In the early autumn of 359, the Persian King Shapur II took by siege the Roman city of Amida in the province of Mesopotamia (modern Diyarbakır, Turkey). By far our most detailed source is the soldier-historian Ammianus Marcellinus, who was trapped in the city and escaped during its sack. An incident in his flight presents seemingly undetected problems in interpretation and text. Ammianus and his two anonymous companions have run ten miles from the city under the cover of night. Ammianus is exhausted—as a gentleman he is unused to travelling on foot (ut insuetus ingenuus, 19.8.6)—when he beholds a horrific sight which nevertheless brings him timely relief. Clark’s text and punctuation are as follows and other editions differ only in minutiae (19.8.7):

J.C. Rolfe’s Loeb translation gives the conventional interpretation of what this means and all other translations, into whichever language, have very much the same sense, although (as we shall see) two of them improve on one crucial element:

Ammianus’ ghoulishness has been noted by readers, as has the entitlement of social privilege by which he, the gentleman officer, gets to ride.Footnote ¹ What has not been observed is that the implied actions of the dead groom make no sense and that there is something wrong with the conventional understanding of the sentence. A moment’s consideration reminds us that a rider controls and directs a horse with a bit to which the reins are attached. Pulling on the reins transfers pressure from the bit to the mouth, tongue and gums, which tells the horse which way to go.Footnote ² But here Ammianus is explicit about the lack of a bit (infreni), which means that any ‘reins’ to which the hapless groom had lashed himself would have been entirely useless.

One must ask what was happening, which leads one to the word habenam—if it produces nonsense when translated as ‘reins’, it makes sense when translated as ‘halter’: this is the advance made by Seyfarth (Halfterriemen, that is, halter straps) and Hamilton (halter), perhaps influenced by the TLL entry by Bulhart.Footnote ³ Roman grooms were not medieval squires riding behind their knights—they walked, along with the support staff, and their primary experience of horses was from the ground. Though it is no use for directing a horse from its back, a halter rope with correctly placed knots is the commonest and easiest way to direct a horse when on foot, as it pulls forward on the neck and throatlatch (and sometimes nose) and shows the horse the way. A groom would have been familiar with the correct use of a halter. But something is still wrong: the words (ne labi possit) ex more habenam qua ductabatur … inligauit clearly imply that the groom has tied the halter by which the horse is guided to his hand in the usual manner, lest he fall off. One may do various things to prevent oneself falling from a horse, but tying reins to one’s wrist is not one of them, because if one does fall, it can lead directly to the fate that befalls the groom here. One might indulge in special pleading and say that in the heat of the moment he lashed the halter to his wrist and mounted the horse, forgetting what might go wrong. But that still will not do, because nothing in the language suggests he made a mistake. Whatever the groom did with the halter ex more was perfectly normal, and what he did was not what one normally does to stop oneself falling off a horse. So the problem lies elsewhere.

Fundamental in establishing the text of Ammianus is his prose rhythm.Footnote ⁴ Ammianus’ accentual prose rhythm expresses itself in two or four unaccented syllables between the last two accented syllables of each clause: it is astonishingly regular—well over 98 per cent of clauses in the unemended text—and the absence of clausulation is as much a guide to modern editors of where corruption may lie as it was a guide for ancient readers as to where the sense-breaks were. The prose rhythm in this passage does indeed imply a problem. In the first sentence (fugaci equo nudo et infreni calonum quídam sédens (ne lábi póssit) ex more habenam quá ductabátur, sinistra manu ártius inligáuit) one finds regular clausulae after ductabatur and inligauit, but not at the clauses with the verbs sedens or possit, for which one would also expect them. The first of the two is solved by an easy emendation: Ammianus never uses the word sedeo for sitting on a horse, for which he uses insido.Footnote ⁵ We should read quídam insídens and mark a clausula, in having been lost after the final m of quidam.Footnote ⁶

That leaves the problem of possit, which can be solved by reading on until one reaches a clausula: ne labi póssit ex móre. On the pattern of previous translations, that would mean ‘in the usual manner to stop him [the groom] falling off’—but we have seen that that is not usual at all. Ammianus’ usage of the verb labor is the key. While the fundamental sense of the verb is something like ‘slide’ or ‘slip’, its two main senses in Ammianus are ‘to fall’ (OLD s.v. lābor ¹, 5) and, almost as frequently, ‘to escape’ (L&S s.v. lābor, I.B.2).Footnote ⁷ Here, it actually means ‘in the usual manner to stop it [the horse] escaping’: a groom would customarily tie a halter round his wrist in order to stop the horse escaping if it shied or tried to bolt and the groom’s grip on the halter failed.

So finally we have sense: a groom has grabbed hold of the halter of a horse not yet tacked up and tied the halter to his wrist as he normally would in the course of his duties. Then, in the desperate hurry of a rout, he had tried to ride it using the halter as one would normally use reins, with the predictably deadly result. The one remaining mystery is how Ammianus was able to jerry-rig a bridle and reins to continue his journey on horseback. Perhaps with the immediate urgency gone, and the horse exhausted, it was more amenable to carrying a rider bareback with improvised reins or with a makeshift hackamore fashioned from cloth or leather. Or perhaps, one of Ammianus’ less socially elevated companions played the part of a groom and led the horse by what remained of the halter.

The following is the emended and repunctuated text as it will appear in Kelly’s forthcoming Oxford Classical Text:Footnote ⁸

We translate as follows in our forthcoming Landmark Ammianus Marcellinus: