What punctuation?

I recently listened to an edition of Radio 4’s “Word Of Mouth” programme, in which veteran presenter Michael Rosen invited punctuation expert Florence Hazrat to explain “how punctuation began”.

It’s an interesting question. And if anyone can be expected to answer it, it’s surely Dr Hazrat, whose publisher calls her “one of the world’s foremost authorities on the history and culture of punctuation”.

I was intrigued by her opening remarks, informing us that the ancient Greeks and Romans had no punctuation at all, until the concept was invented by Aristophanes of Byzantium in the third century BC. Suddenly, it seems, western writers had punctuation.

For context, the work of this Aristophanes has been preserved only in fragments, but it seems that he was particularly concerned with Greek texts of the major playwrights destined for the great Library at Alexandria, of which he was a custodian.

So where might we see this marvellous new punctuation?

Here on the left is a Greek papyrus, known as P. Bingen 45. It was written in 33 BC at the court of Queen Cleopatra, around two-and-a-half centuries after Aristophanes’ “invention” of punctuation.

But there’s no punctuation in evidence at all.

Ah, you may say — but Aristophanes was dealing with poetry, not court correspondence. Maybe his punctuation scheme was adopted only for poetry?

Here on the right is a papyrus (P. Mich. Inv. 6972) from Karanis in Egypt, dating from the second century BC, so again after the time of Aristophanes of Byzantium.

It preserves a section of Homer’s Iliad (specifically Book 10, with lines 421–424 on the left and the beginning of lines 445–460 on the right).

The important point to note is that, again, there is no punctuation in evidence at all. The words of each sentence are simply run together.

Of course, in an inflected language like Greek, it is rather more straightforward to pick out where words begin and end. There is little chance that the reader would encounter any ambiguity of the “eats shoots and leaves” sort.

Not so, apparently, for the Romans.

Seneca the Younger seems to allude to punctuation in the first century AD. Deploring the breakneck delivery of the Greek orators, he wrote that “we Romans, even when writing, have become accustomed to separate our words” (Epistles 40.11). But, far from taking their cue from Aristophanes, it seems that this had always been a Roman tradition.

Take the so-called fibula Praenestina, a gold inscribed brooch discovered at Palestrina (ancient Praeneste), near Rome. It is thought to date from the seventh century BC, thus around three hundred years before Aristophanes “invented” punctuation.

The text is difficult to read, because it has been composed in reverse, right-to-left. But the use of dotted word-dividers similar to colons can clearly be seen separating the first word MANIOS (thought to be the craftsman’s name) from the next word MED, and elsewhere in the sentence.

Of course, as every epigrapher knows, monumental Latin has a history of using interpuncts to separate words, usually abbreviated words, in order to simplify the act of reading. This can be seen as early as 270 BC, with the sarcophagus of Lucius Cornelius Scipio Barbatus (CIL I² 7), which was engraved probably a little before the time Aristophanes was “inventing” punctuation.

All of which shows that we shouldn’t get too excited about Dr Hazrat’s claim that the Romans had no punctuation until a Hellenistic Byzantine scholar invented it for them. Quite the contrary. But nor should we assume that the Greeks suddenly adopted a fully fledged system of punctuation in around 250 BC. Clearly, they didn’t.

Postscript

Florence Hazrat (in her book On the Mark: A History of Punctuation) cites no reference for attributing the invention of punctuation to the Hellenistic grammarian Aristophanes of Byzantium. This is frustrating. So much so that I decided to go to the source in search of an answer: the very writings of Aristophanes himself, preserved in ancient Greek, and unfortunately not translated into any modern language.

Imagine my surprise to find that the answer lies not in the fragments of Aristophanes’ writings, but somewhere entirely different. Grammarians will, no doubt, know all of this (but it surprises me that none of the ancient texts seems to have been translated).

The answer lies, in fact, in a work “On accents”, previously attributed to one Arcadius of Antioch (ed. E.H. Barker, 1820), though some prefer Theodosius of Alexandria. Specifically, Book 20 Περὶ τῆς τῶν τόνων εὑρέσεως καὶ τῶν σχημάτων αὐτῶν:

“The system of vowel lengths, pitch accents, and breathing marks established by Aristophanes was devised both to distinguish words of ambiguous meaning and to guide the melody and harmony of the voice, as if we were singing while we spoke.
“Consider, too, how each of these elements was naturally and aptly named and characterized, much like musical instruments, since they were destined to serve as instruments for speech itself. For he observed that music, too, indicated melody and rhythms in this way, now relaxing and now tightening, and designating some notes as low and others as high. And whenever we sang, either tightening the pitch to the limit or relaxing it, he called this ‘harsh’ and ‘gentle’.
“In this regard, Aristophanes established these initial marks for the written word, so that a standard, a sign of correctness, would accompany the formation of syllables and words. He then divided the movement of the voice into three aspects: duration, pitch, and breath. He likened duration to rhythm, and pitch to musical tones. He assigned specific signs and names to each.
“For the durations, he designated them as ‘short’ and ‘long’ and devised appropriate shapes: a straight, extended line for the long, and a curved line for the short, appearing to enclose the sound from both sides. For the accents, he named the one that rises straight upwards and terminates in a sharp point resembling a launched arrow the ‘acute’, and the one opposite to it the ‘grave’. And since, in the case of circumflexed words, a high-pitched sound is heard at the onset of the utterance, subsequently shifting towards a low pitch, being nothing other than a blending and fusion of both the high and the low, he concluded that the ‘circumflex’ was composed of these two elements, and devised its symbol accordingly.
“He also devised marks for the spoken word, distinguishing between words that are run together, those that are compounded, and those that are separated, such as assigning the mark now called the ‘apostrophe’ to run-together words, the ‘hypodiastole’ to words separated from one another, and the straight line that we call the ‘hyphen’ connecting compounded elements in joined words. The shapes of these marks, along with the others, will be illustrated shortly.
“Furthermore, he ordained that the stigmê should serve not only the word but also the sentence, providing pauses and intervals for the breath and the voice, and he assigned them specific positions and functions. ...”

So now we know.

Nothing new at Masada

Round and round we go. Like a merry-go-round. Academics seem desperate to publish, and if they have nothing new, they pretend they have something new.

I have just read “Le siège de Masada (73–74 après J.‐C.)” by Sophie Krausz, a Professor of Archaeology at the Paris Sorbonne. (The paper appears in HiMA. Revue internationale d’Histoire Militaire Ancienne 14, 2025, 323–342.)

It is, firstly, noticeable that Professor Krausz has chosen to refer to the work of Josephus using the (very) old-fashioned book–chapter–section method, rather than Niese’s (otherwise universally employed) book-and-verse method. Thus, her reference to “Guerre des Juifs VII, 8, 5” (for example) could apply to any one of the sixteen sentences (304–319) that Niese divided this long passage into.

Be that as it may, after some lengthy scene-setting — and confusion over dates (the article title implies a lengthy siege spanning AD 73–74, but the text plumps squarely for 73, although a footnote mentions my 1988 suggestion of 74 without discussion) —, Professor Krausz turns to the Roman circumvallation, for which she uses the peculiarly French term contrevallation, coined in the nineteenth century by Napoléon III. That is certainly her choice to do so.

But she also suggests that the Roman siegeworks utilized an earlier enceinte built by Herod the Great; something for which there is not even the slightest evidence. This is, I think, rather more problematic.
“Hérode aurait pu en effet aménager au cours de la phase 3, non seulement un rempart de couronnement au sommet du rocher, mais également une enceinte basse de contour pour protéger et contrôler l’ensemble des accès à Masada”. (“As part of Phase 3, Herod might have built not only a crowning rampart on the summit of the mountain, but also a low boundary wall to protect and control all access to Masada”. Note that Israeli archaeologists have divided the construction of Herod’s palace into three phases, the last of which they date to 15 BC.)

Her reasoning seems to be: (a) modern scholars have suggested between one and two weeks for the construction of the Roman circumvallation; (b) this seems to her to be too short a time; (c) so there must already have been an existing wall, which required only refurbishment by the Romans.

Professor Krausz then turns her attention to the Roman “rampe d’assaut”, a frequent feature of Roman sieges known as an agger or “embankment” (χῶμα, “heap” in Greek). Here she claims to present “une nouvelle reconstitution de la rampe d’assaut construite contre le rempart à casemates couronnant le sommet de la citadelle” (“a new reconstruction of the assault ramp built against the casemated rampart crowning the summit of the citadel”), along with this diagram.

It doesn’t seem to be new, and the Professor doesn’t indicate which parts she thinks are new. Compare her graphic (above) with this picture (left), which is page 176 of my 2006 book Besieged (based on my 2002 Ph.D. thesis, entitled “Aspects of Roman Siegecraft”).

The only difference that I can see concerns Professor Krausz’s pile of logs (are they logs?) at the top of the embankment, which she has labelled “cavalier”, another peculiarity of nineteenth-century French vocabulary. (Incidentally, I recently discussed the terminology of Roman siegecraft in Brill’s 2019 Companion to Sieges in the Ancient Mediterranean, but this book is not included in Professor Krausz’s bibliography.)

The Professor doesn’t explain this log pile: “Nous n’avons pas d’indications sur sa morphologie” (“we have no details of its morphology”), but she believes that this is Josephus’ ἐπιβάθρα (epibathra, “means of approach”: B.J. VII.307). My reading of Josephus differs from hers. Josephus here describes the entire embankment as the “means of approach”, and singles out “the top” as a βῆμα (bêma, “platform”). It is unclear whether, by “top”, he means the highest extremity or indeed the entire upper surface, and any discussion of the Roman siege really ought to explore this.

Professor Krausz chooses not to. Instead, she simply states: “un assemblage de grosses pierres, ce cavalier mesure 50 coudées de côté et autant en hauteur, soit 22m” (“an assemblage of large stones, this cavalier measured 50 cubits in breadth and the same in height, being 22m”). Ah, not logs but stones.

It is not easy to see how a siege tower might negotiate the final obstacle posed by this 22m-high platform. Professor Krausz tenatively suggests that “il faut peut-être envisager que la tour d’assaut ait été construite au sommet de la rampe, en totalité ou en partie” (“it may be necessary to imagine that the assault tower was built at the summit of the ramp, in whole or in part”). But couldn’t they just have built a taller tower?

And that’s it. End of discussion. In conclusion, Professor Krausz ponders that “la compréhension du siège de Masada passe par une étude minutieuse des données textuelles et archéologiques puis par la confrontation de l’ensemble des connaissances” (“an understanding of the siege of Masada requires a careful study of the textual and archaeological evidence and then a comparison of all knowledge”). An admirable goal, but not one that is represented by this article.

Gallienus’ cavalry resurrected

Today, I discovered a twenty-two-year-old Japanese paper entitled “Regarding Emperor Gallienus’ Cavalry Reforms”.

In brief, the author, Professor Fuminori Inoue of Kyoto, attempted to reconcile two opposing theories: on the one hand, Emil Ritterling’s famous 1903 proposal that the late Roman cavalry units designated as Illyriciani had originated in a special type of independent cavalry (“ever-battle-ready”) created, he conjectured, by Gallienus; on the other hand, Hans-Günther Simon’s 1980 criticism, suggesting that Gallienus had simply fielded “strong cavalry units” alongside legionary vexillations in mixed battle-groups.

I discussed both of these scholars’ ideas (amongst many others) in my 2025 book Phantom Horsemen.

A hybrid fantasy

Professor Inoue admits, in his 2004 paper, that Gallienus’ successor, Claudius II, had multiple cavalry units and explicitly not a unified cavalry force, so he casts doubt on whether Gallienus’ general Aureolus could have been “commander of the entire cavalry” (as the Byzantine writer Zonaras alleges). In my book, I presented what I think is rather an elegant solution to this dilemma, but Professor Inoue ’s rather less nuanced answer is to assert that “the entire cavalry” simply referred to whatever cavalry units happened to be under the emperor’s direct command at the time.

He argues that there was no unified cavalry force under Aureolus’ command, since Zonaras records that Gallienus’ general Aurelian was also in command of cavalry. But, hold on — wasn’t it Zonaras who told us that Aureolus was indeed “commander of the entire cavalry”? Confused? No wonder.

Of course, Professor Inoue follows the herd in his refusal to relinquish the “fact” of Gallienus’ cavalry reform. (Professor David Potter’s Roman Empire at Bay, with its “Gallienic cavalry corps”, appeared in the same year.) But it doesn’t sit well with Inoue’s ideas. His solution is to claim that the “commander of the entire cavalry” (a phrase also used by the late writer Zosimus) was simply whoever grouped a few units together at a given time on an ad hoc basis. Only thus is he able to make sense of the even later writer Cedrenus’ (clearly erroneous) remark, that Gallienus invented cavalry units (plural), not a unified cavalry force (singular).

Myth upon myth

Professor Inoue is convinced that the Roman army absolutely required highly-mobile independent cavalry forces that could suddenly be deployed wherever the emperor wished. (I have questioned the logic of this in my book Phantom Horsemen.)

There is no evidence that this ever existed, so he creates a parallel myth of highly-mobile independent infantry forces under the guise of legionary vexillations — he cunningly gives the false impression that, rather than deployed to perform a particular task at a particular location (as all vexillations were), legionary vexillations were, in reality, roving armies in their own right. Hmmm.

And in a final bizarre twist, he argues that this reform of Gallienus was aimed at creating a permanently functioning mobile army composed of both cavalry and infantry forces, whose powerful officers governed the empire.

To sum up, Professor Inoue proposes that Gallienus did indeed create a cavalry army, but it was an ad hoc grouping of separate units, which somehow became combined with mobile infantry units to create mobile field armies for deployment on widespread battlefronts.

Bringing it up-to-date

This twenty-two-year-old farrago of misplaced ingenuity only came to my attention because of the XIV Convegno Romeno-Italiano 2025 held in Iaşi (Romania), the proceedings of which have just appeared as Volume 21/1 of the journal Classica et Christiana.

I was intrigued by a paper entitled “El ejército romano de Aureliano y Probo en la Historia Augusta”, by Miguel Pablo Sancho Gómez of the Catholic University of Murcia.

The summary of his paper implied an investigation of how the emperor Probus (AD 276–282) capitalized upon Aurelian’s successful military reunification of the empire (AD 270–275), drawing upon evidence from the Historia Augusta.

Crisis and cavalry

Browsing through the paper, my attention was drawn to a four-page section entitled “Crisis of the Third Century and the Rise of the Roman Cavalry”. The scene is set with the usual hyperbole of “fifty years of assassinations, rebellions, secessions, plagues, invasions, and usurpations”. (I have put this kind of pessimistic exaggeration into perspective in my book.)

We then read that, because Rome was where the emperor was, swift and rapid campaigning was required, which in turn made cavalry essential. (I have examined this flawed logic in my book, but you can probably see where this is going ...)

The result was inevitably “the creation of new, large-scale formations of cavalry, previously a relatively secondary arm, which abandoned its almost exclusive auxiliary role in liaison, scouting, and skirmishing to spearhead many of the attacks of the Roman army”. This cavalry also became, we are told, “the emperor’s armed retinue, a training school and an officers’ academy for extracting experienced commanders who enjoyed total confidence owing to their affiliation with the sovereign’s entourage”.

And what is the evidence for this cataclysmic transformation? Why, the 2004 paper by Professor Fuminori Inoue (mentioned above), of course!

Once more the Schlachtenkavallerie

Although all of this mythology has been debunked in Phantom Horsemen, Señor Sancho Gómez may be excused for having missed it. His bibliography is predominantly pre-2010, though he lists Michael A. Speidel’s paper on “The Military in the Third-Century Roman Empire”, which appeared at the same time as Phantom Horsemen.

But he should surely be aware of his fellow Spaniard José Antonio Magdalena Anda, whose book (pictured here) El emperador Galieno y la supervivencia del Imperio romano (Madrid 2022) at least deserves to be cited, particularly in a Spanish discussion.

Like Sancho Gómez, Señor Magdalena Anda seems convinced of “the fact that Gallienus created a cavalry corps at some point between 255 and 260” (El emperador Galieno p. 265).

In his recent, broadly positive review of my Phantom Horsemen book (mentioned here), he doesn’t commit himself on the subject, appearing content to note that I have “identified several (not all) researchers who reject Gallienus’ battle cavalry thesis”. These are, of course, the writers who (like Professor Fuminori Inoue) embrace some form of Hans-Günther Simon’s equally flawed solution.

Señor Magdalena Anda, in fact, seems to equate my position with the one that Yann Le Bohec adopted in his 2009 book L’armée romaine dans la tourmente (a book that I read only recently, as explained here). But Le Bohec mentions neither Ritterling nor Cedrenus in this book. How, then, can he address an ongoing argument that has those two authorities at its root?

Round we go again ...

Let’s see what Professor Le Bohec actually does say. He tells us, first, that cavalry would not have been the answer to any problems experienced by the third-century Roman army. (Note that this is not the same as disproving the prevalent theory that it was the answer.) Secondly, he tells us that the notion of a “mobile army” is absurd, because no army is “immobile”. (Note, again, that this is not the same as addressing the prevalent theory that Gallienus developed a roving, ever-ready task force.)

On the other hand, he tells us that Gallienus created new auxiliary units, “some recruited from the Dalmatae and the Moors, others called scutarii and stablesiani” and a special guard of protectores, and he increased the legionary cavalry from 120 men to 726. I notice that Professor Le Bohec gives no clear references for any of this. (Gallienic scutarii and stablesiani, at any rate, are a theory of M.P. Speidel’s, and H.M.D. Parker can be blamed for the legionary cavalry. I’ve dealt with the protectores in my book.)

Señor Magdalena Anda, for his part, seems to favour the mixed battle-groups theory, though he does at least address the evidence. But predictably, he has not yet shaken off the fundamental myth that equates Gallienus with cavalry in some form. No doubt it will take a while.

Odyssey redux

I recently read an article in The New Statesman claiming that last year’s The Odyssey – Homer by Daniel Mendelsohn “will probably be the authoritative translation for years to come”.

Having finally made its acquaintance, I can see why some might believe this, but how can they avoid the Emily Wilson steamroller? For example, when questioned about “The Odyssey”, Google’s AI Overview doesn’t even mention Mendelsohn. It certainly should.

Sheer poetry

Mendelsohn has gone to enormous lengths to render The Odyssey in as faithful a version as possible, maintaining something of the feel of the original, both rhythmically and stylistically.

Not only has he retained the six-beat rhythm of the original dactylic hexameter in his line-by-line version, but he makes every effort to reproduce Homer’s alliteration and assonace. The effect is simply wonderful.

My favourite epithets

In stark contrast to Wilson (pictured here), Mendelsohn has also realized the crucial importance of Homer’s repeated formulaic epithets — “Odysseus, that man of great cunning” (πολύμητις Ὀδυσσεύς, polymêtis Odysseus), for example, or “Penelope, clear-thinking woman” (περίφρων Πηνελόπεια, perifrôn Pênelopeia) — epithets that Wilson mostly omits entirely in her drive to simplify.

Omits them, or elides them into her narrative. For instance, in conversation with Kalypso, Mendelsohn’s “ever-inventive Odysseus” (πολυμήχανος Ὀδυσσεύς, polymêkhanos Odysseus) is simply “Odysseus”. And when “Odysseus of the many ruses replied to her in turn” (Mendelsohn’s six-beat line — try it), Wilson simply reports that “Odysseus with tact said ...”

Omits, elides, or, worst of all, twists them, as when Mendelsohn’s “Odysseus, that man of great cunning” (see above) becomes, for Wilson (in the opening line of Book 9), “wily Odysseus, the lord of lies”! (She really doesn’t like her protagonist.)

But then, in the same Book 9, “the Cyclopes, an arrogant, lawless lot” (in Mendelsohn’s translation) become, for Wilson, “high-minded Cyclopes, the mavericks”. (Homer’s two adjectives are ὑπερφίαλος — hyperphialos, “arrogant” — and ἀθέμιστος — athemistos, “lawless”, just as Mendelsohn wrote.) Many more of Wilson’s anomalies could be mentioned.

Interestingly, Mendelsohn (pictured here) argues for “Athena of the bright owl-eyes” (γλαυκῶπις Ἀθήνη, glaukôpis Athênê), which I have always believed should be “grey-eyed Athena”. Mendelsohn conjectures that γλαυκός (glaukos, “shining”) is derived from γλαύξ (glaux, “owl”), and refers to that creature’s unnerving stare. I’m not entirely convinced — the logic seems to miss a step —, but I’m willing to give him the benefit of the doubt.

Homer’s text

My thoughts also turned to the source text, not only for Mendelsohn and Wilson, but for all the many translations and versions that fill our bookshops. Whenever I read an ancient text, I like to know whose authority I am relying upon. This is important when dealing with texts that have few source manuscripts: the discovery of a new one can throw everything into disarray. However, The Odyssey (like that other venerable text The Holy Bible) is based on a myriad of original manuscripts, ancient and medieval (the Oxford editor Thomas Allen’s list, published in 1908, runs to two-and-a-half pages). So the appearance of a variant at this late date is vanishingly unlikely.

Meanwhile, as pundits of all kinds seek to benefit from the appearance of Christopher Nolan’s blockbuster movie this month, the University of Manchester’s John Rylands Library have taken the opportunity to reveal their ownership of a truly ancient copy of The Odyssey: An incomplete parchment codex from around AD 300, in which 79 of the (estimated) original 207 leaves have survived with varying levels of damage. Initially acquired in Egypt at the turn of the twentieth century and published in 1911 as P. Rylands I.53, the codex, containing parts of Books 12–15 and 18–24, was probably the second of a two-volume set — ancient precursor of the Loeb edition!

The Manchester Odyssey

Pictured here is the front face (or recto) of folio 81. Carrying lines 140–171 of Book 22, it matches a single page of the Loeb edition almost exactly.

We are near the end of the story. The scene is the megaron hall of Odysseus’ palace in Ithaka. Having revealed himself by stringing the mighty bow and shooting an arrow through the axe-heads, the hero turns on the crowd of suitors. His son, Telemachus, brings weapons for himself, his father, and two loyal bondsmen (the swineherd Eumaeus and the cowherd Philoetius). But he has left the door to the storeroom ajar.

As this page begins, Melanthius the disloyal goatherd runs to Odysseus’ storeroom (you can see the name Ὀδυσεὺς in the middle of the second line — Melanthius’ name is below it in the third line) and seizes twelve sets of weapons for the suitors.

Eumaeus spots him returning for more weapons and asks Odysseus whether he should try and kill him or capture him. On the last line of the page, Odysseus replies that he and Telemachus (you can see the name Τηλέμαχος halfway along the bottom line) will keep the suitors occupied, while the two bondsmen capture Melanthius.

Pictured here is the recto of folio 82, with lines 238–271 of Book 22. With Melanthius captured and trussed up, Odysseus (named in the top line) and his three companions stand with spears ready against the remaining suitors. Six of the boldest suitors now decide to throw their own spears, but all miss their mark. (Their names appear on the badly damaged fourth, fifth, and sixth lines of the page.)

The heroes, in turn, target four others. “Taking careful aim” (in the three six-beat lines of Mendelsohn’s translation) “Odysseus hit Demoptolemos” (Δημοπτόλεμον μὲν Ὀδυσσεύς), “Telemakhos hit Euryades, the swineherd hit Elatos” (Εὐρυάδην δ᾿ ἄρα Τηλέμαχος, Ἔλατον δὲ συβώτης), “and the man who had charge of the cattle was the one to slay Peisandros (Πείσανδρον δ᾿ ἄρ᾿ ἔπεφνε βοῶν ἐπιβουκόλος ἀνήρ).

“So all of them together bit the dust of the wide-spreading ground”. I feel that Mendelsohn’s translation, with its six-beat line, is so much more Homeric (as in so many other places) than Wilson’s weak version: “They all fell and bit the earth”. I can well imagine that, if he were to read Wilson, “Odysseus would cover his head and start lamenting once again” (Book 8, line 92). For we miss the great beauty of Mendelsohn’s Odyssey, which truly deserves to be “the authoritative translation for years to come”.

A voyage around the Odyssey

As we wait, breathless, for “Christopher Nolan’s Odyssey” to be released next month, thoughts turn once again to Homer’s original 24-book epic.

When I last wrote about The Odyssey over a year ago (here), I lamented the fact that Emily Wilson’s 2018 version had taken over as the translation of choice for the modern generation.

I much preferred Lattimore’s 1965 treatment, with its richly poetic vocabulary, though Fagles’ 1996 translation (in my opinion) gave it some stiff competition.

The Odyssey is above all a seafaring epic, and we get our first taste in Book Two, as Telemachos sets sail from Ithaca, searching for news of his lost father. These 400-odd lines of Greek are a microcosm of Homer’s storytelling language.

Homeric language

One of Homer’s best known epithets is surely “rosy-fingered Dawn” (ῥοδοδάκτυλος Ἠώς, rhododaktylos Eôs). Lattimore, true to form, gives us “Dawn with her rosy fingers”. Similarly, Fagles gives us “dawn with her rose-red fingers”. But Wilson merely tells us that “her fingers bloomed”. Ingenious, perhaps. But not what Homer wrote. So we miss one of his famous trademark phrases.

Then there is Odysseus’ guardian deity, the goddess Athena, who is always “grey-eyed” (γλαυκῶπις Ἀθήνη, glaukôpis Athênê), from glaûkos, meaning “bluish-grey” (though many scholars prefer glaukòs, which means “bright, shining”). Lattimore, again true to form, has “grey-eyed”. Fagles, though, describes the goddess “with flashing sea-gray eyes”, thus having his cake while eating it. But for Wilson, she is always “bright-eyed Athena”. (That’s too Watership Down for my liking.)

Telemachos winds up at Sparta, home of “golden-haired Menelaus” (ξανθὸς Μενέλαος, xanthos Menelaos), husband of the infamous Helen of Troy. Fagles prefers to translate the epithet as “the red-haired king Menelaus”, though the adjective elsewhere suggests a lighter colour, like a chestnut horse or a tawny lion. Curiously, Wilson again breaks ranks with “ruddy Menelaus”, which she alternates with “flushed” or “high-coloured Menelaus”. This simply won’t do.

Another classic is the “wine-dark sea” (οἶνοψ πόντος, oinops pontos), which Fagles duly gives us. Wilson, too. Amazingly. (Lattimore preferred “the wine-blue water”.) Diggle, in the excellent Cambridge Greek Lexicon (2021), glosses the term as “having the appearance of wine”. If there’s one word in Homer that cries out for explanation, it must be this one. Of course, the language is meant to resonate. But it is also meant to evoke a mental picture, like the rosy fingers of dawn across an early morning sky. It’s notable that Telemachos is setting sail at dusk, precisely when the sea will appear purplish and dark. Only a few lines later, Homer tells us that the “purple wave” (πορφύρεον, porphyreon) sang along the prow of the ship, ingeniously using a word that encapsulates both the colour of porphyry and the action of surging or heaving. So the “wine-dark sea” is specifically an evening phenomenon.

Finally, I couldn’t leave the subject without mentioning Homer’s epithet for Helen of Troy, which is always “white-armed” (λευκώλενος, leukôlenos). If the rumour of Nolan’s audacious casting choice for Helen turns out to be true, perhaps his Athena will have rosy red eyes and his Menelaus will be bald.

When is a siege not a siege?

Archaeology is full of mysteries. The best we can expect of scholars is that they set aside personal prejudice and weigh up the available evidence, presenting it in a way that permits logical analysis. But there is a major caveat in this last phrase, because it’s the contemporary logic that we must apply, not our preferred twenty-first-century logic. And this is where the sheep diverge from the goats.

Fifteen years ago, I devoted a whole series of articles in the magazine Ancient Warfare to this very phenomenon. The series was entitled “The Debate”, because each article presented the available evidence to illuminate a particular topic, and assessed the ongoing debate about how that evidence should be analyzed.

For “The Debate” in December 2011 (in Ancient Warfare vol. V, issue 6), I took as the theme the Roman military operations at Burnswark in south-west Scotland (known as Birrenswark in the older literature), where the upstanding remains of two camps are imprinted on the landscape.

The article (available here) was called “Hillfort under attack: Roman siege warfare or training exercises at Burnswark?”, because these were the two explanations that had been put forward in the ongoing debate. (I say “ongoing debate”, but it has always been a little one-sided, as proponents of the “training exercises” theory insist that all evidence favours their case.)

From antiquarians to archaeologists

Over a century ago, in 1914, Francis Haverfield, the pre-eminent Romano-British archaeologist of his day, had voiced the opinion that “the well-known and remarkable earthworks at Birrenswark have long been explained as Roman siege-works round a native hill-fort” (though he failed to name those authorities by whom it had “long been explained”).

A year earlier, the site had been visited by the German archaeologist Adolf Schulten, who — fresh from his eight seasons of excavations at the elaborate Roman Republican siegeworks of Numantia (see here) — immediately confirmed Burnswark as the site of a Roman siege. (It should be mentioned, however, that it was his belief in a circumvallation, disproved soon afterwards, that prompted his analogy with Caesar’s siege of Corfinium in 49BC, a comparison that is still occasionally mentioned to this day.)

Only a decade later, Haverfield’s pupil R.G. Collingwood, investigating the site in 1925, confirmed that “the siege theory gives the only possible explanation of the camps and their relation to the hill-fort”.

In the swinging sixties, anything goes

Then, at some point in the 1960s, the peculiar idea took root — popularized by Roy Davies in a 1972 paper entitled “The Romans at Burnswark” — that the only siege at Burnswark had actually been a practice siege!

When I was researching my PhD in 2001, I identified a lecture delivered in 1963 by Kenneth Steer (Secretary General of the Royal Commission on the Ancient and Historical Monuments of Scotland) as the first time this idea was mentioned. “It is at least arguable”, Steer pronounced, “that the camps were simply training-quarters for troops engaged in storming native fortifications”.

Indeed, Roy Davies’ PhD thesis (entitled Peace-time routine in the Roman army, Durham 1967) — for which he began research in 1963, the very year of Steer’s lecture — has a chapter on “Manoeuvres”, in which he admitted that “the theory that Burnswark may not in fact be the scene of a siege carried out in time of war was suggested to me by Mr G. Jobey”, an archaeology lecturer at Newcastle University who had been excavating there in 1965 and 1966.

For the next forty years, the “practice siege” theory held sway, despite the fact that — unlike other Roman army manoeuvres — there was not a single jot of evidence to support the existence of such a thing.

For example, Gordon Maxwell gave it prominence in his 1998 book A Gathering of Eagles, from which this illustration (right) is taken. Here we see the north rampart of the South Camp, where the author has depicted catapults mounted on mounds in the open area outside the camp.

Siege or no siege?

In 2003, I published a paper — “The Roman Siege of Burnswark” (available here) — in which I restated the case for a genuine episode of warfare. I rather foolishly imagined that this might have put the entire matter to bed. But I hadn’t reckoned with the argumentative energy of those who had set their minds and staked their reputations on the “practice siege” theory.

That is not to say that my research went unheeded. In 2009, for example, Nick Hodgson appealed to “the convincing demonstration by D. Campbell that there is no reason to believe that the siegeworks at Burnswark were not intended to function in earnest”.

But while the “training exercises” proponents insist on having the last word, I find their dogmatic belief in a “practice siege” curious. Even if we suppose for a moment that such a phenomenon ever existed, how would we tell it apart from a genuine siege?

A misconceived argument

Whenever anyone claims support for the “practice” theory, discussion always comes back to the “Three Brethren”. These are the earthworks positioned in front of the three north-facing gateways of the South Camp (depicted on the plan above).

In 1925, Collingwood pronounced them “emplacements for artillery bombarding the hill-fort”, which he (erroneously) called ballistaria. These were a particular hobby-horse of Collingwood’s, an obsession that he bequeathed to his student and collaborator, Ian Richmond. They were keen to find such “gun-platforms” in any camp or fort that they happened to be investigating. (I disproved five of them in a 1984 paper.)

The “Three Brethren” struck Collingwood as particularly suitable for this function, as his 20th-century logic told him that artillery should be thrust forward and elevated as far as possible. But to a Roman, neither of these characteristics appears to have been at all desirable.

Advancing artillery beyond the safety of the camp was foolish and reckless, while elevating it above ground-level was only necessary if its purpose was to shoot at the parapet of high fortifications. We can see this from the ancient technical manuals. Any self-respecting Roman engineer would have constructed a siege-tower for this purpose. But in the case of Burnswark, we may add the clinching fact that the “Three Brethren” are too small and precarious to be used as “gun-platforms”.

Maxwell’s illustrations (above and right) cleverly show a solid mound, carefully constructed from layers of turf, with a large consolidated surface area to support his catapult with a crew of four or five men. In reality, the mounds appear, on the contrary, to have consisted of piled up earth measuring some 10–12 feet (3.5m) in diameter at the top, which is a tight fit for a 2-pounder ballista, with no spare room for the crew to work effectively. (How did they even get the catapult up there to begin with?) In the case of this one-armed onager, it is unclear how the rear of the machine would be supported, far less operated.

Back in 2003, I gave a talk to the Trimontium Trust in Melrose on this subject, in which I plotted the trajectory of a missile from a 2-pounder Roman stone-throwing catapult located safely behind the rampart of the South Camp, where it ought to be. I demonstrated that the hillfort of Burnswark was well within the range of such machines, without having to position them precariously on mounds dangerously exposed outside the Roman defences.

The cross-section view of the terrain at Burnswark also demonstrates very clearly the danger posed to the Roman camp by an enemy located on the hilltop. The gateways were especially exposed, in this regard, and required robust protection. In fact, we know of a genuine Roman description of such robust defences being employed to deflect heavy and destructive objects rolled downhill against Roman fortifications. There is even an illustration of this on Trajan’s Column.

But the “training exercise” proponents object on principle to any contrary argument. David Breeze cannot disprove my interpretation of the “Three Brethren”, so he simply asserts that “this, of course, is based on the assumption that Burnswark was the location of a siege”. Thus, any suggestions may be swept aside on the grounds that Burnswark is a “practice siege” — a classic case of circular reasoning. Alan Wilkins likewise claims that “the design of the Three Brethren and their position immediately in front of the three gateways is the most important evidence that this is not a real siege”. For him, the catapult balls and copious quantities of sling bullets are “the types of missiles that might be expected at a training camp”. There is no replying to this level of debate.

How rumours start

In 2011, Rebecca Jones published a book entitled Roman Camps in Scotland. I recently acquired a copy (thank you, Ruth and Ruairi) and was surprised by her discussion of Burnswark (pages 154–156). It wasn’t her characterisation of the “practice siege” theory as the most likely interpretation, as I fully expected this. Rather, it was her allegation — in the context of the “Three Brethren” — that “he [meaning me] has accepted the possibility that they are ‘ballistaria’!

This is quite impossible, for two reasons. First, I have never wavered in my scepticism of Collingwood and Richmond’s “gun-platforms”. Over forty years ago, I demonstrated that the variety of archaeological features that they wished to identify as “artillery platforms” were no such thing. The “Three Brethren” fall into this same category.

And second, more importantly, in my 2003 book on Greek and Roman Artillery 399 BC–AD 363, I explained the meaning of the word ballistarium: “for a long time scholars interpreted this as an ‘artillery platform’, but it was probably a storehouse or workshop for catapults”. So, definitely not a large pile of earth defending a camp gateway!

Postscript: A siege fantasy

Today, I came across a still image from the “Burnswark VR Experience”, a cinematic experience hosted by Trimontium Museum in Melrose. Under the lowering skies of Roman Scotland, waves of legionaries ascend the hillfort of Burnswark under fire (literally) from their own heavy-calibre artillery.

The foreground is occupied by an enormous gun-platform, fenced in, massively fortified, and protected by a deep trench. On top, three legionaries can be seen operating an old-style pre-Antonine stone-throwing ballista, which — from its size — appears to be something like a 40-pounder, designed to project stones weighing around 13kg. Such stones are larger, far larger, than those discovered at Burnswark. Moreover, the missile just launched from this machine is somehow capable of leaving a smoking trail.

Unfortunately, viewers of this fanciful scene will take away quite the wrong idea about any military action that might have taken place on Burnswark two-thousand years ago.

A book by any other name

In 1994, I wrote a somewhat critical review of Yann Le Bohec’s L’armée romaine sous le haut-empire, which had recently appeared in an English-language version. (The review is buried in a long-forgotten edition of Liverpool Classical Monthly.)

The original book was published in 1989, and apparently appeared in a second edition in 1998, a third edition in 2002, and a fourth in 2018 (all unbeknown to me). These later versions don’t seem to have been reviewed anywhere.

Having already purchased the original (twice) — and having struggled to enjoy Le Bohec’s book about the third-century army —, I am understandably reluctant to buy the new one, which (in paperback) retails for around £32 (€40).

Imagine my surprise, then, on finding (in my favourite online lending library) a 2004 Spanish version, El ejército romano, with “revisión científica” by the historian (mostly prehistorian, it seems) Francisco Gracia Alonso of Barcelona University. Could this be Le Bohec’s third edition?

Textbook or manifesto?

In 1989, L’armée romaine sous le haut-empire wasn’t just a book about the Roman army. Rather, it was a book about Le Bohec’s vision of the Roman army. These turned out to be two different things.

Here, in this edition, Le Bohec still promises to follow “a principle that we shall call ‘totality’”. (Incidentally, I have translated all quotations from El ejército romano for convenience.)

In fact, I found, thirty-odd years ago, that this wasn’t the comprehensive handbook that Le Bohec’s globalité (‘totality’) suggested. The question is, has anything changed in this (let’s call it the third) edition?

(And if you protest that this is all very academic, as Le Bohec’s fourth edition might be a completely new book, please remember that there will surely be enough people still reading the third edition, Spanish or not.)

Archaeology, but not as we know it

One of my original criticisms was about the lack of archaeological discussion, or even an appreciation of the contribution of archaeology. Unfortunately, this shortcoming has not been addressed.

Actually, Le Bohec has a peculiar idea of archaeology, summed up in his opening comment, that “excavations provide us with more than just inscriptions”. But for him, inscriptions are definitely the highlight, along with monumental sculpture, which is also subsumed under “archaeology”. (Later in the book, he remarks that “archaeology has revealed some interesting reliefs”, in reference to Trajan’s Column!) Great importance is also accorded to aerial photography, chiefly by Père Poidebard (whose 150th birthday is coming up soon — couldn’t he think of someone more recent?).

However, admittedly, a new paragraph has been added, with the revelation that “Recent work has drawn attention to a previously overlooked reality: soldiers need not only weapons, but also a range of items that constitute military equipment”! This has, at least, prompted Le Bohec to excise his original reference to Paul Couissin’s vintage Les armes romaines of 1926 (though it still appears in his bibliography) and replace it with Michel Feugère’s Les armes des Romains, a book that (as I pointed out elsewhere) has its own problems.

Having introduced the subject of military equipment, Le Bohec has added a paragraph to the “Tactics” chapter, explaining that “legionaries and auxiliaries possessed equipment that was much more similar than previously thought, although it did exhibit great diversity”. So ... the same but different.

Otherwise, archaeology is only deployed, as it was in 1989, in a facile attempt to gauge the size of the Praetorian Guard by providing a “decisive argument” (p. 29 = 1989 p. 21): the praetorians cannot have been as numerous as a legion, because their fort in Rome is smaller than a legionary fortress!

And archaeology is briefly mentioned in the section on sieges. But it is symptomatic of this idiosyncratic book that the original perfectly-judged footnote referring to the work of Adolf Schulten on the siege-works at Masada, and the comments of archaeologist Sir Ian Richmond based on aerial photographs taken by the RAF, have been replaced by a reference to a 1995 book by a French historian: Mireille Hadas-Lebel’s Massada.

Changes, but not for the better

Le Bohec displays a pride in French scholarship. We can see that in his preference for French publications over more pertinent non-French publications. That’s perhaps understandable. But references like “Carcopino, Mél. P Thomas, 1930” are not very helpful for Spanish (or, for that matter, any non-French) readers.

But this example serves a dual purpose.

It is one of the many niggling inaccuracies that permeate the book. The reference accompanies the statement that “Auxiliaries served twenty-five years in the time of Augustus, twenty-six from the middle of the 1st century onwards, and twenty-eight from the time of Caracalla”. Any moderately well-informed editor would immediately have realized that this is a description (admittedly, somewhat garbled) of the terms of service not of auxiliaries, but of fleet personnel.

(The 1930 Carcopino article, incidentally, is his publication of the military diploma of 30 April AD 166 that became CIL XVI, 122.)

Another peculiarity is the tendency to present insufficiently explored ideas as facts. For example, in one of the few newly inserted paragraphs, four sentences are added (on p. 17) on the subject of military diplomas, essentially repeating what already exists on p. 14 (= 1989 p. 10, where diplomas were first mentioned, with a blanket reference to L’Année épigraphique, as if each volume of that journal were packed with diplomas), although the citation of “M. Roxan, Roman Military Diplomas” has been updated to reflect the availability of “4 vols., 1978–2003”. (Six have now been published.)

However, Le Bohec takes this opportunity to repeat his peculiar belief that discharge from service entailed the issuing of an epistula missoria (no evidence is provided) before the lex data (the edict preserved in the military diploma) was enacted. (He has perhaps muddled up the handful of ‘discharge certificates’ that Nesselhauf gathered in 1936 in the Appendix to CIL XVI, which are quite obviously distinct from diplomas and nothing to do with them. But nor are they epistulae. The mystery remains unsolved.)

Addenda?

There has certainly been an attempt to mention more recent publications, even if their content is not reflected in the text. New footnotes appear, for example, in the Introduction, flagging up Le Bohec’s 1998 article about Onasander (but not C.J. Smith’s paper from the same year), C.R. Whittaker’s Les frontières de l’empire romain (but not the English translation), S. Perea Yébenes’ Collegia militaria (sensible for Spanish readers), M.A. Speidel’s Die römischen Schreibtafeln von Vindonissa (which are not discussed anywhere in the book), and the Mons Claudianus ostraca graeca et latina volumes I–II (likewise; incidentally, a third volume appeared in 2000).

But the main text seems rarely to have been altered. The paragraph that previously discussed the peregrini as spies and assassins who occupied the castra peregrina in Rome has been redrafted and now denies the existence of these men, but includes instead a class of “displaced soldiers” who “do not have Roman citizenship and correspond to the Latin term peregrini”, which all seems rather odd.

All of this gives the impression of a book that has been tinkered with, rather than revised.

No corrigenda?

Near the beginning, Le Bohec’s complaint that “researchers occasionally rely upon erroneous information” seems deeply ironic, given that the same mistakes that I noted thirty years ago (and there were many) are still there and the whole book now feels curiously outdated. To name a few examples, Le Bohec’s auxiliaries still wear leather tunics “sometimes covered with metal plates”, his trierarchi are still “undoubtedly” the same as centurions of the fleet, we are still told that the equites singulares “constituted a reserve and a training school for non-commissioned officers”, and the emperor Antoninus Pius still enjoys “a great reputation as a general”. None of these errors has been corrected.

Likewise, although he continues to scold those who “enjoy using Latin terms for any reason — and, naturally, most of the time for no reason at all — the precise meaning of which they do not know”, his own usage (which, as I charitably observed in my 1994 review, “occasionally lapses”) remains uncorrected.

The discussion of the three-element Roman name still includes the praenomen “Caius”, although (as I explained here) no such name ever existed. Readers are still (mis)informed that the trecenarius took his name from having commanded the 300 Praetorian speculatores. And the word vallum (“rampart”) is still translated as “palisade of timber”. More examples could be given.

But it is tedious to go through all of this again, and a little depressing.

To do list

Whatever is in the new edition of the book, I truly hope that any outdated bibliography has been excised. When the subject is soldiers’ epitaphs, we really shouldn’t be sent in search of Jean-Jacques Hatt’s 1951 book on La tombe Gallo-romaine or Jean-Marie Lassère’s mammoth 1973 paper in Antiquités africaines. There simply must be something more appropriate by now.

Le Bohec’s temptation to cite his own (rather specialized) 600-page tome on La Troisième Légion Auguste on the subject of dating inscriptions may be understandable. But perhaps the likes of Lawrence Keppie’s 1991 book Understanding Roman Inscriptions might better fit the bill?

We can only hope that the fourth edition will bring “a readable, informative and well-balanced work that will be of interest to scholars as well as general readers” (as one reviewer of the first edition rashly claimed).