An Update on Javanese Crossbows

A while ago I posted something on the blog about crossbows in the Indo-Malaysian archipelago. I looked primarily at the textual evidence and came to the following conclusions:

  1. No unambiguous word for ‘crossbow’ existed in the region’s languages in the Middle Ages.
  2. Words that might have referred to crossbows, particularly the Old Javanese word gaṇḍi and its Malay relative gandi, probably referred to other weapons (slings, etc.) most of the time.
  3. To my knowledge, no pre-sixteenth-century reliefs or other images of crossbows could be found in the archipelago.

Well, predictably, it turns out that my knowledge was not complete. I was alerted to the existence of a Javanese relief apparently depicting a crossbow by the prehistorian Sofwan Noerwidi (@sofwannoerwidi) on Twitter. The relief in question, an andesite one at Candi Mleri near Blitar in East Java, is discussed at length in an article by Willem Frederik Stutterheim (1935), and two of the photographs he used have been digitised and hosted online by the Leiden Digital Library. We’ll have a look at those in a moment.

Before we do that, though, we need to have some background on caṇḍi Mleri itself (Figure 1).

Fig. 1 — A map showing the location of caṇḍi Mleri in East Java, Indonesia.


Mleri’s Old Javanese name seems to have been Waleri. It appears as such in the kakawin Deśawarṇana, composed by Mpu Prapañca in 1365 CE, which says (41:4 — adapted from Pigeaud 1960:31):

śakābda kha-nawa-wāni-kṣiti Bhaṭāra Wiṣṇu mulih iṅ śūrālaya pjah,
dhinarmma ta sire Waleri Śiwawimbha len Sugatawimbha muṅgwiṅ Jajaghu

‘In the Śaka year “air-nine-earth-earth” (1190, AD 1268) Lord Wiṣṇu returned to heaven, having died;
He was enshrined at Waleri as a Śaiwa image, and as a Buddhist one at
Jajaghu(In Stuart Robson’s translation [1995:54])

(I’ve put the Old Javanese vocabulary for these lines in an appendix to this post — see below.)

The Lord Wiṣṇu here is Wiṣṇuwardhana (IAST: Viṣṇuvardhana), also known as Ranggawuni, the third king of Siṅhasari, who, Prapañca tells us, died in 1268 CE. (The Pararaton gives a slightly different date.) The combination of Buddhist and Hindu worship implied in the last line is quite typical of élite Javanese religion at this time, and East Javanese-era kings were often considered to be embodiments of the Buddha and various Hindu deities simultaneously.

Most of the reliefs at Mleri date to the late twelfth century CE — a number of inscriptions at the site date it to around 1180 — and so a good deal earlier than the enshrining there of Wiṣṇuwardhana in 1268. Still, this reference tells us that the temple was being used and updated into the late thirteenth century and was still revered in the mid/late fourteenth.

A Scene from the Kakawin Arjunawiwāha

As might be expected from Prapañca’s description, the reliefs at Waleri/Mleri are ‘Hindu’ rather than ‘Buddhist’, with most depicting stories from the eleventh-century Old Javanese kakawin Arjunawiwāha (IAST: Arjunavivāha), the oldest surviving kakawin from the East Javanese period. Stories from the work, which is based on the Mahābhārata, were often depicted in Javanese temple reliefs, including the relief we’re concerned with here, which shows the aftermath of an encounter between Arjuna and the god Śiwa (IAST: Śiva).

In this episode, Arjuna, the hero of the Arjunawiwāha, is charged by a wild boar while in seclusion in the forest. He shoots it dead with his bow. Suddenly, a hunter appears, claiming that he had shot the boar instead (Figure 2). An argument ensues, which turns into a brawl, during which Arjuna comes to realise that the hunter he’s fighting is none other than Śiwa himself. Impressed by his devotion, Śiwa grants Arjuna pāśupata (or pāśupatāstra), a superlatively powerful weapon usually interpreted as a kind of magical bow.

Fig. 2 — A relief depicting the argument between Arjuna (L) and Śiva (R) at caṇḍi Kedaton, Probolinggo, East Java (site dated 1292 Śaka/1370 CE). Note that Śiva and Arjuna are both clearly carrying recurved composite bows. Leiden, UBL, OD-3402.

The relief from Mleri that appears to show a crossbow is now in the Museum Majapahit at Trowulan, and it depicts Arjuna receiving pāśupata from Nārada, messenger of the gods, at the end of this encounter. In the photographs of the relief in Figures 3 and 4 below, Śiwa appears on the left and Arjuna on the right. Nārada is in the middle, facing Arjuna and holding a strange weapon that has been interpreted as a crossbow. This is rather surprising: although fundamentally a magical weapon directed by the mind, pāśupata is usually depicted as a bow and not a crossbow.

Fig. 3- A relief depicting a scene from the Arjunawiwāha at caṇḍi Mleri. Leiden, UBL, P-045022.
Fig. 4 — A detail from the Arjunawiwāha relief at Mleri in Figure 2. Leiden, UBL, P-045024.

W. F. Stutterheim’s Interpretation

As I said above, this relief is discussed in Stutterheim (1935), an article I don’t have access to in paper form (the library here isn’t working because: coronavirus) but which can be accessed in slightly janky HTML fashion here. My comments here are based on this online version of the article, which may or may not accurately reflect the original.

Anyway, Stutterheim credits Josef Knebel with identifying the weapon in the relief as a crossbow. Stutterheim himself had some misgivings about this identification, asking,

…waarom heeft de Paśupata den vorm van een kruisboog, indien wij tenminste Knebel moeten volgen?

(‘…why does pāśupata have the form of a crossbow, if we must follow Knebel at least?’)

He comments that he had never seen a crossbow before in a Javanese relief; that the arms of the ‘bow’ are extremely short while the stock, if that’s what it is, is excessively long; that the string is connected to somewhere near the middle of the limbs rather than the tips; and that the weapon as depicted would be difficult to wield and shoot as a crossbow. The weapon appears to have a defined central ridge — more like a sword than a crossbow — and it even seems to have a handle shaped somewhat like a sword hilt at the bottom (1935:134).

Stutterheim also says, however, that crossbows are commonly found in reliefs from mainland Southeast Asia, which is certainly true, and that the fact that he knew of no other crossbows from Javanese reliefs ‘of course says little’ (dat zegt natuurlijk weinig). He even floats the possibility that both a sword and a (really tiny) bow are being depicted here, which would be peculiar but would fit with the more standard iconography of pāśupata. The central ridge could equally well be the underside of a crossbow — but even then the shape would be quite unlike any crossbow known from anywhere else in the world.

This is certainly not an open-and-shut case, anyway, and Stutterheim advises us not to jump to conclusions.

I’m not entirely convinced that the weapon in the Arjunawiwāha relief at Candi Mleri is a crossbow. The interesting thing from a philological perspective, though, is that we are confronted with an image of an unknown weapon with a known name. It suggests to me that there is a very real possibility that crossbows were known and used in medieval Java, but that they were known by the names of magical superweapons and ambiguous ‘projectile weapon’ category names (like gaṇḍi, a word we looked at in some detail in the last post on crossbows), and so are effectively invisible in the textual record.

A. J. West — Leiden, May 2020

Appendix: Old Javanese Vocabulary

Here I’ve listed all the words found in the extract from the Deśawarṇana (41:4) above, just in case you’re interested in learning some Old Javanese. Notice that many of the words here are originally Sanskrit (Skt). The interpretations are mostly taken from the online version of Zoetmulder’s Old Javanese-English Dictionary (1982) and the numbers in brackets refer to their locations in that dictionary. You can find Willem van der Molen’s 2015 Old Javanese textbook here if you’d like to learn more of the language.

abda (2:1 — Skt) year.

awani (166:8 — Skt) the earth.

bhaṭāra (224:8 — Skt bhaṭṭāra) God, Lord, deity; the Lord.

dhinarma (367:11.8 — from Skt dharma) to be enshrined.

kha — Robson translates this as ‘air’, although it seems to me (disclaimer: I’m very much not an expert) that it might be better translated as ‘cavity’ or ‘hollow’ (as in Monier-Williams 1899:334 sub ख). Both would be appropriate ways of writing ‘zero’ in candra sangkala format (see van der Meij 2017:443–446).

kṣiti (905:16 — Skt) the earth, ground, land.

len (1021:2) other, different, otherwise, and also.

mulih (2111:5.6) to reach (go to) one’s destination, go (come) home, go (come) back.

muṅgu (2125:1.1 — muṅgwiṅ = muṅgu + iṅ ‘in, at’) to be positioned, be situated, dwell.

nawa (1178:14 — Skt) nine.

pĕjah (pjah — 1335:3) death, dead.

śakābda (1603:2 = Skt śakakāla) the Śaka calendar, Śaka year.

sira (1785:2 — sire = sira + i ‘in, at’) 3.SG pronoun (‘he, she, it’).

sugata (1833:6 — Skt) the Buddha.

surālaya (1863:3 — Skt) abode of the gods, heaven.

ta — (1892:2) separating particle.

wimba (2281:10 — Skt bimba, wimba) disk, esp. of sun or moon, orb; image, reflected form, picture, portrait, statue. (I discussed this word in an earlier post on some of the inscriptions from Kawali, West Java.)


Meij, Dick van der. 2017. Indonesian manuscripts from the islands of Java, Bali, Madura, and Lombok. Leiden: Brill.

Monier-Williams, Monier. 1899. A Sanskrit-English dictionary. Oxford: Oxford University Press.

Pigeaud, Theodore. 1960. Java in the 14th century. Volume I. The Hague: Martinus Nijhoff.

Robson, Stuart. 1995. Desawarnana. Leiden: Brill.

Stutterheim, Willem Frederik. 1935. Enkele interessante reliefs van Oost-Java. Djåwå. 15:130–144.

Zoetmulder, P. J. 1982. Old Javanese-English dictionary. The Hague: Martinus Nijhoff.

Posting about ancient and medieval Indonesia, up to ~1500 CE. Mainly into 14th & 15th century stuff, but earlier is fine too.