THIS IS AN EXTRACT FROM CHAPTER 9 OF PROFESSOR PHILIP
SABIN’S BOOK LOST BATTLES: RECONSTRUCTING THE GREAT
CLASHES OF THE ANCIENT WORLD (HAMBLEDON CONTINUUM,
2007). FULL COPYRIGHT RESTRICTIONS APPLY, AND IT MAY NOT BE
POSTED ELSEWHERE OR OTHERWISE DISTRIBUTED IN ANY FASHION
WITHOUT EXPRESS WRITTEN PERMISSION.
HYDASPES, 326 BC
The River Hydaspes was much wider than the Granicus or the Pinarus, so when he
encountered Porus’ Indian army on the other side, Alexander had to undertake a
surprise dawn crossing 150 stadia (27 km) upstream.1 He then smashed a force of
cavalry and chariots led by Porus’ son, and marched back down the river to meet
Porus himself. As I said in the Introduction, this engagement is plagued by
ambiguous and contradictory sources and scholarly disagreements, and is an
archetypal ‘lost battle’. Arrian (Anab. V.15-18) is again our principal source, with
some 1,500 words on the battle itself, based mainly on Ptolemy and Aristobulus.
Diodorus (XVII.87-9), Plutarch (Alex. 60), Curtius (VIII.14) and Polyaenus (IV.3.22)
have shorter and more anecdotal accounts which nevertheless contribute something,
often by challenging the details in Arrian.
There is the usual scholarly controversy over exactly where on the modern Jhelum the
engagement took place, with the shifting course of the channels posing familiar
complications.2 Fortunately, this is immaterial for our purposes, since there are no
indications that terrain (even the riverbank) impinged on the actual battle. A more
important issue concerns the weather conditions.
The sources describe a
thunderstorm during the night, and Arrian (Anab. V.15) tells how the soft mud made
the chariots under Porus’ son useless. Curtius (VIII.14.4-8 & 19) confirms this story,
and also claims that the slippery ground hindered the Indian archers during the main
battle. However, he runs the two engagements into one, when in fact a number of
hours must have passed before the main battle, considering how far both armies had
to march in order to meet. Arrian (Anab. V.15) explicitly says that Porus drew up on
sandy ground to avoid the deep mud present elsewhere, and the subsequent cavalry
(and perhaps even chariot) manoeuvres all the way from one flank to another do not
chime at all with the bogging down described in the earlier encounter. The scenario
below hence begins in overcast conditions, with no restrictions operating at first, but
with a continuing risk of further wet weather which will asymmetrically handicap
Porus’ less well commanded force of chariots and archers.3
The size of both armies is a matter of considerable dispute, complicating attempts to
use either one as a reliable yardstick against which to measure the other. Arrian
(Anab. V.14 & 18) says that Alexander crossed the river with 5,000 cavalry and 6,000
infantry, and he repeats the figure of 6,000 foot when discussing casualties after the
battle. However, he claims (Anab. V.12) that Alexander’s infantry force before the
crossing consisted of the Hypaspists, archers, Agrianians and two phalanx taxeis,
which should in theory amount to considerably more troops. Some writers think there
really were just 6,000 foot - Lane Fox suggests that only the Hypaspists and light
1
Arrian (Anab. V.11).
Bosworth (1995) pp.265-9 and Devine (1987) p.96.
3
See Head (1982) p.136 and Bosworth (1995) pp.291 & 293.
2
infantry made the crossing, while Bosworth and Daniel argue that the various
contingents must have been significantly under strength, perhaps because of the
limited capacity of the available boats.4 Other scholars challenge the figure of 6,000,
claiming that it referred only to the heavy infantry or that some of the other troops
which Arrian (Anab. V.12) says were left at intervals along the bank managed to cross
and join up before the main battle. Hammond, Green and Ashley estimate that there
were 9-10,000 infantry in the engagement, while Devine and Hamilton suggest 15,000
and Montagu goes as high as 23,000.5
Arrian’s figure of 5,000 cavalry is less controversial, though there is still debate over
contingent sizes. Estimates of the number of Companions range from 1,500 to 2,700,
with the rest being horse from the eastern satrapies including 1,000 Dahae mounted
archers.6 Our model allows users to experiment with all these possible variations, but
the scenario below takes an intermediate approach, assuming there were 6,000 heavy
and 2,000 light infantry, plus 2,000 Companions and 3,000 Asiatic light horse. The
Hypaspists were surely by now armed with the sarissae which Curtius (VIII.14.15-16)
and Diodorus (XVII.88) describe being put to good use against the elephant line.7
One could argue that more than just the Hypaspists, Companions and Agrianians
should by now be classed as veterans, or conversely that some of the Asiatic horse
should count as levies, but the scenario instead uses an average rating throughout.
With Coenus as an average leader to reflect his prominent supporting role, the army’s
fighting value at a multiple of 3 hence comes to 85.
This gives us at least some kind of yardstick against which to judge the varying
numbers for the Indian forces. Arrian (Anab. V.15) says Porus fielded 30,000
infantry, 4,000 cavalry, 300 chariots and 200 elephants, while Diodorus (XVII.87)
gives him 50,000 foot, 3,000 horse, over 1,000 chariots and 130 pachyderms. Curtius
(VIII.13.6) agrees with Arrian on infantry and cavalry numbers but says there were
only 85 elephants, and Plutarch (Alex. 62) gives the lower totals of 20,000 foot and
2,000 horse. Modern scholars disagree just as much, with Hammond and Lane Fox
endorsing Arrian’s claims while Devine and Green go with Plutarch’s much lower
numbers. As regards elephants, Hammond and Hamilton think there were 200, Green
opts for 130, Devine and Ashley choose 85, and Bosworth suggests less than 50.8
At a multiple of 3, each elephant unit in our model represents 15 beasts with light
infantry support or 30 without. The only way 200 pachyderms would be practical
would be if we discount the infantry support which these beasts seem to have had. At
the other extreme, giving Porus less than 50 beasts would mean a fighting value of
only 12 for this key component of Indian strength. Six units representing 85
accompanied elephants seems a much more appropriate total. The Indian cavalry
fought bravely and for a long time despite being outnumbered, and must surely be
classed as average, but the foot archers were low status troops who should clearly
4
Lane Fox (2004) p.355, Bosworth (1988) pp.127-8 and (1995) pp.277-9, and Daniel (1992) p.52.
Hammond (1989a) pp.210-13, Green (1991) p.393, Ashley (1998) p.318, Devine (1987) pp.98-100,
Hamilton (1956) p.26 and Montagu (2006) p.151.
6
Hammond (1989a) p.213, Devine (1987) pp.98-9, Daniel (1992) p.51, Bosworth (1995) pp.293-5 and
Aperghis (1997) pp.143-4.
7
See Bosworth (1995) pp.59-61 on the debate over the weapon used to kill Cleitus 2 years earlier.
8
Hammond (1989a) pp.211-13, Lane Fox (2004) pp.357-8, Hamilton (1956) p.27, Devine (1987)
pp.101-2, Ashley (1998) p.318, Green (1991) p.396 and Bosworth (1995) p.292.
5
count as levies like their Persian counterparts, especially since the best 4,000 or so
infantry are assumed to be in front with the pachyderms.9 Even if we have 30,000
foot as well as 3,000 horse and up to 300 of the large chariots, then with Porus as an
average commander, the army’s fighting value comes to 66, which still gives
Alexander a clear advantage in what was by all accounts a hard fought battle.10
Devine’s smaller Indian army would have a value of only around 54, which as usual
for him seems over-cautious given that the value of his inflated Macedonian force
would exceed 100.
This brings us to the vexed issue of army frontages. As I mentioned in the
Introduction, Arrian’s figure of 100 feet between elephants (Anab. V.15) would give
his infantry line alone a length of over 6 km.11 Hammond reduces the intervals to the
50 feet claimed by Polyaenus (IV.3.22), while Ashley cuts his elephant numbers to 85
and retains Arrian’s spacing, but both these writers still suggest an Indian infantry line
around 3.5 km long, dwarfing their 1.0-1.2 km line of Macedonian foot, and with the
unengaged majority of the Indians conveniently remaining far more inert than
Artaxerxes’ right did at Cunaxa.12 Devine’s 85 more closely packed elephants would
cover only 1,300 metres, short enough to be matched by his force of nearly 15,000
Macedonian infantry, but Bosworth thinks there were only 6,000 such foot, with a
frontage of just 700 metres, and he makes the Indian line equally narrow because he
believes that Alexander trapped Porus’ entire deeply arrayed army in a complete
double envelopment which (if true) would outshine Hannibal’s achievement at
Cannae.13 Reputable scholarly estimates of the length of the Indian infantry line thus
differ by a factor of 5 or more!
Our model strongly suggests that both Hammond and Bosworth are wrong, and that
the reality lies somewhere in between these two radically different extremes. With a
frontage of 500 metres per zone, an Indian deployment of 85 elephants at 50 foot
intervals, backed by infantry in around 20 ranks, and with cavalry and chariots split
evenly between the wings, makes perfect sense. It is also entirely plausible that
Alexander would counter this with his usual oblique attack on his right with the
Dahae, Companions, Agrianians and Hypaspists, while holding his 8 rank phalanx
and flanking light infantry back in the centre to guard the Hypaspists’ left, and while
detaching Coenus to tie down the Indian right until the main attack could take effect.14
Arrian (Anab. V.16) makes clear that the majority of the cavalry were with Alexander
on the right, but putting all the Asiatic horse there seems rather like overkill given the
limited frontage available, and it makes more sense for at least some of them to
accompany Coenus, as Bosworth suggests.15 Our model also shows how the other
9
Head (1982) pp.135-9.
Polyaenus (IV.3.22) describes Porus as riding the leftmost elephant of the line, but elephants and
levies cannot be rallied by leaders in our model, and Arrian (Anab. V.18) makes clear that Porus
remained on the field as long as a single unit fought on, so it is more appropriate to treat him as a
commander.
11
Hamilton (1956) p.27.
12
Hammond (1989a) pp.212-4 and Ashley (1998) pp.322-6.
13
Devine (1987) pp.101 & 113 and Bosworth (1995) pp.292-3 & 296.
14
Hammond (1996) pp.27-31 brings some interesting insights from Polyaenus (IV.3.22) to bear on
this.
15
Bosworth (1995) pp.294-5.
10
Asiatic light horse have the flexibility to reinforce the central phalanx if its flank is
threatened by the Indian right.16
The biggest controversy about the tactical course of the battle concerns Arrian’s
description (Anab. V.16-17) of the Indian right wing cavalry being shifted across to
reinforce the left wing horse, and being pursued and attacked from behind by Coenus’
men, who then joined up with Alexander’s cavalry as the Indian horse withdrew to the
shelter of the elephants. Some, like Hammond, think that at least one of these
movements took place between the two opposing infantry lines, but others such as
Hamilton believe that both forces rode round the back of the Indian line.17 In our
model, the former possibility is best simulated as part of initial deployment, with both
sides’ cavalry already massed on one flank by the time the main lines get so close
together. Given the limited frontage on the right, it would actually make more sense
for Coenus to be placed on the other wing regardless, so the scenario below is based
on the alternative possibility of movements behind the Indian line. If the Indians get
sufficient commands on turn 2, they are indeed able to advance their whole main line
and then make a free about turn with some or all of their right wing horse and move
them all the way to their left rear or left wing using double moves, relying on the
elephants to protect the right against Coenus. He in turn can exploit Alexander’s
capacity for turn reversal to curve round all the way into the Indian rear, thereby
catching the Indian horse between two fires as happened historically. This places the
Macedonian cavalry in an ideal position to encircle the left and centre of the Indian
line. Meanwhile, the Macedonian pikemen and light infantry will engage the
elephants in the bitter frontal struggle which the sources describe, until the crowded
mass of infantry, cavalry and pachyderms finally gives way under attack from all
sides.18 Indian casualties were enormous, but the sources put Macedonian losses at up
to 280 cavalry and 700 infantry, and the bravery of the wounded Porus in defeat
stands in stark contrast to the alleged timidity of Darius.19
This battle illustrates perfectly how our model can contribute to a vexed scholarly
debate which has produced radically different reconstructions using more traditional
approaches to the obscure and often contradictory evidence. Despite the complexity
of the engagement, which involved units of every single type and class, we can
compare proposed armies according to their overall fighting values rather than just
their numbers and presumed frontage, and we can study suggested deployments and
manoeuvres dynamically and as the product of rational command choices rather than
just through blocks and arrows on a map. The results are not conclusive, and there is
plenty of scope to experiment with different hypotheses, but our model casts severe
doubt on some of the starkly contrasting ideas put forward by writers like Hammond,
Bosworth and Devine, however erudite and impressive their scholarship.
16
Arrian (Anab. V.18) claims that 200 of his 230 cavalry losses fell among horsemen other than the
Companions and Dahae.
17
Hammond (1989a) pp.211-14, Fuller (1960) pp.192-7, Judeich in Kromayer & Veith (1922-9)
Griech. Abt. Blatt 7.7a & b, Warry (1991) pp.76-9, Ashley (1998) pp. 323-5, Hamilton (1956),
Scullard (1974) pp.69-70, Green (1991) pp.397-9, Devine (1987) pp.102-7 and Bosworth (1995)
pp.299-300.
18
Arrian (Anab. V.17-18).
19
Arrian (Anab. V.18), Diodorus (XVII.89), Devine (1987) pp.108-9 and Bosworth (1995) pp.304-10.