¹ All references to Arrian in this paper are to the Anabasis. The best modern work on this battle is by Tarn, W. W., Cambridge Ancient History VI 379 ff.Google Scholar, and 595 (Bibliography), and now in Alexander the Great, Part I Narrative, and Part II Sources and Studies (Cambridge, 1947)Google Scholar references to which will be shown by the abbreviations Tarn AGN and Tarn AGSS. Indispensable also are Kromayer, J., Antike Schlachtfelder IV 377 ff.Google Scholar; Judeich, W. in Schlachtenatlas zur antiken Kriegsgeschichte, griechische Abteilung, Blatt 7 (ed. Kromayer, J. and Veith, G., Leipzig 1922)Google Scholar; Beloch, K. J., Griechische Geschichte III2 1. 642 ff., IV² 2. 290 ff.Google Scholar These versions contain some important differences of opinion.

I am greatly indebted to Dr. Tarn for allowing me to see his most recent work (AGN and AGSS) while it was still in proof, and for reading this paper in MS.: his detailed comments and criticism led me to make a number of additions, omissions and alterations. My thanks are also due to Professor F. E. Adcock, who read my MS. and helped me greatly by his suggestions, particularly in the interpretation of several difficult passages of Arrian: and to Professor D. S. Robertson for kindly giving me his opinion on a point of Greek translation.

² Cf. Judeich loc. cit.

³ There can be no serious doubt that the words, in this military connexion denote a bend or angle in a line of battle, whether a bend forward in the case of the outflankers, or a bend backward in the case of the outflanked (as here): cf. Arrian I 1 9. 2 and 11. 1 (Issus); Xen., Hell. IV 2. 20Google Scholar (Corinth); Cyr. VII 1. 6 (Croesus and Cyrus); Anab. I 8. 23 (Cunaxa); Diodorus XVII 57. 5 (Gaugamela); Polybius V 82. 9 (Raphia); VI 31. 2 (lay-out of a Roman camp). The defensive was called by the later tacticians a definition of which seems to illustrate Alexander's dispositions here; Aelian, , Tactica XXXI 4Google Scholar— The meaning of ὑπό is ‘behind’: for τρίπυλον (a three-piece gate), see Liddell and Scott,⁹s.v.

⁴ Tarn, , AGSS p. 184Google Scholar.

⁵ Arrian III 12. 4.

^{page 79 note 1} A small but awkward difficulty is a doubt which must present itself as to the meaning to be attached to the prepositions πρό and ἐπί when they are prefixed to τάσσω. έπιτάσσω clearly can mean either ‘post behind’ or ‘post next to’ (cf. Liddell and Scott⁹s.v. for the two groups of examples), and only the context can indicate which meaning is to be preferred in each case. But I am inclined to think that we must look for a similar ambiguity when we meet προτάσσω too, though I do not know that anyone previously has felt this doubt (Liddell and Scott⁹s.v. gives only the obvious meaning). The fact is that whereas (for example) the Persian scythe-chariots must certainly have been ‘posted in front’ (προετετάχατο, Arrian III 11. 7), the word is used also of certain cavalry units of Alexander, and seems to me to give better sense if translated ‘posted on the extreme flank’ (Arrian id. 11. 8, 12. 3 and 5): this is particularly so of its application to Alexander's in relation to the remaining mass of the ‘Companion’ cavalry at a moment when they are quite clearly in line, not in column (11. 8–9). The plan shows in which cases I have thought that πρό = ‘on the flank’ gives the better sense, and a careful reading of the cited passages in Arrian will (I hope) indicate why.

Although it is not easy in individual cases to interpret this ambiguity (if I am right in thinking it exists), it is also not difficult to see its origin, if we remember that a line-of-battle is really only a column-of-route in which each man has obeyed an order ‘Right (or left) turn’. This must, in fact, have been the commonest (because the simplest) way of getting into line-of-battle, by moving in column on to the required position, halting, right (or left) turning, and then making any necessary adjustments (such as ‘dressing’ spacing, and deepening the phalanx if desired). In a force moving in column, are the people in front; but when the column has halted, turned, and become a line-of-battle, have become the people on one wing. As for the mere paradox of πρό meaning ‘on the flank’, it can easily be paralleled, or even surpassed, in contemporary military language. What recruit has not heard the command ‘Company will advance–about turn’? It sounds like nonsense, but it makes good sense when you know what has gone before; and this is true also of my proposed paradoxical meaning for πρό.

It will be objected that here at Gaugamela Alexander had units προτεταΥμένοι on both wings. But the objection disappears when it is realised that in the preceding paragraph it is only for the sake of simplicity that I have spoken of one column as forming the line-of-battle. Anyone familiar with the movements of troops will realise that an army of even 20,000 men in one column would take hours to get anywhere or form anything. In practice an army deploying for battle would often deploy in a number of columns, to each of which individually my remarks above will apply. At Issus, Alexander brought his whole army into the plain of Issus in one column, but only because the narrow pass through which he had to march gave him no choice: as soon as the plain broadened out enough to give him room, the several contingents ‘peeled off’ from the single column under his direction and went the shortest way to their battle stations (Arrian II 8. 1 ff.). Nevertheless a clear reminiscence of occasions when the whole phalanx moved in one column can be seen in the later practice of calling the right wing the ‘head’ and the left wing the ‘tail’ of the phalanx (Arrian Tactica VII. 2), where ‘head’ and ‘tail’ becoming ‘right’ and ‘left’ correspond exactly to these alternative meanings of πρό and ἐπί.

^{page 79 note 2} For the probable strengths of these units of the right flank-guard, see p. 83, n. 20 below.

⁶ On the site of the battle, see now SirStein, Aurel, Geographical Journal 1942, p. 155CrossRefGoogle Scholar. The battlefield had been in part artificially levelled (Arrian III 8. 7). For the numbers of the Persian cavalry, see Tarn, W. W., Hellenistic Military and Naval Developments, pp. 153 ff.Google Scholar, where a maximum figure for the Empire of 50,000 is proposed: this is a paper strength, and not the number for this or any other particular battle, and Dr. Tarn points out that it may be an over-estimate. For his most recent views on the numbers of the cavalry at Gaugamela, see now AGSS p. 188, where he gives reasons for believing that the Persian cavalry units here were comparatively small, This is evidently true of certain picked units, e.g., the picked Bactrians and the Royal Guard (each 1000), and probably also the Indians who operated with the Royal Guard, the Sacae who operated with the picked Bactrians, and the Armenians and Cappadocians who did a similar job on the other flank. This does not account, however, for by any means all the cavalry, as a glance at the list will show: and Arrian's implication (III 13. 1) that Darius' line of battle was about twice as long as Alexander's does not suggest small numbers in general. Since Darius had no means now of getting good infantry, his best policy would seem to have been to make himself as strong as possible in cavalry, which he could get. For these reasons I would suggest that an estimate of 25,000 cavalry with Darius here may not be too high. Beloch's estimate (op. cit. III², 643, note 1) was 12–15,000. Curtius (IV 12. 13) gives 45,000 (with 200,000 infantry); Arrian (III 8. 6) gives 40,000 (with 1,000,000 infantry). Dr. Tarn is clearly right in his view that no Greek writer's total for a Persian army has any chance of being right, since the official version is just as liable to exaggeration (for obvious reasons) as the vulgate.

⁷ The hills at Issus had provided this insurance, though they did not relieve Alexander altogether of the necessity of guarding his right flank, since they were already occupied by a Persian force (Arrian II 8. 11; 9. 2–3).

⁸ Arrian III 12. 1.

⁹ Id. 13. 1, with 11. 5. Diod., XVII 58. 1, says that in the order of battle Darius was opposite to Alexander. He later (59. 2) says that Darius commanded the Persian left, which must be a mistake in view of the Persian tradition of the King in the centre. For a suggested explanation of the mistake, see note 16 below.

¹⁰ Arrian III 13. 2–4.

¹¹ Tarn, , AGSS pp. 185 f.Google Scholar

¹² Arrian III 14. 3. ….

The Granicus passage is Arrian I 15.4— here the word used for the defensive ‘pushing’, the repelling of an enemy out of one's own line. See Thuc. V 72. 3 for a confirmatory use of ἐξέωσαν for the ‘pushing’ of an attacking force.

¹³ III 14. 4. = ‘the lef wing of Alexander's army’.

III 15. 5. = ‘the baggage of Alexander's army’.

of III 14. 5 sub finem refers actually to Thracian unit (cf. III 12. 5). I am indebted to Dr. Tarn himself for pointing out these instances to me in a letter: I had not noticed them previously.

¹⁴ Arrian III 13. 5–6.

¹⁵ Id. 4.

¹⁶ It is possible perhaps that a hint of this development is to be found in the mistake of Diodorus (59. 2—see note 9 above) in making Darius command the Persian left. The mistake occurs just when Diodorus begins his description of a cavalry battle on the Macedonian right in which the Persians have the best of it, and in which the personal troops of Darius are engaged. If this is really his version of the beginning of the outflanking attack of which I have just quoted Arrian's version, one can see how Diodorus (or his source) could have got the idea of Darius' being in command of the Persian left.

¹⁷ Arrian ibid., 13. 4 ff.

¹⁸ Id., 14. 5 and 15. 1 ff.

¹⁹ Id., 15. 1 ff.

²⁰ See Tarn, , AGSS pp. 157 ff.Google Scholar, for an analysis of the Balkan (non-Macedonian) cavalry contingents at Gauga-mela: there were four (the other three being the Paeonians, and the Thracians and Odrysians of the left wing), and together they cannot have amounted to more than 1300–1400 horse, but this contingent of Aretes may well have been by far the largest of the four.

²¹ These figures cannot be regarded as certain, but it seems probable that the mercenaries of Oleander here (—Arrian III 12. 2) are the contingent which crossed into Asia with Alexander originally, 5000 strong (Diod. XVII 17. 3). The archers and Agrianians together numbered 1000 (Diod. ibid.).

²² While I hesitate to place reliance on the account of Curtius, so full of confusions (see note 26 below), I would point out that one long passage (IV 15. 18–23), when allowance is made for all its inaccuracies and misplacements, does give the impression of a hard battle on the right flank, not only before but also after Alexander himself (and of course the ‘Companions’) were already committed: it may be true, as Curtius says (ibid. 21 ff.), that the Agrianians here served him well.

²³ The only other body of troops near enough to support it was the ‘second line’ of infantry (Group D); but they were soon fully occupied in facing about and driving enemy cavalry out of the Camp (Arrian III 14. 6).

²⁴ Arrian III 16. 1. Tarn, AGSS p. 187Google Scholar. The figures of Curtius (IV 12. 6 f. and V 8. 4) for the Bactrians in the battle and later in Media are of little value in this connexion, since there may have been many desertions after the defeat. I n any case, they cannot be allowed to stand in the way of our accepting this statement of Arrian.

²⁵ The phrase here used does not necessarily mean that they did break through the second line of infantry in addition to riding through the gap in the first line.

²⁶ Arrian III 14. 4 ff. This is the view of all the modern writers cited above. This Indian and Persian cavalry, however, is in itself something of a problem, because it is not easy to see how it found itself anywhere near the gap when it appeared. The enemy opposite to the fifth taxis of the phalanx were the troops of the Persian right-centre (the ‘inside’ units of Mazaeus). When Arrian describes the return of this cavalry force a little later (15. 1), he calls them ‘the Parthians and some of the Indians and the most and best of the Persians’. Now the Parthians were in the right place, originally, to take advantage of this gap: it is the Indians and Persians who were not (see Plan I and Key). The Indian cavalry were in the true centre, with Darius, and ‘the most and best of the Persians’ must surely be the King's Guard, as Dr. Tarn says in his account of the battle (AGSS p. 187). But the Indian cavalry and the Royal Guard must (one would think) have taken part in the great outflanking movement described on pp. 81 f., and this movement took them into battle somewhat to the left (Persian left) of their original position in the centre, whereas the gap in the Macedonian phalanx occurred to the (Persian) right of their original position. It does not make sense. The solution I propose (though with the utmost diffidence) is as follows: When the gap in the phalanx appeared, the cavalry which used it was the Parthian, which was in a good position to do so. At the time when that happened, the Indians and the Royal Guard were already involved in the flank battle; but at some moment they, some or all of them, broke through the flank-guard (and with the ‘Companions’ now engaged there was little to stop them), and rode to the Camp, When the time came for looting to cease in the Camp, all the enemy cavalry returned together, and met Alexander and the ‘Companions’, now disengaged.

If this explanation is right, Arrian has omitted one of the two break-throughs, and has transferred the agents of this one (the Indians and the Royal Guard) to the other one, which really belongs to the Parthians. The account of the battle by Curtius is such a nightmare of confusion that I hesitate to use it either to confirm or to stultify any explanation of any particular incident. But allowing for the fact that sometimes (but not always—that would be too simple) he says right wing when he means left and vice versa, I think it cannot be denied that Curtius believed (if he ever thought about it at all) that there were two break-throughs by enemy cavalry, one by cavalry of the wing commanded by Mazaeus (he calls them Cadusians and Scythians, IV 15. 5, 9 ff., 12 ff. and 18 f.), the other by cavalry of the wing commanded by Bessus (these he calls Bactrians, IV 15. 20 and 22). The fact that all these bodies of cavalry (Cadusians, Scythians, and Bactrians) in reality came under the command of Bessus (see Plan I and Key) is not perhaps an insuperable obstacle, if we can bring ourselves to use Curtius at all for incidents in the battle, to our using him here in support of the view which I have just suggested, that there may have been really two break-throughs by enemy cavalry to the Camp.

The interesting thing about all this is the question of generalship which it raises: this time, of Persian general-ship. It has long been recognised that it was a piece of good luck for Alexander that a break-through by enemy cavalry should have wasted itself on the Camp instead of winning the battle for Darius or at least trying to win it; and the motive for the mistake was no doubt the political motive of trying to recover the King's family from its captivity (Diod. XVII 59. 7; Curt. IV 14. 22). But what if there were two break-throughs? That would mean, in a sense, two pieces of good luck, in the sense that a second body of cavalry should have followed the political rather than the military aim. It would not be two mistakes by the Persian command, but the same mistake committed twice, by two of its executives. If the thing really did happen twice over, it reveals either a pitiful incompetence in the Persian command (in this case obviously Darius himself), or else a failure by the command to make its wishes clear to its fighting leaders, and of the two alternatives the second is perhaps the more likely. It is incredible that Darius should have said ‘Rescue the Royal Family even if it means losing the battle’. But it is comparatively easy to believe that he said to Bessus and Mazaeus, ‘It is vital to rescue the Royal Family’, perhaps naming a prize for the man or men who should do it. In this case, the two generals would no doubt hand on the message to their unit commanders and they to their units: and the handing-on process, particularly when it is handing-down, is a peculiarly vulnerable one. It could well have ended in an intense rivalry among the cavalry units (Persians, Indians, Parthians, Bactrians, Cadusians and the rest), all determined that it should be they who won this prize. But this is conjecture, based on nothing more solid than a reading of human nature and an experience of its occasional impermeability to all except the most exact of ‘briefing’.

²⁷ Arrian III 15. 1 f.

²⁸ The second line of infantry later faced about to prevent it, and then drove them out of the camp (Arrian ibid. 14. 6).

²⁹ So (e.g.), Judeich loc. cit.

³⁰ It will be realised that this little plan is highly schematic, and that the reality must have been much more untidy: my A, B, and C are not intended to represent three still intact battle lines, but to cover the three main groups of the army, which was by this time much broken up by the manoeuvres of the several units in attack, pursuit or (in the case of A) in self-defence.

³¹ It may be said that the gap created by the wheeling (on my interpretation) of the ‘Companions’ would still exist if, instead of wheeling, they had charged straight forward ‘in pursuit’ (presumably of the fleeing Darius): in this case the enemy cavalry returning could still use this gap, and still meet Alexander as he returned ‘from pursuit’ to help Parmenion (which is certainly what Arrian says he did—loc. cit.). The objection to this, and it is a strong one, is that although in these conditions the two forces could have met if both had wanted to meet, it is certain that the enemy cavalry wanted anything rather than this. They were not now trying to do damage, they were trying to escape; and this meeting, as Arrian says (III 15. 2), meant that they had to fight for their lives instead of merely riding for them. My interpretation, however (see Plan), allows of the probability of their being surprised by the meeting, since Alexander could appear rather suddenly out of the dust created by the other formations on the extreme right of the Macedonian army, cf. Diod. XVII 60. 4 and 6 for the dust of this battle.

³² Arrian ibid., 15. 2.

³³ Id., 14. 4.

³⁴ See H. Berve, Das Alexanderreich auf prosopographischer Grundlage, No. 508.

³⁵ Arrian ibid., 12.3. Curtius (IV 5. 12) mentions this Menidas as moving with his mercenary cavalry to the rear at an early stage in the battle, to drive enemy cavalry out of the Camp. It seems impossible that this can be right, unless Arrian's account of the battle is to be jettisoned entirely in favour of the confused story of Curtius. In any case, Curtius makes Menidas return to Alexander almost immediately.

³⁶ As for how a mistake about ‘the pursuit’ can have arisen, it seems very unlikely that Ptolemy did not know the truth about Alexander's movements and their motive; but I suggest that a description by Ptolemy of a wheeling movement by Alexander to the right when Darius fled, if it were not phrased carefully and explicitly, could perhaps have been misinterpreted by Arrian as the beginning of a pursuit of Darius.

³⁷ I am assuming of course, what all our sources indicate clearly, that Alexander was still unaware of the state of the battle on his left flank: the position there was serious, but he did not know this until he got Parmenion's message.

³⁸ Arrian ibid., 15. 1: Curt. IV 16. 1 ff: Diod. XVII 60. 7: Plut., Alex. 33Google Scholar.

³⁹ Diod. id., 4 and 7.

⁴⁰ Id., 7 … ….

⁴¹ See A.J.Ph. 98 (1937) 109 f.Google Scholar, where A. J. Robinson deduces a legal justification from Curtius VI 11. 20 supported by Arrian III 27. 1 ff.

⁴² See in general Beloch Gr. G. IV² 2. 294 ff.

⁴³ Plut., Alex. 33Google Scholar = Callisthenes frag. 37 (Jacoby).

⁴⁴ Arrian IV 8. 6.

⁴⁵ Curtius and Plutarch, locc. citt.

⁴⁶ So Kornemann, E., Die Alexandergeschichte des Königs Ptolemaeus I von Aegypten, pp. 56 ff., and p. 130Google Scholar.

⁴⁷ Tarn points out most acutely (AGSS p. 177, n. 1) how Ptolemy may have damaged the reputation of his (later) great enemy Antigonus by merely omitting all references to his great work as satrap of Phrygia while Alexander was in the Far East. But he had no motive for depreciating Parmenion.

⁴⁸ Loc. cit., pp. 300 f.

⁴⁹ Notice particularly his preoccupation at Issus, too, with the danger of being outflanked on his right (Arrian 118. 7; 9. 3 f.).