authentic.  Only if they can satisfy themselves that its provenance, and therefore provenience, is secure, will they be able to publish it; a fake cannot by definition satisfy these criteria.  The investigation of the given account of an object’s history is, far from a formality, a professional obligation and sine qua non for publication.

Papryologists, at least in America, have followed suit: in 2007 the American Society of Papyrologists passed a resolution condemning the illict trade of papyri, including similar language prohibiting the purchase of post-1970 materials by members of the Society.  Publication of such material is not allowed under the Society’s auspices (so in its Bulletin or at its Annual Meeting), ‘unless the author, speaker, or curator includes a frank and thorough discussion of the provenance of every item’.  Though this weakens the language to some extent, it is still worth nothing that Obbink’s editio princeps of P. Sapph.Obbink (which does not include the word provenance once) would presumably not have been eligible for publication in the Bulletin of the American Society of Papyrologists.  The German periodical in which it was published, Zeitschrift für Papyrologie und Epigraphik, offers no ethics statement on its website.  


A secure provenience is the sole guarantor of authenticity; a secure provenance is the guarantor of that provenience.  If, therefore, the provenance does not seem secure, then provenience and authenticity can come into question.  There is no better way to ensure an artefact’s authenticity than to trace a secure line from its current owner to the site where it was excavated.  Indeed, the market’s uncritical acceptance of flimsy provenance in the case of looted antiquities makes it uniquely susceptible to infiltration by fakes (as my friend Richard Bott demonstrated in his excellent Macquarie MRes thesis).

Both looted and fake artefacts are often sold with fake provenances, and only by investigating these can their true nature be determined.  Because the necessity of constructing a fake history for both types of artefacts is essentially the same, evidence that the provenance may be false does not, in and of itself, suggest that an artefact is fake, only that it could be.  At this point, internal evidence will often be adduced one way or another, the conclusions of which will often dictate the tenor of further investigation.

In the case of the Gospel of Jesus’ Wife, questions of authenticity emerged almost immediately, and Karen King’s publication of the editio princeps was in fact delayed to incorporate and be published alongside the results of scientific testing.  In that same issue of the Harvard Theological Review, however, the case was made  from internal evidence that the text of the papyrus could not be genuine.  The tension between philological approaches, which consistently suggested against authenticity, and scientific tests, which showed the ink was appropriate and the papyrus itself ancient, brought the situation to something of a stalemate.  That it is now considered a fake is the result of research into its provenance by Ariel Sabar in an Atlantic article well worth reading (a book will be published in August).  Faced with his results, even Karen King conceded it was unlikely to be real with the remarkable statement that ‘Your article has helped me see that provenance can be investigated.’

Now, it’s fair to say that not everyone has the time and resources to conduct the type of research that Sabar did (though I wonder whether his research was actually any more costly than the tests performed on the papyrus).  But it is also fair to say that only this investigation was able to quell any further debate.  And, of course, if false provenance could be offered for a fake artefact, the exact same could have been offered for a looted one.  All the scientific tests and grammatical infelicities in the world will fail to identify an artefact’s modern history.  And yet there was never any concern by those who believed in its authenticity, to my knowledge, that she may have published a looted artefact, though this was of course not an issue for those who thought it was fake.

In rather stark contrast, the authenticity of P.Sapph.Obbink was essentially never at question.  In his initial announcement of the poem in the TLS Obbink did offer a brief defence:

Sappho Authenticity

Within the context of the article, however, this is presented as almost an after-thought, and the argument is ultimately bloodless.  In a later discussion of the poem’s authenticity, Obbink allows Martin West’s initial impression to represent the communis opinio: ‘My initial impression was that it was very poor stuff, and linguistically problematic. But the more I look at it, the more OK it seems. It’s certainly not one of her best, but it has her DNA all over it’.  Controversy was rather immediately centred on its provenance — or rather the lack thereof.

The variants and permutations of the provenance given at different stages defy the possibility that the full story is known.  I can hardly improve on the account given by Bettany Hughes in a recent Guardian article, which ultimately shows that the cartonnage from which the poem was supposedly removed was still unsold after other fragments of Sappho ‘extracted’ along with P.Sapph.Obbink in the same hand had been waved around by Scott Carroll, then associated with the Green Collection.  Indeed, though that same article does note that some thought that it may be a fake (more on this later), the controversy of its provenance and potentially looted origin presupposed that it was authentic.

That provenance is, bluntly, a headache.  The most recent overview, and newest information, can be found in an(other) Atlantic article by Ariel Sabar, though earlier posts by Brent Nongbri and Roberta Mazza offered a more detailed look at various aspects.  As reported by Sabar, the owner of both P.Sapph.Obbink and the fragments purchased by the Green Collection was the Turkish antiquities dealer Yakup Eskioglu.  The article further demonstrates that no significant Green Collection papyri were actually extracted from mummy cartonnage, as was famously claimed, but rather purchased from Obbink or Eksioglu.  The destruction of mummy masks was, it seems, both literal and figurative papyrus laundering.  In light of this, it seems likeliest to me that the same happened with P.Sapph.Obbink, and the convoluted cartonnage narrative was simply a red herring to legitimise a recently looted papyrus.  Paul Barford offers a good overview.

As was mentioned, all of this rather presupposes something: that the papyrus is authentic.  But, of course, a shaky provenance could equally belong to a fake, and speculation on Twitter recently turned in that direction when Clara Bosak-Schroeder (@thaumatic) tweeted:

This sparked a fair bit of discussion, including ultimately this post.  I stand by my immediate response, though it does not in fact make a great argument:

And, of course, we have this from Armand D’Angour

… but a faker would say that, wouldn’t they?  Ultimately, any statement about how a forger thinks is likely to be tendentious.  Some forgers may go for the sensational; others, perhaps more wary of being caught, the hum-drum.  Arguments can be marshalled in either direction, limited only by the ingenuity of those posing them.  Psychologising our putative faker gets us nowhere.

(Not to cast aspersions on Armand, of course, who who surely would have written something much more interesting.)

In spending more time with the poem prior to this post, I have found with West that familiarity does improve it, though it ultimately lacks something of the vivid immediacy and subtle imagery of her more justly famous poems.  I do not doubt that it is real, though it is not my intention here to make a comprehensive case for the text’s authenticity.  At any rate, I have no particular investment in its authenticity; I have never published on it, and doubt I ever will.  It is, ultimately, much less exciting to argue for the authenticity of a text, especially when there are no major issues that need defending.  That is very much the case with this poem.  The most lethal objection, the linguistic problems noted by West, is not so bad as all that.  These are generally points of orthography where the text does not conform to the Lesbian Aeolic dialect.  These sorts of errors are only to be expected in a papyrus written in the third century AD, perhaps 800 years after the poem was written; few today would offer to copy out Sir Gawain and the Green Knight and guarantee there would be no errors, especially if the possibility presented that the exemplar may need correcting.  This is essentially what has happened, and these sorts of errors are incredibly common in Greek literary papyri.

The case is further bolstered by the fact that the Brothers Poem, and other Green Collection papyri in the same hand, join textually with previously known fragments.  This is not a smoking gun — our faker may have been deeply familiar with Sappho — but the effort would be quite extraordinary.  Given what amounts to, I think, fairly significant evidence for its authenticity, the case that it may be a fake needs to be made explicitly, and with something more than just dubious provenance.  As has been stressed, this is a shared feature of both fake and looted antiquities, and so problematic provenance is not evidence, in and of itself, that an artefact is fake.  As in the case of the Gospel of Jesus’ Wife, knowing the true provenance would solve the problem, but in the absence of that arguments have to be made from internal evidence, and, for all those who’ve expressed doubt, I’ve yet to see any case actually made.

(Throughout I have used ‘authentic’ to mean that it was collected in the first book of the Alexandrian edition of Sappho and circulated as one of her works in antiquity.  The question of whether is it from the hand of Sappho ipsa, or an ancient imitator, is a very different one, and not immediately relevant to the authenticity of P.Sapph.Obbink.  While there are some divisions within the Sapphic corpus based on metrical and dialectal differences [the so-called ‘abnormal’ poems], there is no reason to count this poem among them — and indeed no consensus that they couldn’t be by her anyway.  But we are coming here to the limits of my expertise, and these arguments are probably better made by others.)

On the other side, what we do know about its provenance points to the eminent possibility that it was looted.  A connection with Yakub Eksioglu is nearly guaranteed, given that he was the one who sold the related fragments to the Green Collection, and it’s fairly clear that had access to material looted from Egypt.  That P.Sapph.Obbink is an illicit, looted document is the most economical explanation — and the basis of the entire controversy.

Indeed, that’s why I find speculation as to its authenticity rather confusing.  As Enrico Prodi noted…

If P.Sapph.Obbink was faked, not looted, then the entire ethical problem is obviated.  The pivot is drastic, especially all that we know about other illegal dealings by all involved.  What does it benefit to reframe the conversation in terms of authenticity? At this point, with every reason to think that P.Sapph.Obbink is looted, and no positive evidence in favour of forgery, it simply confuses the issue.  This is not to say that we shouldn’t worry about fakes, but only when the evidence points us there.


So, then, what is the worse crime —  publishing a looted artefact, or a fake one?  This question, I think, has a clear answer, but the issue itself is not simple.  After all, the scholar who publishes a looted artefact can make the claim that they are performing a service to scholarship — if they didn’t publish it, the argument goes, it would disappear into a private collection, and the world would be (e.g.) one poem of Sappho poorer.  Publishing a fake as authentic, however, distorts the corpus, and so accomplishes rather the opposite effect.  It is almost inevitably malicious if done knowingly; at at best it is an act of scholarly irresponsibility.  Fakes have the ability to alter scholarly narratives, especially when opinion is divided, which (at best) suck oxygen away from more interesting debates, or (at worst) encourage the wrong questions and guide us to the wrong answers.

Without discounting the difficulties that fakes offer, their significance is generally limited to the pages of academic periodicals and monographs.  We can rank them of the utmost importance only if we consider academia a closed system, removed fromt the causes and consequences of the wider world.  But while a fake, if made by a scholar, may exist within this microcosm, when antiquities dealers intrude they bring the world with them.  The scholarship on looting and the black market in antiquities is now extensive, but the simple fact is that, beyond the inevitable damage to archaeological sites, it is a brutal form of transnational crime that leads to death, murder, and the exploitation of war-torn countries.

In this case it becomes, I think, absolutely clear that engaging with looted antiquities is worse than faked ones.  Looting in Egypt (where our papyrus must have been preserved) follows the same patterns as it does elsewhere, and was particularly bad in the disruption following the Arab Spring revolt of 2011.  Children employed by looters to dig through sites put their lives at risk, and occasionally lose them, and two guards hired to protect a site were shot and killed by looters in 2016 (see further this excellent article by Roberta Mazza).  It is not even the looters, often driven to the activity by desperation, who benefit; they are paid pennies compared with what those capable of getting objects out of the country and to the major auction houses in London and New York.  The sensationalised reporting of new artefacts, such as P.Sapph.Obbink, or knowledge that the Green Collection was willing to pay vast amounts for papyri, fuels and perpetuates this process.

As mentioned above, archaeologists are at this point well used to the idea that looted artefacts should not be published; the discipline has decided that whatever knowledge may be lost in this way is less important than avoiding contact with organised crime.  But in this the archaeologist who turns down an artefact for publication has an advantage over the papyrologist asked to do the same.  Archaeological finds offer, on their own, very limited information; we understand them through the context in which they were excavated — where were they found? what were they found with? One of the most spectacular recent finds in Classical Archaeology is the Combat Agate from Pylos.  Whatever its merits as a work of art, however, analysis would be far more limited if it had appeared without context.  We may place it on Crete, based on the quality, or else at Mycenae; Pylos would not be the first guess.  Understood as part of their proper context — the burial goods of the so-called Griffin Warrior — it is helping us re-assess our understanding of the early Mycenaean period in Greece.  In a different universe, where the tomb was looted and a scholar was offered the Combat Agate alone to publish, they would be rejecting only a(n exceptionally) pretty rock, not this entire scholarly opportunity.

The situation with papyri is obviously different.  Even those that have a recorded archaeological provenience are typically assessed based on what they say, not where they came from (the first papyrological publication based entirely on the building in which the papyri were found was published in 2018).  And so I think it is fair to say that the papyrologist who turns down a looted find is making a greater sacrifice than the archaeologist who does the same.  While, I think — as, albeit, an archaeologist, not a papyrologist — that rejection is the clear and necessary option, it is perhaps also fair to say that coming to that decision may not be easy.

Final Thoughts

But if the artefact is published, what then?  Archaeologists, in general, are happy to ignore looted artefacts, not least because they tend to lack the sort of contextual information that enable analysis.  But it is perhaps glib to suggest the same standard for papyrologists and textual scholars.  We could, I suppose, teach Sappho without the Brothers Poem, especially at the undergraduate level; but should we also ignore fr. 5 (relatively well-preserved, as things go) because one of the Green Collection papyri now records the first word?  Or fr. 16, the famous ‘what one loves’ poem, which has gained words in the same way?  It is not clear to me that this situation is tenable.  Because these looted texts were published, all who study Sappho are now implicated in the ethical question, and must make their peace with this.  I first read Sappho in depth this fall, and I devoured as much as I could — including the lovely Tithonus poem, published in 2005 — but deliberately avoided the Brothers Poem, a strategy which worked for about a month before it appeared on the midterm.  Is there a moral to this story? I don’t know — but I’m glad that I’m not likely to run into the same problem again soon.

While Obbink’s situation is clearly not resolved, and the ultimate fall-out (whatever it may be) will not be based on his publication of this poem, a comparison of his current circumstances with Karen King’s is in some ways instructive.  In many ways, she is lucky that doubt presented immediately, and the situation was resolved with (in academic terms) great speed.  The Gospel of Jesus’ Wife did not fully enter the corpus, but maintained a liminal position, and lasting damage seems to be minimal.  At any rate, she remains Hollis Professor of Divinity at Harvard, and there has been no public consequence beyond, presumably, a fair deal of embarrassment.

In some ways she got lucky.  Had Sabar’s investigation found that the papyrus was not a fake, but had left Egypt much more recently, the narrative would be very different.  By accepting a piece presented by the stranger and the story he offered, this was the risk she took.  Provenance remains the only relevant question when dealing with an unknown artefact, and is far too important to accept at face value.

My thanks to Richard Bott for proofreading and years of valuable discussion on issues of fakes and forgeries.  All references to the excellence of his work are mine.

The Tacitean Trump

Pagnion of an April’s night.

in ea tempestate accedit ad imperium princeps haud credibilis, gente natus divite se ipsum peperisse prosperitatem abritratus: magnae opes isti inopum vindici. adeo sapiens ut, quod pro venatione veneficarum habuit ipsas veneficas semper invenisset. sed alitus non solum per populos plebesque, sed cultores eius senatores, obstrincti tanto amori iudicium. suffragatores eius bonos viros appelavit, ab aliis deplorati. consensu stultissimorum capax imperii, imperavit stultissime.

A Phaistos Disk Primer

The Phaistos disk is something of an embarrassment to students and scholars of the Aegean Bronze Age.  Its celebrity has given it an inordinately central position on the discipline, the extent of its fame something like inversely proportionate with its evidential value to scholars, to say nothing of certain knowledge.  Only Atlantis can vie with its centrality to pseudo-scientific narratives of the Bronze Age, and new ‘decipherments’ are often lauded by the media in breathless tones, ignorant, it seems, of the oblivion to which most previous attempts have been consigned.

The proximate cause of this post is the announcement of (yet another) decipherment by Gareth Owens, which he will announce at a ‘Cambridge Lecture’ that is nothing of the sort:

At a deeper level, though, the need for this post reflects the general absence of public-facing scholarship on the Aegean Bronze Age, and especially its scripts, in recent years.  This was noted four years ago, now, by Dimitri Nakassis, and while I think the situation has improved in some ways (the CREWS Project has an excellent blog, and the number of Linear B scholars active on Twitter has at least tripled), the fact remains that ancient scripts are inevitably considered abstruse and difficult, even by those in cognate disciplines.

So far as the Phaistos Disk is concerned, the best online resource (to my knowledge) is a page maintained by John Younger as part of a larger website on the Minoan scripts, with other sections dedicated to Linear A and Cretan Hieroglyphics.  Anyone looking for a well-illustrated overview of the disk and its signs should certainly start there (he also, incidentally, offers a rebuttal of one of Owens’ earlier attempts at decipherment).  I aim here to complement it as a source, and so largely defer discussion of sign forms and their comparanda in other Minoan scripts, except as these pertain to the possibility of decipherment and the question of authenticity.

(While I was writing this, Anna Judson posted an excellent overview of the various scripts used in the Bronze Age Aegean, and how much we know about them.)

img_1497.jpg(Side A of the Disk on display at the Heraklion Archaeological Museum.)

As ‘mysterious’ as the disk is, there are some facts which brook little disagreement.  On total count, it bears 241 signs, 45 of which are discrete.  This is quite too many for an alphabet, which renders consonants and vowels separately, and too few for a pictographic (or logographic) script, in which each sign represents a full word.  Rather, it suggests a syllabary, where each sign represents some combination of consonant and vowel.  We should think this even if all available evidence for the other Aegean scripts indicated that they, too, were syllabic in nature.  For Linear B and the Cypriot syllabary (a niece or nephew of the former, in crude genetic terms), we know this to be the case because we can read them.  Because of their relation to the other scripts, especially Linear A, we can be quite certain in their cases, too.  Both internal evidence (the count of signs on the Disk) and external evidence (our knowledge of writing in the Bronze Age Aegean) corroborate the same conclusion.  By processes such as these, it is possible to draw logical inferences that have some probability of being correct; this is how work should be done on the Disk.  For other examples of conclusions (never radical) which have a greater likelihood than others of being correct, and with which any decipherment attempt should in large part conform, see this excellent article by Yves Duhoux.

But that is not to say that, by working carefully out from plausible hypotheses, we can decipher the Disk.  On that front, I cannot improve greatly on Judson’s summary:

The Disc is a completely unique inscription whose function (and therefore likely content) we don’t know, and although some signs do bear resemblances to other Cretan writing systems, its exact relationship to these (also undeciphered and poorly understood) writing systems is very unclear.

There is one further problem: because the Disk is the only known representative of its writing system, even if we somehow arrived at the correct reading, there would be no way to prove it.  This would require external verification, which is prima facie impossible when dealing with a unique artefact.  Some of the more clever would-be decipherers attempt to side-step this by incorporating a ‘decipherment’ of Linear A which, on the grounds that (as deciphered) they represent the same language, this offers the necessary external verification.  Given the difficulties inherent in deciphering Linear A, this only has the effect of making their argument less likely.  Ultimately, the Disk cannot tell us anything certain about Minoan culture, and the ingenuity wasted on it sucks the air from much more interesting discussions about writing in the prehistoric Aegean.

(Side B of the Disk, Heraklion Archaeological Museum.)

For the last decade or so, another controversy has hounded the Disk.  This is the question of authenticity, provocatively questioned by Jerome Eisenberg in a 2008 Minerva magazine article.  Some context:  this was not a peer-reviewed article, nor did the author have to convince an editor of its merits, as Eisenberg himself plays that role.  This has no direct bearing on the strength of his argument, of course, but this is not a source, for example, that an undergraduate would generally be encouraged to cite.

I had (perhaps naively) not thought this view had won many adherents, but some recent discussion on Twitter suggested it had rather more traction than I had thought.    It is therefore, I think, worth offering an overview of the controversy.  I believe it is real, but some senior scholars do not, and the conclusion on either side must rest on a balance of probabilities; as so often in archaeological questions, we want for certainty.  For point-by-point discussion, readers should refer to Eisenberg’s original article (cited above) and this insightful response by Pavol Hnila.   Here I will simply review one aspect (scientific testing) untouched by Hnila, and the evidence of more recent scholarship.

Those who argue in favour of a forgery do claim that certain evidence is attainable, and take the Heraklion Museum’s refusal to allow the requisite scientific testing as a tacit admission of guilt (as it were).  Eisenberg’s views, and frustration, are laid out in this NYT article.  The fact of the matter remains, however, that the requested test (thermoluminescence dating)  is destructive, no matter how much Eisenberg seeks to minimize this aspect:

Dating by thermoluminescence would not really damage the disc, as it would require drilling a small hole or two on the edge that could afterward be infilled so that it would be virtually invisible.

It need not, I think, be said that the Heraklion Museum has many reasons beyond fear of the result to want to avoid destructive tests on, yes, one of its most famous artefacts.  However invisible Eisenberg imagines the damage will be, this is a test which requires taking a drill to a three-and-a-half thousand year old piece of baked clay.  It is further not clear that a single sample would be sufficient to determine the age of the Disk.  Contemporary discussion on the AegeaNet mailing list likewise demonstrates some uncertainty as to the infallibility of such a test; it depends on the clay being evenly fired, and so multiple samples, rather than one, may be required.  Despite the popular sense that scientific testing is able to offer the last word on controversies in the humanities, circumstances in reality are very rarely so clear-cut.  A test might be conclusive, but the destruction and expense might equally lead down only a dark alley.

Further contextual evidence was also offered in 2017, when the Middle Minoan stamped pottery from Phaistos was published by Alessandro Sanavia.  While it was already known that the use of stamps was paralleled both at Phaistos and elsewhere, we now have a much larger corpus available for study, and more of the Phaistos Disk signs have external comparanda (see especially pp. 89-94 of Sanavia’s article).  Though no less unique as a written document, neither the technology nor iconography of the Disk appear now so exceptional as could once be thought.

Linguistic evidence, comparing sequences of signs on the Disk with those found in Linear A records, has also been used in an innovative way to argue for linguistic continuity across the two scripts (Davis 2018 — unfortunately pay-walled, but get in touch if you’re particularly curious).  Needless to say, it would be quite remarkable if it were within our forger’s ability to reproduce accurately an unknown language.  This sort of approach, patient and ‘unspectacular’ in the broad sense (more of the article is taken up with arguing for the validity of the method than anything else), both demonstrates how responsible work can be done on the Disk, and ‘scholarly’ methods, rather than just scientific, contribute to the discussion.

I have not set out to answer the question of authenticity here, only compile and present the best available arguments and evidence.  For me, they make a convincing case for authenticity, but enough uncertainty remains (especially around the exact context of its discovery, for which see Hnila and this AegeaNet email by Vincenzo la Rosa) that I can see room for doubt.  Given the scope for misinformation when it comes to the Disk, it is my hope that this post, at least, may serve as a reliable first port of call for the curious.

A Lost Mycenaean Record

When I arrived in Ann Arbor this fall, word eventually made it to me that the Kelsey Museum of Archaeology had in its holdings an old photograph, terribly faded, of what appeared to be some sort of inscription.  My own interest in Linear B being well known, I was able to get a look at it; little, indeed, could be said except that some features (or were they shadows?) gave the dim impression of an Aegean script.  Disappointed, but not greatly surprised, I was soon occupied entirely by coursework, and between Tacitus’ torturous Latin and the great lacunae of knowledge in the Greek Early Iron Age, the damaged photograph very quickly passed from my mind.

All of that changed last night, when I found myself walking the halls of the Kelsey.  A dream, to be sure — not least because I had left Ann Arbor for home at the end of last week.  Something was different, though: details were wrong; the old wood looked fresh.  I soon came to realize I had no control over my wandering feet, and so relaxed, a passenger in my own body, or some oneiric simulacrum thereof.  As watched, the sense of strangeness increased, and the more I saw the more I felt out of time, as though I were seeing the old stones in much younger days.  Soon enough I was in the basement, and then the archives, and in my hands was an album; with a thrill I recognized it as that which hold the photo.

I opened it, or rather watched as it was opened, and soon I was faced with a clear photograph of a Linear B tablet, its crisp lines matching the dim outlines I had managed to perceive in the picture viewed now some months ago.  Along the bottom was written ‘Tablet in the Minoan Linear Script of Class B.  Gift of A. Evans to F. Kelsey.’   I forced my eyes back to the signs, insistent that I should not forget them, and after what may have been thirty seconds or minutes of dreamtime, just as I felt certain I could not forget what I had seen, all faded to white and I was staring instead at my ceiling.  With a jump I leapt from bed, grabbed a nearby pad of paper, and sketched what I had seen.  That it was indeed a true vision I had no doubt, but this of course falls well short of sufficient evidence for scientific publication, so I offer my drawing and findings here:

ChristmasWe may transliterate thus:

.1 po-me-ne                         VIR 12
.2                                          OVIS 60
.3 ma-ko                              VIR 3
.4                                          AUR M 1
.5                                          RI
.6                                          SU
.7 tu-pa-ni-ti-jo                  PUER 1


  • po-me-ne  is transparently ποιμένες, shepherds, a common word in Linear B records.  VIR 12 indicates that we are dealing with a dozen of them.
  • The following line, seemingly incomplete, must represent a continuation of the previous; that is, OVIS 60 (60 sheep) must represent animals accompanying the shepherds.  This is, of course, not a particularly coherent number; we have only 5 sheep per shepherd.  These cannot represent their whole flocks, but only a small portion, taken somewhere — as a gift?  It is not clear.
  • ma-ko is not a word known in Linear B, and must be a (probably nominative) plural, given VIR 3 (3 men) at the end of the line.  It is not clear what we should read; options include μάργοι, ‘mad men’, or perhaps more reasonably, μάγοι, ‘wise men’.
  • AUR M 1 is quite clear: roughly a kilogram of gold.  As with the sheep listed below the shepherds above, we must imagine that this accompanied the ma-koRI is, however, curious.  It would typically indicated λίνον (ri-no), ‘linen’, but the absence of a measure is curious.  The use of a logogram on its own is known in other contexts to indicate a full measure, but this never occurs with RI.  Something other than linen is, I think, being indicated here; may we conjecture λίβανος, frankincense?
  • SU must likewise be an unknown logogram.  If λίβανος is in fact correct for RI, then might we read *su-mu-na, σμύρνα — myrrh? These, I think, must be gifts, which may suggest that the curiously small number of sheep per shepherd are likewise gifts.
  • tu-pa-ni-ti-jo is another hapax graphomenon.  Likewise curious is the logogram, which takes the conventional form of VIR, man, but is much too small, from no apparent necessity.  This small size, and the ending –i-jo, which suggests the diminutive -ίον, have led me to suggest a reading PUER, which is to say ‘child.’  tu-pa-ni-ti-jo must therefore correspond; the only word which comes to mind is a diminutive of τυμπανιστής.  I would suggest, therefore, τυμπανιστίον, ‘little drummer boy’.  He is singular, and unlike the other groups, seems to have no gift to bring.  Since the situation seems to have demanded as much, perhaps he rather played his drum.

The record, then, records gifts given by shepherds, wise men, and a ‘little drummer boy’ to an unnmamed recipient.  What could have brought such an eclectic group together is only to be wondered at, as indeed is the identity of the one receiving these gifts.

A Cucurbitaceous Linear B Inscription

I set off home from the Kelsey Museum of Classical Archaeology last night under a drooping dusk, the streetlights just flickering into life as the leaves tumbled in their autumnal dance.  The air was cold, and a firm wind seemed to find my face no matter which way I turned.  Tucking in my chin, I set a deliberate pace with thoughts of a boiling kettle and hot tea on my arrival.  But as I passed through the deepening shadows between lights, my eyes were drawn past a decrepit fence to a house unusually dark and grim.

What guided my next steps I do not know, but soon I was standing before a gate which hung from its post only, it seems, by force of will.  It swivelled with a groan in the wind, and strange curiosity took me past it and into a yard, the air full of leaves, swirling as though levied for some dim purpose.  Still, my attention was rapt, and I squinted through the leaves and vesperal gloom to see — I know not what.  The house, perhaps, to catch the root of my purposing.  But before I could grasp illumination from the shadows, my attention was snatching away by something before my feet.

More substantial than a leaf, I looked down and saw a broken piece of pumpkin, curiously scratched — almost, I should think, inscribed.  But as I bent to take a closer look a shape leapt from the leaves to my right in a flurry of fur and fury; I could but watch as the darting squirrel snatched the pumpkin and ran off behind me.  I turned in indignation, and gave what pursuit I could — back down the path and through the tired gate.  But man for all his virtues cannot conquer a squirrel in the chase; soon it was halfway up an oranging oak, lost in the shadows of its remaining leaves.  From above, I could hear — or did I only imagine it? — the dim sound of gnawing.

Disappointed, I stood for a moment until with a sudden glare the streetlight above finally flared into life.  Startled, I turned with a sudden hope that the light might illumine the strange house, and all the night’s curiosity not be frustrated.  But as I looked back along the street I could not find the weary gate, nor the fence’s planks, ragged and raw; my eyes met only only a row of happy houses, their windows gayly gleaming.  I blinked, and shook my head, but the vision — had it been nothing more? — of only a moment prior was gone.

The walk home did not clear my thoughts, nor could I shake the curious sense that I had recognized something earlier as I clicked on the kettle; I stood, distracted, as it merrily steamed and bubbled, and only as I sat down and sipped at the brew did my mind return to the scratches on the pumpkin.  With a start I grabbed a piece of paper, and with a strange certainty I sketched them out from memory.  Only then did my thoughts come together: staring back from the page, clear as day, was the Mycenaean Linear B syllabary.

I have more questions still than answers, but here follows my drawing and what sense I can make of the signs, in hopes some sense might eventually be made of this evening most curious.


.1 a-re-ti-no-ta-to
.2 a-ko-ro

.2 me-ka-ra ko-ro-ku-ta


  • There is no ruling, which is odd for a Linear B inscription, but as this medium is unprecedented, we should perhaps not be surprised.
  • a-re-ti-no-ta-to a-ko-ro is clear enough in interpretation, though the sense is obscure.  The reading ἀληθινώτατος ἀγρός (/alethinotatos agros/, ‘most sincere field’) is suggested readily enough, though how or why a field should be sincere I cannot say.  Perhaps more helpful is a reading ἀληθινωτάτῳ ἀγρῷ (/alethinotatoi agroi/) in the dative/locative, situating events in this curious field.
  • me-ka-ra ko-ro-ku-ta is likewise straightforward as readings go, and only slightly less opaque in sense.  We may read μεγάλα κολοκύνθα (/megala kolokuntha/), which evokes the strange image of a ‘great pumpkin’.  How we might conceive of this figure is far from my understanding, as it is association with the ‘most sincere field’.  The vagaries of Linear B allow us to read this in a number of ways — a nominative, perhaps of rubric, or else a dative of reference (μεγάλᾳ κολοκύνθᾳ, /megalai kolokunthai/); was something offered to this figure?  Perhaps we may imagine the pumpkin itself was an offering; perhaps, indeed, we should conceive of the ‘field’ as a pumpkin patch.  The ‘great pumpkin’ might then be a tutelary deity or genius loci.  What is offered, then, is an offering to the ‘Great Pumpkin in the most sincere pumpkin patch’.

Linear B Translated: MY V 659

Another tablet in translation for your reading pleasure (you can find a full index here).

This one is fairly well served in modern handbooks (it’s in Documents and the more recent Companion to Linear B), but there remain some differences in interpretation — and it’s a personal favourite.  It was found in the West House at Mycenae, outside of the citadel proper, and indeed most of our records from that site come from outlying “houses”.  While on initial discovery it was thought that these recorded private transactions of the households, as more (such as this one) were found, it became clear that the same people were named on some of the tablets, suggesting a broader (probably palatial) level of control.

MY V 659 drawn

.1       wo-di-je-ja   ,      de-mi-ni-ja          1
.2       ma-no   ,   a-re-ka-sa-da-ra-ka           2
.3       ri-su-ra   ,   qo-ta-qe                         2
.4       e-ri-tu-pi-na   ,   te-o-do-ra-‘qe’       2
.5       o-to-wo-wi-je  tu-ka-te-qe               2
.6       a-ne-a2   ,   tu-ka-te-qe                     2
.7       pi-ro-wo-na  ki-ra-qe                       2
.8       p̣ụ-ka-ro  ke-ti-de-qe                         2
.9                               ]-ri-mo-qe                 2
.10                             ]ma-ta-qe                  2
.11                                ]*8̣2̣                         1
.12                                        ]-q̣ẹ                  2
.13                                                                 ] vac.
inf. mut.
lat. dex.
]  ,  i-ri-[•]1̣        ke-ra-so   ,   ki-ra-qe 2

.1 <For> Rose: 1 Bed
.2 <For> ma-no <and> Alexandra: 2 beds
.3 <For> ri-su-ra and qo-ta: 2 beds
.4 <For> e-ri-tu-pi-na and Theodora: 2 beds
.5 <For> o-to-wo-wi-je and <her> daughter: 2 beds
.6 <For> a-ne-ha and <her> daughter: 2 beds
.7 <For> pi-ro-wo-na and her infant daughter: 2 beds
.8 <For> pu-ka-ro and ke-ti-de: 2 beds
.9 <For> someone and ]-ri-mo : 2 beds
.10 <For> someone and ]-ma-ta: 2 beds
.11: <For> ]-*82: 1 bed
.12 <For> someone and someone]: 2 beds
.13 (empty)
Damaged below.
On the right side:
] <for> i-ri-[?]: 1 bed.  For Cherry and her infant daughter: 2 beds


  • While Duhoux (Companion to Linear B vol. 1, pp. 290-4) takes issue with the conventional interpretation that this tablet records the allotment of bedding, I see no problem with it.  He objects that the key word, de-mi-ni-ja (δέμνια, demnia, plural as is most often the case from Homeric poetry onwards) should, if really the key word, come first.  But it sits here quite naturally where a logogram would sit, after the name and before the numeral.  Its omission in subsequent lines also matches the omission of logograms on other tablets (cf. KN Fp(1) 48 where the sign for oil is only given once); it is perhaps even more explicable here given the length of the word.  He further objects that the names are all in the nominative, rather than the dative, which would be odd in a case where items are being distributed; but this is no real objection, as the so called “nominative of rubric” is common in the records, which simply indicates the person involved.  Hence the supplied <for> in the translation, which could likewise be omitted without any great injury to understanding.
  • Ma-no, along with a-ne-ha and ke-ra-so (“Cherry”) are also listed on MY Fo 101, which records the distribution of oil.  As Fo 101 was found in the House of the Oil Merchant, not the West House as was this one, it is clear that these records were not limited in interest to the “house” in which they were found.
  • In line .2, we can be quite certain that the scribe has made a mistake.  They have written a-re-ka-sa-da-ra-ka, “Alexandra-ka”.  In subsequent lines, the postpositive “qe” is used to link the two items (cf. alphabetic Greek τε, Latin -que).  The signs for ka and qe are perhaps the most similar in the syllabary, as can be seen in the drawing — the former a circle with a cross, the latter a circle with (generally) three or four dashes.  This, combined with the fact that “Alexandra” is such a recognisable name, makes it quite clear they did not intend an obscure alternative on the lines of “Alexandraka”.
  • Theodora joins Alexandra among the names still in use today.  As Chadwick already noted in the initial publication, names form from the stem θεός (theos, god) are hard to parse in Linear B (there is only one other), and it must have some other valence from the modern name with its overt Christian connotations.
  • The word tu-ka-te (θυγάτηρ, “thugater” — daughter) is one of the few which indicates that we should take the names as nominatives (the dative would be tu-ka-te-ri, θυγατέρι).  Mainland scribes don’t conventionally write the -ι in diphthongs, so (e.g.) te-o-do-ra could be read as nominative Θεοδώρα or dative Θεοδώραι.  Since all of the nouns on the tablet read most naturally if taken in the same sense, we should assume that if tu-ka-te is nominative, then so too are all the others in cases of ambiguity.
  • ki-ra, here translated “infant daughter” is somewhat obscure.  It seems unlikely to be a name, since it appears twice on the tablet (line .7 and the right side), and is therefore perhaps better taken as a noun akin to tu-ka-te.  If it corresponds to an alphabetic Greek word, it will be γιλ(λ)ά (gilla), attested in the form νεογιλός (neogilos) in the Odyssey.  Hesychius, an Alexandrian who compiled a dictionary of rare Greek words, glosses it as “newborn”, and a scholion to the Odyssey verse clarifies “nourished by milk”; it might also correspond etymologically to a Lithuanian verb meaning “to suck.”  If this interpretation is right (and it seems to be on fairly firm footing), then pi-ro-wo-na must have had a daughter still breastfeeding.  It is difficult to understand why a child so young would need a separate bed, but the exact sense of de-mi-ni-ja is somewhat opaque — bed, bedding, and other similar words remain possible, so we should not be too dogmatic in denying a child’s need for it.
  • In line .11, it seems only 1 name was recorded; it was probably too long to fit alongside another (cf. a-re-ka-sa-da-ra-ka, but in that case joined with the very short ma-no.)
  • We might, in the damaged space at the end, expect that the scribe had totaled the amount of bedding (vel sim.) distributed: to-sa de-mi-ni-ja 25 (τόσσα δέμνια, “so many beds”).  This could even have been done in .13, since it need not have run all the way to the right edge of the tablet.  Complicating this interpretation, however, is the fact that we have writing along the right edge, which is not common.  It is possible that the scribe reached the end and did not want to rule the opposite side to record only two lines of information; this assumes that line .13 was used normally and we have simply lost any trace of the 1 or 2 de-mi-ni-ja allocated (which is fully possible).  Alternatively, if .13 was used for a totaling formula, the scribe may have realised too late that he did not leave himself enough lines to record everything and total it, and so appended the information to the side.  But this is only so much speculation.

On a personal note, since my name is Theo(dore), this tablet has always held a special fascination for me — not least now because my fiancée’s name is Rose, so both of our names (in a sense) are recorded here.

Linear B Translated: KN Fp(1) 6, 7, 16, and 48.

This post (and others which will hopefully follow) is part of the initiative started by Dimitri Nakassis to make translations of Linear B tablets more widely available online.  While he so far has focussed on longer tablets, I wanted to take a look at some shorter ones, and offer in addition to translations the methods of reading by which we can arrive at interpretations.  To that end, I chose four tablets from the same series — they were found together, deal with very similar topics, and were all written by the same scribe.  Given these distinct similarities, we can be on quite sure footing when illuminating the ambiguities in one tablet by reference to one or more of the others; because the records can be so cryptic and abbreviated, this sort of contextual information plays a significant role in our interpretation of the evidence.

The series in question is Fp(1), a designation they received even before decipherment based on their ideographic content (now known to be OLE, the sign for olive oil).  The best known tablet in this series is Fp(1) 1, which includes such high-profile names as “Dictaean Zeus” and “Erinys”.  But it has received translation and discussion in every major Linear B publication, so I will largely pass it over (those interested in a recent discussion should see Joann Gulizio’s dissertation, pp. 201ff.).  What interests us here is that it is a “totalling document”, which collates and sum(marize)s other tablets, giving an overall total quantity of oil dispensed.

Many of the other tablets in the series are preliminary records, like Fp(1) 7:

.1 ka-ra-e-ri-jo , ‘me-no’ [
.2 di-ka-ta-de ,    OLE S 1̣[

.1 In the month of ka-ra-e-ri-jo
.2 To Dikte: ≥ 9.6 litres of olive oil.


  • me-no appears to be a genitive (μῆνος, menos), but the name of the month is almost certainly an adjective in the nominative.  On other tablets, the names of other months appear in the genitive, including Fp(1) 1 by the same scribe.  It may have been possible to refer to months in different ways (the “Marchy-month” or the “month of March”), or else the scribe was simply unscrupulous in his record-keeping.  This linguistic problem has no real bearing on our reading of the tablet, however.
  • ka-ra-e-ri-jo, like many words in Linear B, offers no obvious Greek interpretation.  This shouldn’t surprise — knowing only English, future scholars should make nothing of July either.  A link with Κλαριών (Klarion), the name of a month at Ephesus in the historical period, is possible but beyond falsification.
  • The tablet is damaged, which makes the exact quantity of oil difficult to determine.  Linear B counts using (essentially) an elaborate tally system, so while part of one mark is preserved, more could be lost.  This is where other tablets in the set can perhaps be of help; on FP(1) 1, which records the distributions for a different month, the amount of oil sent to Dikte is S 1 (or 9.6 liters).  In other cases it is clear that each recipient receives the same amount of oil in different months (so, for example, qe-ra-si-ja receives S 1 on both Fp(1) 14 and 48, in the months of a-ma-ko-to and wo-de-wi-jo respectively).  The context seems to suggest, at least, that only S 1 was sent in ka-re-ri-jo, and nothing is therefore missing.

With this tablet in mind, we can make sense of Fp(1) 6:

.1 ka-ra-e-ri-jo / pa-si-te-o-i S 1
.2 qe-ra-si-ja      OLE S 1

.1 <In the month of> ka-ra-e-ri-jo, for all the gods: 9.6 litres of olive oil.
.2 For qe-ra-si-ja: 9.6 litres of olive oil.


  • There is nothing in .1 to indicate that ka-ra-e-ri-jo represents the name of a month.  Given the shared syntax, scribe, and content with Fp(1) 7, however, we can safely provide it from that context.  It seems to be the case that Linear B tablets, especially preliminary records like this, were predominantly aide-mémoires for the scribes who wrote them.  If this was indeed the case, the records needed only be intelligible to one person, and such can could be understood in that light.  Of course, we could likewise write “June” instead of “In the month of June” and cause no confusion.
  • pa-si-te-o-i, “all the gods” (πᾶσι θεόhι, dat. pl.) seems to be a variant on a Near Eastern practice used when propitiating foreign gods.  We might therefore be seeing here evidence for the continuation of Minoan religion (or at least the worship of Minoan gods) into the Mycenaean period at Knossos; see further Gulizio and Nakassis 2014, pp. 123f.
  • qe-ra-si-ja admits of various interesting interpretations, but none is falsifiable.  The q- sound at the beginning would in Ionic dialects have become a t-; if aspirated, th-.  Therasia (Θηρασίαι, dat. sing.) is therefore a possible alphabetic reading; this could be linked to the word for beast, θήρ (ther), and thus to Artemis as πότνια θήρων (potnia theron, “mistress of animals”).  The use of an epithet alone to refer to a deity would be, however, without parallel in Linear B.  Since Artemis is also attested on other tablets under her own name, what we may be seeing here is a case of syncretism, where two formerly separate figures became joined to each other over time.  The name might alternatively be associated with Θήρα (Thera), ancient Santorini, or perhaps an early form of the name Teiresias (Kwερασίαι > Τερασίαι > Τερεσίαι > Τειρεσιαί), but the linguistic changes required for the latter are a bit suspect.  Ultimately, we can only say with certainty that we have no certain Greek interpretation.  Such agnosticism is, unfortunately, a large part of reading Linear B.

Next we have the very similar Fp(1) 16:

.1 wo-de-wi-jo , ‘me-no’ , pa-si-te-o-i S 1
.2    qe-ra-si-jo          OLE S 1

.1 In the month of the roses, for all the gods: 9.6 litres of olive oil.
.2 For qe-ra-si-<ja>: 9.6 litres of olive oil


  • We finally have a month with an intelligible name: Ϝορδήϝιος (Wordewios), derived from an early form of ῥόδον (rhodon), rose.
  • The reading qe-ra-si-jo is problematic.  This is its only attestation (hapax graphomenon), and it is very tempting to emend it to qe-ra-si-ja; the scribe may well have been distracted by the final -jo of wo-de-wi-jo directly above where he was writing.  Errors in the tablets are rare (there being no corruption introduced by later copying), but far from unknown.  Here’s a picture, with the two signs circled; it’s not hard to see how the top may have influenced the bottom in a moment of lapsed concentration: jo
  • But to reiterate: errors are rare, and our first instinct should not be to emend.  While qe-ra-si-ja is conventionally understood (as it has been here) to be a first declension noun, it might also be a feminine adjective, in which case qe-ra-si-jo would be the masculine equivalent.  But that qe-ra-si-ja appears so frequently in other tablets of the same series, especially Fp(1) 6 which is functionally identical, is a strong impetus to read this as scribal error.  But there is another tablet in this series which we must take into account:

Fp(1) 48:

.1       wo-de-wi-jo , ‘me-no’ / si-ja-ma-to    OLE    S    2
.2 pa-de ,    S  1     qe-ra-si-ja     S  1      pa-si-te-o-i      S    1̣
.3       a-mi-ni-so-de   , / pa-si-te-o-i       OLE    S    1

.1 In the month of the roses, for si-ja-ma-to: 19.2 litres of olive oil
.2 For pa-de: 9.6 litres of olive oil; for qe-ra-si-ja: 9.6 litres of olive oil; for all the gods: ≥ 9.6 litres of olive oil
.3 To Amnisos, for all the gods: 9.6 litres of olive oil

  • The relation of these two tablets to each other is not entirely clear.  Both record activity in the same month, and since Linear B records never seem to have been kept for more than a year (at least in clay), it seems likely that either these two should be combined or else Fp(1) 48 supercedes Fp(1) 16.  If that is the case, then we should certainly emend qe-ra-si-jo to -ja, since that is what we find on Fp (1) 48.
  • This tablet is of a different type than the others assessed so far, being both physically larger and recording more information.  It differs from Fp(1) in not providing a total amount of oil disbursed at the end, but in approach it seems closer to a record of total transactions for the month than a preliminary record such as Fp(1) 6, 7, and 16, which were perhaps written at the storeroom as the oil left.  But the fact it is clearly not a totaling document of the same type as Fp(1) must give us pause in thinking of this as an archival record.  It is possible it was simply a more elaborate sort of preliminary record, and was (or would have been) combined with Fp(1) 16 when the final record was made.
  • I am inclined to think that this was closer to a final, “archival” copy than a preliminary record, should be taken to supercede Fp(1) 16, and that qe-ra-si-jo therefore be emended.  There is no overlap in any of the other known preliminary records, and other tablets of this type show total numbers closely matching those in the totaling document Fp(1) 1.  That said, we do not have all the evidence, and nothing precludes the opposing interpretation.  Agnosticism again: Linear B is a game of relative probabilities.
  • Amnisos is one of the few place names recorded in Linear B that remains attached to the same place today.  The identification of proper nouns has played in a key role in the decipherment of many ancient scripts, and the story for Linear B was no different.  The identification of the sequences a-mi-ni-si-ja and a-mi-ni-si-jo with ethnic adjectives linked to Amnisos was a key stage for Ventris in beginning to fill out his grid, following on Alice Kober’s work in identifying signs with shared consonants but different vowels.

By way of a conclusion, I want to emphasize something about reading Linear B: the tablets are not purely philological documents, to be understood and edited on the same principle as, say, literary papyri.  They are also archaeological artefacts, and contextual information such as findspot and scribe allow us to associate ambiguous tablets with, and illuminate by way of, each other.  This sort of analysis, when done with tablets not so closely linked, becomes much less illuminating, if not downright misleading.  Only by marshalling the right evidence, in the right way, can we hope to make sense of these wonderful, enigmatic records.

The Theran Eruption: Chronology and Controversy

I’m a bit late to the game here, but the Late Bronze Age eruption of Santorini (ancient Thera) has been in the news again lately, in relation (as so often) to the controversy surrounding when, exactly, that eruption happened. This post is intended a necessarily simplified primer to that debate, in an attempt to identify the few things we do know for certain as well as outline the reasons for uncertainty underpinning the issue as a whole. It will conclude with an assessment of how the new study changes our understanding of the situation.

Chronologies: Absolute and Relative

When an archaeologist is talking about chronology, there are in fact two separate issues they may be discussing. The first is perhaps what most people would immediately assume: the absolute dating of an event, which is to say when it happened in calendar terms. This is what radiocarbon dates offer (at least within a band of uncertainty). In some areas of classical archaeology, inscriptions, coins, and written records can narrow this down to a given year, month, or even day.  In the Bronze Age Aegean, though, we generally lack this sort of granular data (Linear B records narrow things down to year, and even in some cases to a month, but they are too specific — and of course late — to have any bearing here). As such, to arrive at an absolute date in the absence of modern scientific methods, we have to chart correlations and exchanges with a region where the evidence for dating is much more abundant: Egypt.

There, the detailed knowledge of regnal lengths, as well as their correlation with astronomical events, allow for a fair degree of certainty in dating individual pharaohs. When finds associated with their reigns are found outside of Egypt, or Aegean imports are found alongside such items in Egypt, we are then in a position to give at least an earliest possible date (terminus post quem) for that archaeological context or period. They don’t really help with a lower boundary (terminus ante quem), because fine, inscribed objects such as faience scarabs bearing royal cartouches often became heirlooms, and could finally be deposited decades or more after their manufacture.

Unfortunately, only relatively few such items have been found in the Aegean, and most are certainly later than the eruption; the same goes for items of Aegean manufacture (or at least style) in datable Egyptian contexts. It is around these very few “fixed” points that the traditional (or low) chronology is constructed.  The crux of the controversy is that these dates, as currently construed, clash violently with the ranges of dates allowed for the eruption yeilded by radiocarbon dating of biological remains from the destruction layer. More on all this later.

The other type of chronology is relative, and here we can speak with much greater confidence. At its simplest level, this is simply stating whether one event occurred before or after another. When dealing with a single site, this is established on the principles of stratigraphy: each successive period of habitation (or destruction, or abandonment) at a site is reflected in successive layer (stratum) of dirt, the most recent on the top, the oldest on the bottom. Here, for example, is a section of the mound at Hisarlik (Troy):

3868DB6B-CC81-4C7F-A26B-B62674C1B63E(Picture taken from livius.org.)

The various strata are easy to see, and represent the relative chronology of the site: Troy I antedated Troy II, and so on and so forth. Also visible is the “shaving” of the mound that occurred in Roman times for the construction of the temple to Minerva (visible on the left). This destroyed all the evidence for Troys VI and VII at the top of the mound, which correspond to the layers most commonly associated with the Late Bronze Age destruction some would link with the Trojan War, and vividly illustrates why our understanding of those florescent periods is so patchy. Of course, for those who aren’t careful, the notion that the oldest layer will be the deepest can be dangerous: Schliemann cut through a large part of the sixth and seventh cities looking for his “historical” Troy, which he identified with the impressive but much-too-early second city. Only once Wilhelm Dörpfeld took control of the excavations at the end of Schliemann’s life did the truth of the situation become clear.

Working across sites, relative chronology becomes slightly trickier, but is still based on the same broad principles. Here, the extensive pottery record of the Bronze Age Aegean takes centre stage: if pottery associated with the same style is found exclusively in strata at two different sites, then (generally) we can say that those two strata are contemporary. While not all strata offer a clear enough picture, and regional styles and variations have to be considered, this sort of evidence provides quite a firm basis for cross-site dating.

As mentioned, the controversy concerning the chronology of the eruption is linked to the absolute date. The relative date, however, is prefectly clear, and has been since Spyridon Marinatos began excavations at Akrotiri on Thera in 1967. While it’s a journalistic cliché to call archaeological sites “the Pompeii of…”, in this case it is perhaps a forgiveable label, for Akrotiri was buried by the eruption in much the same way Pompeii was. It therefore provides us with a remarkable picture of the city on the very day of its destruction.  Since it was such a heavily Minoanized settlement (though its exact political relation to the palatial centres of Crete is unclear), the style of pottery and frescoes uncovered there should enable us to correlate those with a specific period of Minoan history.

And indeed they do: the latest finds from the eruption layer on Thera are all Late Minoan (LM) IA. For those who wanted to posit a direct link between the eruption and the so-called Minoan Collapse, this presented a grave chronological complication: whatever the cause of that collapse, it most certainly occurred at the end of LM IB, some fifty years later (at least). For a time, regional variations were cited as a possible explanation; could not provincial Thera, only a colony or emulator of Cretan customs, not have lagged behind the artistic developments of the main palatial sites? No: the discovery of Theran ash in LM IA contexts on Crete put paid to that theory. New finds continue to cement this, such as the recent discovery of tsunami damage at Palaikastro in Eastern Crete at the same chronological horizon (for an account that isn’t behind a paywall, see here).

The evidence is therefore overwhelming that the eruption occurred in LM IA, not LM IB, and therefore will not admit of any interpretation that posits a direct link with the Minoan Collapse. Whatever the absolute date may be, this cannot and will not change. Of course, indirect links are quite possible; the classic exposition of this hypothesis is Jan Driessen’s and Colin MacDonald’s The Troubled Island (1997 – Aegaeum 17; you can read a review by Ilse Schoep here and Driessen’s follow-up here). While this interpretation is not universally accepted, I think on balance of probabilities that the eruption was simply too big and too close to Crete not to have a lasting impact, but the evidence does not exist (and probably never will) to confirm that interpretation. And, while we are here, it bears stating that we are not dealing with the origins of the Atlantis myth here, and anyone who tells you otherwise is lying.

Absolute Chronology: The Two Camps

Before radiocarbon entered the picture, the limited evidence for interconnections with Egypt had been used to establish an apparently consistent chronology for the Aegean Bronze Age. By this interpretation, Thera erupted in c. 1500 B.C., during the reign of Hatshepsut in Egypt, and a short LM IB period followed, ending in c. 1450, early in the reign of Amenhotep II. The arguments put forth to support this dating are immensely complex, and to assess them fully requires a comprehensive grasp of Egyptian, Aegean, and Cypriot pottery styles.  While the data can be marshalled to create an image that is internally consistent, it should be stressed that there is no smoking gun, and indeed very few pieces of evidence that are truly unambiguous.

That said, we should not have much cause to doubt this interpretation if not for the radiocarbon data. This, quite explosively, pointed to a date around 1630 B.C. for the eruption, and was entirely incompatible with the traditional date of 1500 B.C. This, it should be re-iterated, had no bearing on the relative date: the eruption still must have occurred in LM IA. The consequence, therefore, is that LM IA must have been much earlier than we thought, and LM IB much longer.

This is the origin of the controversy. Various attempts have been made to cast aspersions on the specific tests used — the key evidence was, at first, an olive branch which (it was argued) may have been long dead at the time of the eruption, and (it is universally agreed) is not the ideal sort of wood for radiocarbon dating, since its annual growth rings are small and inconsistent. It was further argued that the eruption may have “tainted” it because of the quantities of carbon dioxide released. However, the early date was maintained by each successive test, and we now possess more floral and faunal remains associated with the eruption, including some from Crete, all of which support the early date when tested.

Because of this consistency, the archaeological evidence has come under increasing scrutiny, with successive articles, chapters, and monographs revealing the gaps of our knowledge and the paucity of unambiguous evidence. The same vase can support the low or high chronology, depending on whether we assume the coffin closest to the door is the earliest or latest in an Egyptian tomb — there seems to be no conclusive evidence. Broadly speaking, those who support the high chronology can read the evidence in such a way that it just about works if you squint, while those who support the low maintain that our understanding of radiocarbon is insufficient at this stage. This may seem old-fashioned, but it should be remembered that, in the early days of radiocarbon dating, when it clashed with the traditional Egyptian chronology, it was the latter that won out.

With that said, the possibility has emerged in recent years of a “compromise high” chronology, which would place the eruption in the early 16th century B.C.  This emerged from the realization that some evidence once through to corroborate the radiocarbon date, derived from tree rings and ice cores, did not quite hold up, and the fact the confidence intervals from various tests consistently allowed it. It also allows an understanding of the archaeological evidence influenced by knowledge of the radiocarbon dates (such as the “requirement” for a longer LM IB period than previously thought) to fit the picture in a more satisfactory manner. But the facts of the matter are not democratic, and a solution which makes the most people somewhat happy is not obviously true.

New Developments: Refining the Radiocarbon Data

Earlier this month, a new study was published in the journal Science Advances which refines the radiocarbon curve for the second millennium B.C. — you can read it here. As a consequence, the radiocarbon date ranges for the Theran eruption are moved away from the late seventeenth century B.C. and into the 16th. This is an important development, and seems to be the sort of “evolving understanding” that defenders of the low chronology insisted would come. Various media sites have in fact framed it in this manner. But this is not entirely the case. Here is a graph from the paper, illustrating the old date ranges (in pink) against the new ones (in blue):


While we have clearly moved towards the sixteenth century, the higher probabilities remain earlier rather than later. As the authors of the study put it: “We do note, however, that our data indicate that a date for the Thera eruption more recent than c.1510 BCE is highly unlikely, which remains at odds with certain archaeological arguments, and credible intervals do not exclude an eruption in the late 17th century BCE.” A table further illustrates that the new calibration brings the mean dates (variously derived) down by somewhere around 30 years in every case:


While almost every range is now much larger, and so admit of lower chronologies, the data still seem on aggregate to support the high chronology more than the traditionally low; indeed, the compromise high seems to come out quite well.

Two things, I think, bear noting in conclusion. The first is that this study does not support the low chronology to the exclusion of the old. The second is that it is far from the last word on the matter. Only the evidence from Thera has been re-calibrated; while I expect the effect should be similar on data from elsewhere, I do not know that this will be the case. There are also other records, such as the growth of stalactites and bristlecone pines that might help narrow down the range of possible dates. While we are undoubtedly moving closer to a full understanding of events, we are very definitely not there yet.

The Origins of the Greek Alphabet

A rather strange review was published by BMCR this morning: Barry Powell on Martin West’s The Making of the Odyssey. This post will be best read with it in view, though there is little to recommend it. It is (spoiler alert) negative in its outlook, which is unsurprising given the idiosyncratic nature of West’s views and the force with which he presented them. This is not the strange part. What is unusual is that it is not an assessment of West’s arguments as presented in the book (which one might naively assume to be purpose of a review) but rather spends a great deal of time explaining why West must be wrong based on Powell’s own views on the origins and purposes of the Greek alphabet.

A careful reader of West’s book and other publications will have cause at many points to question whether Powell could similarly be described, though his error in reporting the year of West’s death (2015, not 2017) is probably just a typo. But it is not my place nor goal to assess that part of the review. Rather, it is his statements about the origins of the Greek alphabet and the nature of writing systems that particularly caught my attention. The study of writing systems in the abstract is relatively young, and something that slowly grew to dominate an entire chapter of my Master’s thesis. Since Powell has hinged his review on these issues, and they are not something with which many classicists regularly grapple, I hope this post may prove fruitful.

Powell’s fundamental objection is that West “gives no consideration to the technology that made Homer possible, the Greek alphabet, what kind of writing this was, its position in the history of writing, how it came into being, when it came into being, where it came into being, and what it was used for from the beginning.”  He continues: “[West] seems unaware that the Greek alphabet was a very odd form of writing unprecedented in its attention to phonetic accuracy, which can more plausibly be explained as a technique created expressly for imprisoning in signs oral verse.” His ultimate argument is that “Homer” can be dated by the creation of the alphabet, because the Greek alphabet was invented to record the Iliad and Odyssey.

When Powell claims that the Greek alphabet was “unprecedented in its attention to phonetic accuracy,” what he really means is that it records consonants and vowels separately and consistently. This was an innovation. Earlier writing systems fell into two categories: syllabaries, which recorded consonants and vowels in conjunction (as in Linear B, which would render Θεόδωρος te-o-do-ro); and abjads, which recorded only consonants (so would render it θδρς). Into the former category fall the cuneiform script in its various manifestations (except the Ugaritic cuneiform abjad), Luwian Hieroglyphs, and the Aegean scripts: Linear A and B, Cretan Hieroglyphics, Cypro-Minoan and the later Cypriot syllabary. Abjads are represented by Egyptian Hieroglyphs, and from them Proto-Sinaitic and the very Phoenician script from which the Greek alphabet was derived.

At this point, we much consider what was known to the inventor (probably singular, though perhaps a committee) of the Greek alphabet. They were bilingual and literate in the Phoenician script (these conditions being a sine qua non of adaption). They realised that an abjad could never do for Greek what it did for Semitic; vowels were much too essential to both the declension of nouns and conjugation of verbs to be eschewed, to say nothing of their natural preponderance in Greek words. A solution was required.

Now, it is possible, if unlikely, that our inventor knew of syllabaries. They could not have known that Linear B was once used to encode Greek, but the Cypriot syllabary was in contemporary use. In any case they were in a poor position to develop one: to mark the 16 consonants they could borrow and the five invented vowels, they would require 85 signs from an exemplar of 20; if they wished to mark vowels for length, 170. The solution was the most economical and practical possible, based, not on a desire to encode verse, but the fact they had an abjad before them but needed a system of recording vowels. Very simply: take the signs of the Phoenician script and use those for which the Greek language had no phonetic equivalent to mark vowels.

The key question, therefore, is whether, as Powell claims, this most “plausibly [can] be explained as a technique created expressly for imprisoning in signs oral verse.” It is true that reducing the first line of the Iliad to consonants (as Powell has illustrated elsewhere) produces an impenetrable cipher: ΜΝΝ Δ Θ ΠΛΔ ΧΛΣ. But prose would be no less effected, as already noted: the systems of nominal declension and verbal conjugation would entirely collapse. A syllabary can, because it renders vowels, support them and thus enscribe the Greek language; an abjad cannot. That the system developed proved useful for the recording of Greek verse is an accident based on the fact it was useful for the recording of the Greek language.

Powell also makes some interesting claims about the nature of reading and writing in the early days of literacy. To be sure, it was only centuries later that Greek society embraced all the possibilities of this new technology. But I do not think the evidence supports the claim that the early Greek alphabet “could not be read with the eye, as scholars read Homer today, but was understood through the ear as the reader puzzled out and pronounced aloud the phonetic equivalents of the marks.” Exact phonetic representation was never a requirement for reading (or Linear B could never have functioned), and the notion of experienced readers “puzzling” over words like first graders is patently absurd. The way we read has not changed, except insofar as we now do much more of it.

In Powell’s favour, it is the case that verse inscriptions started appearing very early. But even earlier are simple names (as indeed is the case in most secondary scripts), and indeed these represent 73% of the earliest Greek inscriptions with an interpretable sense, three times as common as hexameter verses (none of which is Homeric). The potential for recording poetry certainly occurred early to the newly-literate Greeks. But there is no reason to think it came first, and it certainly was not the use to which the script was most widely applied. To claim that the alphabet was developed not just to record verse, but Homer’s verse, is a distortion that pays insufficient heed to the historical situation of “the Greek alphabet, what kind of writing this was, its position in the history of writing, [and] how it came into being”.


An Aethiopian at Pylos?

The decipherment of Linear B shed new light not only on the Late Bronze Age world it so elliptically recorded, but also on the development of the Greek language.  Reconstructed proto-forms featuring digamma and labiovelars were spectacularly confirmed, and a window was opened into a new historical dialect from centuries before the Homeric poems were put to writing.  But the tablets offer only brief snapshots of their world, and even when the linguistic data are clear, it can be tough to know what to do with them.  This is not a problem, in and of itself, but it becomes one when people try to use the simple fact of a word’s appearance in Linear B as evidence for some concrete aspect of Mycenaean society.  That we have a ϝάναξ does not mean we had Agamemnon, nor anyone who looked at all like him constitutionally.  As spectacular as the correspondence with the Iliad seems, anyone who wants to understand the Bronze Age on its own terms must be very careful about using evidence from the Homeric poems; likewise, the tablets are not always a sure guide to words in those later works.

Nowhere is this more clear than when we come to a man named a₃-ti-jo-qo, who held office and land at the Pylian district of Sphagianes.  Here he is, on PY Eb 846 (top row, first word):


(Photo taken from the CaLiBRA database and copyright the University of Cincinnati.)

While it is often difficult to be sure of names due to the elasticity of Mycenaean spelling rules, his is distinctive: the ‘doublet’ a₃ has to be the dipthong αι, and the final -jo-qo shows the typical treatment of a (labio)velar, which alone of Mycenaean consonants are recorded at the end of syllables alongside a ‘dead’ vowel, matching that used in the syllable before (-jo-qo).  Even the -ti-jo offers little room for misunderstanding, since dentals are (uniquely) marked for voicing in the syllabary (hence the voiced d- series and unvoiced t- series).  Given all of this, plus context which guarantees the nominative case, a reading Αἰθίοκʷς is all but assured, which corresponds exactly to historical *Αἰθίοψ following the loss of the labiovelar (the same general phenomenon is at play in πέντε, cognate with Latin quintus).

The problem here, often insufficiently addressed, is that we have simply no idea what to think Αἰθίοψ meant in the Bronze Age.  Etymology may be thought to help, but the question is vexed.  There is no doubt that, past a certain point, the historial Greeks interpreted the word as the LSJ records: “burnt-face”, from αἴθω, I kindle/burn, and *ὤψ, face (the certain but unattested nominative; the Homeric poems have it frequently in the accusative singular and compounds).  That this is the true etymology of the word is frequently asserted; Stephanie West, commenting on Odyssey 1.22, notes that it is “a properly formed Greek compound, and, despite some uncertainty about its derivation, the interpretation ‘with burnt face’ is the most probable.”  The uncertainty is greater than she allows.  The case against it was most recently and fulsomely laid out by Beekes (1995/6), though he is not as novel in his doubts as he claims:


So J. R. R. Tolkien in 1932.  But Beekes’ doubts are essentially the same: αἴθω means to ignite or to burn, not to be burnt, -οψ with a short vowel cannot mean face, and the -ι- is at any rate unexplained.  His article is not a masterpiece; many of his arguments strain credulity, and there is a nasty racial undertone in places, but the etymological discussion is sound.  Those with access to JSTOR may certainly profit by reading it in full, but it may be summarized:

  • There is no word with the root αἰθ- with a passive sense (i.e. ‘burnt’) except the rare αἰθός, which seems to have this sense when used by Aristophanes (Th. 427: a slave is singed, and exclaims “αἰθὸς γεγένημαι”, “I am burnt!”).  But in Pindar (Pythian 8.47) and the Iliad (in the form πάναιθος, 14.372) it must have the sense “shining”.  That it might have a sense closer to “burning” (should we imagine the slave’s skin glowing red?) in the Aristophanes passage is therefore a possibility.
  • The -ι- has been explained as the so-called Caland-i, a phenomenon in the Indic languages whereby -ρο- can alternate with -ι- in compounds.  Beekes does not believe this applies here; I will return to this point later.
  • There is no word in Greek where -οψ means face.  While it is true that we do not have the nominative singular *Αἰθίοψ preserved, so a form *Αἰθίωψ is possible, we do not see the ablauting pattern -ωψ, -οπ- in other compounds that definitely contain *ὤψ (so ἑλίκωψ, -ωπος) or in *ὤψ itself (acc. sing. ὤπα).  Since the ancients interpreted the word in light of *ὤψ, it is unlikely that they would have treated this word any different from other words with the same element.

The most fulsome rebuttal to these points is by Simon Pulleyn, in his edition and commentary on Iliad 1 (pp. 229-31).  He is not convinced by any of them, and ultimately defends the traditional interpretation.  As with Beekes’ article, it should be read in full by those with a deeper interest.  Summarizing again:

  • The active, transitive sense of αἴθω is not guaranteed in a compound; the epithet τερπικέραυνος given to Zeus can hardly mean “he who delights the thunderbolt.”  Further, “shining faces” could refer to the sheen of a black person’s skin, rather than the specific colour; the Aithiopes are listed alongside Libyans and the Μέλανες (literally “Blacks”) in a fragment of Hesiod’s Catalogue (150 M-W).  Aeschylus also has compounds where the verbal element must be passive (βλαψίφρων, “whose mind has been harmed”, Septem 725), so “burnt faces” might be possible.  We might also think of the first element as an adjective *αἰθι-; no such word exists, but αἰθός with the sense burnt is attested in Aristophanes (though cf. above; we shall return to this).
  • Given the adjective αἶθρος, there is no reason to think the -ι- can’t be a Caland-i.
  • In light of the unattested nominative singular, we might well imagine that it was *Αἰθίωψ, and the short vowel in the oblique stems is by analogy with ἡγεμών, -όνος vel sim. There would then be no problem deriving the second element from *ὤψ.

These are not arguments to be dismissed lightly, least of all by an amateur.  But there is, I will suggest, enough uncertainty to give us pause.

The first argument, that the verb can be taken intransitively, is I think convincing.  The sense should be “people whose faces shine/burn”, not the absurdity “people who burn their faces.”  The Aeschylean βλαψίφρων also suggests against being too dogmatic in claiming the first element cannot be taken passively, though it is a tragic hapax and from later than we should like.  As for “shining face” being taken as a reference to the sheen of their skin, this is largely unfalsifiable.  The Hesiodic associations adduced as evidence for black skin are, moreover, less than straightforward:

Αἰθίοπάς⌋ τε Λίβυς τε ἰδὲ Σκύ⌊θ⌋ας ἱππημο⌊λγού⌋ς. (Fr. 150 M-W, 15)

“He saw the Aithiopians and Libyans and mare-milking Scythians”

Though linked with the Libyans, the presence of the Scythians speaks against too straightforward a geographic interpretation here.

. . . . . . . .] Μέλανές τε καὶ Αἰ[θ]ίοπες μεγάθυμοι
ἠδὲ Κατου]δαῖοι καὶ Πυγμαῖ[οι] ἀμενηνοὶ
. . . . . . . .] κρείοντος Ἐρικτύπου εἰσὶ γενέθλης. 
(Fr. 150 M-W, 17-9)

“… the Blacks and the greathearted Aithiopians,
the Subterraneans and feeble Pymies
… are descended from mighty Poseidon.”

Here the groupings are genealogical, so again the association of the Aithiopes with the Blacks is not necessarily meaningful (the existence of the Blacks as a separate people does not guarantee that the Aithiopes could not also have had black skin, but it certainly does not favour that argument).  Moreover, the one character in Homeric poetry who is almost certainly black, Eurybates, Odysseus’ herald, μελανόχροος, οὐλοκάρηνος (“black-skinned and wooly-haired”, Od. 19.246) has nothing at all to do with the Aithiopes.

As for a potential adjectival root *αἰθι-, this is possible but again not immensely likely.  If it existed, we should still expect it to mean “shining/burning”, as αἰθός did before Aristophanes (and perhaps even there).  The usage of αἰθός in Aristophanes is also late – certainly after the Aithiopes were identified with the historical Ethiopians, and the sense “burnt” for αἰθός may then be derived from the very folk etymology it is now used to support.

I do not mean to suggest that none of Pulleyn’s suggestions is possible; I merely wish to stress that there are grounds for doubt.  Moreover, any explanation of one element is still contingent on the other two being likewise explained.  We must walk a very fine line.

The classification of the -ι- as a Caland-i relies on the adjective αἶθρος, and moreover depends on this being an adjective in -ρο- form from the stem αἰθ-.  The trouble is that there are many derivatives of αἰθ- that feature ρ, and it may be better grouped with them.  The noun αἰθήρ means “clear, bright sky” (so Beekes), and αἶθρος is almost certainly more closely related to this than αἴθω.  The word only appears in a clear context once:

… τοῦ γὰρ φίλος υἱὸς ἐπελθὼν
αἴθρῳ καὶ καμάτῳ δεδμημένον ἦγεν ἐς οἶκον,
χειρὸς ἀναστήσας, ὄφρ᾽ ἵκετο δώματα πατρός. (Od. 14.317-9)

“… for his son came to me
brought low by cold and weariness, and taking me by the hand
led me homewards, until he reached his father’s halls.”

The development of the sense is traced by Arie Hoekstra, commenting on 14.318: from “clear sky” it came to mean the attendant temperature; in winter, of course, the coldest days are the clearest:

Cold(Aἶθρον ἦμαρ ἐν Ϝιννιπέγι.)

The relation to αἰθήρ is further supported by the adjective αἴθριος, which refers to a clear sky in Herodotus.  It seems unlikely, therefore, that αἶθρος is derived directly from αἴθω in the way that required for the Caland system to be at work in Αἰθίοπες.  Doubt must remain.

Pulleyn’s final argument is quite reasonable.  The ablauting system that yields a long vowel in the nominative stem but a short one in the oblique forms is incredibly common in Greek, and even Beekes admits that we should expect it in a compound of this form. A nominative *Αἰθίωψ is thus highly possible.  But, unhelpfully, other nouns with this element have in fact standardized the long vowel across all forms, so ἑλίκωψ, -ωπος, not ἑλίκωψ, -οπος.  I find it hard to explain why a noun thought to contain the same element would be treated differently, either maintaining the original pattern or else remodeled on analogy with a completely different word like ἡγεμών, -όνος.

Where does this leave us?  Pulleyn, I think, presents a strong case that the traditional interpretation is not as untenable as Beekes would have us believe.  It is certainly possible.  But it relies on a confluence of factors that are perhaps more more possible than probable. To return to the original question: where does this leave our friend a₃-ti-jo-qo?  The answer must be: in a state of some uncertainty.  Etymology cannot provide a sure guide.

Can we turn to Homeric poetry for recourse?  The answer, of course, is yes, but this is a solution of despair.  It is no sure guide to the Bronze Age.  And here we may be especially sure that he has nothing to offer, for the Homeric Αἰθίοπες are not citizens of the world but dwellers on Ocean, who truck rather with gods than mortals:

Ζεὺς γὰρ ἐς Ὠκεανὸν μετ᾽ ἀμύμονας Αἰθιοπῆας
χθιζὸς ἔβη κατὰ δαῖτα, θεοὶ δ᾽ ἅμα πάντες ἕποντο. (Il. 1.423-4)

“For Zeus went yesterday to Ocean, to feast among
the noble Aithiopes, and all the gods with him.”

… εἶμι γὰρ αὖτις ἐπ᾽ Ὠκεανοῖο ῥέεθρα
Αἰθιόπων ἐς γαῖαν, ὅθι ῥέζουσ᾽ ἑκατόμβας
ἀθανάτοις, ἵνα δὴ καὶ ἐγὼ μεταδαίσομαι ἱρῶν. (Il. 205-7)

“… For I will go to the land of the Aithiopes
on the shores of Ocean, where they make hecatombs
to the gods, so I too may share in the feast.”

ἀλλ᾽ ὁ μὲν Αἰθίοπας μετεκίαθε τηλόθ᾽ ἐόντας,
Αἰθίοπας τοὶ διχθὰ δεδαίαται, ἔσχατοι ἀνδρῶν,
οἱ μὲν δυσομένου Ὑπερίονος οἱ δ᾽ ἀνιόντος,
ἀντιόων ταύρων τε καὶ ἀρνειῶν ἑκατόμβης. (Od. 1.22-5)

“But [Poseidon] is visiting the Aithiopes far away,
The Aithiopians, a divided and distant people,
Who live, half at the sun’s setting, half at its rising,
And offer hecatombs of bulls and rams.”

This, it should be clear, can tell us nothing about a historical person; if anything, it may be clear that in these poems the Aithiopes are not, in fact, conceived of as a historical people.  We have already seen what little Hesiodic poetry has to add.  It is certainly irresponsible to use the appearance of the name in Linear B to flesh out the picture, as all too many have.  So again Stephanie West’s comment on Odyssey 1.22: “Negroes are depicted in frescoes from Cnossus and Thera… So the Mycenaeans must have had a word for ‘negro’, and there is nothing against supposing this to have been the original meaning of Αἰθίοψ.”  But this is surely disingenuous.  We must imagine that the Mycenaeans met black Africans and coined a name for them, “Burnt Faces,” that transparently meant black-skinned.  Then, we must suppose, both the fact that they were real people and the transparent meaning of their name were forgotten.  Much was lost in the aftermath of the Mycenaean collapse, but an understanding of the Greek language was surely not among the casualties.  After an interval of many centuries, the Greeks must then have encountered black Africans again and suddenly remembered they had a name for them which they had most spectacularly misplaced.

The absurdity of this situation is all the more remarkable for the popularity of its variants.  Here is Wolfgang Kullmann in 2005 (p. 15): “Does not the etymology of the name Aithiops, “burnt face”, and its Mycenaean attestation, suggest that a realistic geographic knowledge of people with black skin was originally responsible for the name?” And here is Bruno Currie in 2016 (p. 60, n. 130): “Although the Ethiopians are removed from the world of the heroes in the Iliad, this does not necessarily reflect an older strand… the personal name Αἰθίοψ is found in Mycenaean.”  All of this presupposes a great deal about the rather over-taxed a₃-ti-jo-qo.  The fact is that we do not now enough about the Aithiopes of early myth to suggest what the word might have meant in the Bronze Age, and we do not know enough about what the word meant in the Bronze Age to illuminate its meaning in early myth.

The appearance of a word so explicitly linked with the Greek mythic tradition in Linear B is always fascinating, and we cannot rule out that a₃-ti-jo-qo was indeed an Aithiop as the later Greeks came to understand the word.  It would not do to rehash Beekes’ argument (1995/6, p. 29) that a black African could not have risen to the status of a₃-ti-jo-qo in the Bronze Age Peloponnese.  This assumes knowledge of Mycenaean race relations that we simply do not have, and can only be stated on anachronistic (if not racist) grounds.  But the word Αἰθίοψ is elusive, and no context shines brightly enough to illuminate another.  So we are as Tantalus, endlessly enticed by remarkable possibilities that must remain ever beyond our grasp.  Such is the joy and frustration won by the decipherment of Linear B.



Beekes, R. (1995/6). “Aithiopes.” Glotta 73, pp. 12-34.

Currie, B. (2016). Homer’s Allusive Art. Oxford: Oxford University Press.

Heubeck, A. and A. Hoekstra (1989). A Commentary on Homer’s Odyssey, Vol. 2: Books IX-XVI. Oxford: Oxford University Press.

Heubeck, A., S West, and J. Hainsworth (1988). A Commentary on Homer’s Odyssey, Vol. 1: Books I-VIII. Oxford: Oxford University Press.

Kullmann, W. (2005) “Ilias und Aithiopis.” Hermes 133, pp. 9-28.

Pulleyn, S. (2000). Homer: Iliad I. Oxford: Oxford University Press.

Tolkien, J. R. R. (1932). “Sigelwara Land.” Medium Aevum 1, pp. 183-96.