WATER FROM THE ANCESTORS: IRRIGATION IN EARLY JAVA

AND BALIl

Jail Wissemall Christie

Introduction
Any discussion of water management in the early states of South East Asia must,
given the history of European thought on the subject of Asian states, also address
the larger issue of the nature of those states and the explanations which have been
advanced for their origins. Two broad schools of thought have, over the past few
decades, dominated the study of early South East Asian states, Java and Bali
included. One tends to treat inigation as the key to centralized state power, and to
view rice-growing states as 'hydraulic societies' which shared certain characteristics
of royal claim to ultimate ownership of the land arising from the ruler's functional
role as builder and maintainer of the water system. This view was first advanced in
connection with Java by Raines (1817) and still lies at the heart of Marxian
assessments of early states on the island (Tichelman 1980). The mainstream
writings of such professional South East Asianists as Coedes (1968), Heine-
Geldern (1943) , Benda (1962) and their associates have tended to reflect similar
views in rather vaguer terms. South East Asianists made uncomfortable by this
school of thought have more recently been offered the alternative model of Geertz's
negara (Geertz 1980), generated from ethnographic studies of modern Bali and
projected backwards into the history of the region. This school focuses attention
upon the Balinese subak, or irrigation society, which manages water distribution at
a local level, excluding direct royal interference and thus undermining any
functionalist royal claims to ultimate ownership of the land.
The only work of recent years to deal extensively with rice cultivation and
irrigation in early Java - that by van Setten van der Meer - has drawn upon the
arguments of both schools, proposing a model which combines the presence of
subak societies with the provision of royal construction of and control over larger
inigation projects (Meer 1979:42 and 97). This mixed model appears to be based
upon two assumptions made by the author: first, that the Balinese subak society can
be viewed as a survival of a more widely-spread and ancient Indonesian institution,
and second, that irrigation provides the explanation for the development of supra-
village political power which, in tum, led to the formation of the first proto-states
on the island of Java. For this latter idea she is indebted to van Naerssen (1977),
whose explanation for the creation of the first Javanese states runs as follows:

The detennimmt factor in Cenu·al Java, East Java and Bali for the formation of
Hindu-Javanese kratons, was the relationship between the population living in
agrarian communities (e.g. Wi/nUll) and the person(s) who had the disposal of pmt
of their labour and produce ... This authority ... emerged from the previous
equalitarian society - long before the intrusion of Hinduism - when the sawah
with its severe water control system was introduced ... The introduction of sawah,

1 I wish to thank the British Academy for the funding that made possible some of the research
upon which this paper is based. I also wish to express my gratitude to the members of the Pusat
Penelitian Arkeologi Nasional in Jakarta and Yogyakarta, and to those of the Suaka Peninggal<m
Sejarah d~m Purbakala in Yogyakarta ~Uld Cenu·al Java for tile aill ,md facilities provided. My great
gratitnde goes to Drs. Boechm·i and Drs. Mund;u~jit() for sharing their time anu iueas witi1l11e.

7
The Gift of Water

i.e. the cultivation of rice on irrigated fields , reljuired a complicated system of

inigation and is therefore based on the co-operation of several cOllununities or
wanu(/.\·, all depending on the water of !lIe same river or its tributaries. It goes
without saying that such an ilTigation system needed a head whose authori ty
reached beyond tha t of a single wanu(/. Such a head could have extended his
power over a federation of wanuas ,md be considered a., the rakll or 'older brntiler'
of tile ramas of tile wanua. It was essential that tile raka had the right of disposal
of the produce and tile labour of his subjects to fulfil his fun ction. Hence he
became !lIe ruler of that federation (Naerssen 1977:27 and 37-8).

This, succinctly put, is probably the most widely held view of how the first states
of Java and Bali were formed, and of the centrality of irrigation of a fairly
complicated nature to the agrarian econ omies of those and subsequent states.
Naerssen's model is both attractively simple and intemally logical. Meer's thesis,
which pulls in both directions at once, is rather more muddled, but it too
emphasizes the centrality of water control to political development. But do the data
actually support this view'! Was a 'severe water control system ' a necessary feature
of early states of Java and Bali'l Did early rulers playa direct role in building and
managing irrigation sys tem s" Or are .l" uba k societies as ancient and once
widespread as has been suggested? Archaeology, unfortunately, can provide very
little in the way of answers to these questions at present, since surviving irrigation
works are too difficult to date and no abandoned systems (such as constructions
which must lie under the layers of volcanic ejecta in southern Central Java) are
accessible. There does, however, exist a substantial body of epigraphic material
which provides a great deal of information concerning agliculture, taxation, and the
relations between the state and the village, covering a period fro m the 9th thr()ugh
the 15th centuries, and focused up on the political and economic heartlands of
Central and East Java and of Bali 2 . Given the nature of the bulk of the surviving
inscriptions, one would expect to find , at the very least, oblique references to water
management, if the activity did indeed provide a foundation for political control or a
justification for taxation. The attention given to the transfers of tax rights by most
inscriptions should also highlight such external obligations of the villages affected
as would be involved in the support of subak-type irrigation societies. The
discussion below will focus on the period of the 9th through 13th centuries , when
the great majOlity of the surviving inscliptions were wlitten.

The ecological context and agrarian responses

Most of the early states on the two islands for which epigraphic data exist were
centred in three areas: on the uplands skirting Mounts Merapi and Merbabu and the
Perahu Massif in Central Java, in the Brantas River valley and on the adjacent
Malang uplands of East Java , and on the southem slopes of the Balinese mountains
(figures 2 and 3). Although the climate grows progressively drier as one moves
east from Central Java to southem Bali, all three regions share a relatively long dry
season and depend upon the south-west monsoon for their heaviest period of rain.
In all three regions the mountains playa key role in the productivity of the fields
below them. They trap the rain clouds, increasing rainfall on their Hanks. Since
this appears to happen even out of the monsoon on a regular basis, this effect

2 For discussion of the Javanese ch~u·ters see BmTe tt Jones 1984, Christie 1<)82; for those of Bali
see Goris 1965, Sukm"to 1<)86.

8
 Water from the ancestors: inigation in early Java and Bali

'"~ Q;'"~ ~
Q; E 0
E
0
0
0 "'
~
0
N
;;;
., ~ 0
>
0
a."
E c "c
" --''" --''"
~

I nl

Figure 2. Historical sites and geography ()lJava

9
The Gift of Water

increases substantially the available surface water on the mountain slopes, much of
which feeds into numerous small rivers draining the high regions. Most of the
rivers flowing south from the Perahu and Merapi-Merbabu peaks in Central Java
and the central Balinese peaks remain small, draining separately into the sea or
converging near the coast. However, those Howing south from the Malang
Highlands are blocked by limestone ridges near the southem coast and are det1ected
into a giant spiral, circling the western fringes of the volcanic massif and collecting
uibutaries from other mountains to fonll the great River Brantas, which debouches
into the sea through a large delta just to the north of the Malang Highlands (figure
2). States centred in East Java used both the highland and the lowland zones, with
an apparent shift of population from the uplands to the delta and river valley
occuning after the beginning of the 10th century. These states were thus faced with
more varied, and less stable, water control regimes than were their Central
Javanese and Balinese neighbours.

Land over 200 metres

_ La nd over 1000 metres

20 km
~--~,

Figure 3. Historical sites and geography of Bali

The mountains near the hearts of the states in the three regions share a
further key characteristic: they are all volcanic, and some at least in each grouping
are still active. The structure of the cones produces a spling line girdling the peaks
at about the 400m level, which raises further the surface water supply below that
altitude, multiplying water resources, but confining them to an optimum belt.
Above 400m, the incline grows markedly steeper-and water resources fewer, while
below 100m separate small waterways give way to larger rivers with fewer small
tributaries. Distribution of known early historic remains in Central Java and
southern Bali seems to indicate that most early settlement was located within that
optimum altitudinal band (Mundarjito: personal communication), and pollen
samples from the site of Borobudur in Central Java indicate that by the 9th century

10
 Water from the ancestors: inigation in early Java and Bali

the region sUlTounding the temple was reasonably well settled, containing a mixture
of grain fields, house gardens, palm groves and forest similar to that of the modem
landscape, though perhaps in somewhat different proportions (Nossin and Voute
1986:858). A striking feature of the disttibution of early sites in Central Java and
southern Bali is the fact that all settlement so far mapped hugged rivers and springs
very closely indeed (Mundarjito: personal communication). There cannot have been
many rice fields of the period located at any distance from natural water supplies,
and it seems unlikely that any inigated lice culture along the banks of the numerous
small streams and rivers of the area would have called for elaborate artificial
constructions. If no more than one ilTigated crop was grown annually, there should
also have been little need for major collaborative efforts between villages.
The volcanic nature of the Javanese and Balinese mountains affects the
fertility of the surrounding uplands, and the lowlands through which their waters
drain. The volcanoes of these regions only infrequently explode violently. Most
normally produce intermittant nows of volcanic muds and sands. This ejecta,
usually described as lahar, is neutral-basic in composition, and it adds immensely to
the fertility of the soil it renews. The /ahar is canied beyond the immediate region
of the nows by the waterways draining the volcanic massifs, and several
centimetres of this silt may accumulate annually in sawah fields inigated by tapping
these streams and small livers. The regions under discussion now support what is
probably the greatest density of agricultural population in the world - in excess of
1,000 per sq km over large areas and more than 2,000 per sq km in some districts
(Hugo et al. 1987:56). Since the present population of the islands is probably a
good thirty times what it was in the later first and the beginning of the second
millennium, early farmers in the region were clearly not straining the carrying
capacity of the land, nor were they under pressure to intensify crop production.
Given the low population in the past, there is no reason why wet rice
regimes need have been used at all. Indeed, the thesis has been advanced by a
number of geographers that ilTigated rice as a major crop isrecenl in Indonesia, and
that it could not have been the chief staple of the early Javanese and Balinese states
(Men·ill 1954, Hill 1977). This assessment, however, appears to have been based
upon an erroneous identification by Steinmann (1934: 584) of the grain depicted in
the reliefs on the Central Javanese temple of Borobudur as millet (Setaria italica,
Beauv.). The suspicion that the grain in the Borobudur relief is actually a stylised
representation of rice is supported both by the archaeological and by the epigraphic
evidence. Bricks dating to the 8th and 9th century were tempered with rice chaff,
just as they are today. More importantly, many of the inscriptions of that period
and later make specific mention of sawah (h uma in Bali), that is, irrigated rice
fields. Sawah was the most important revenue-producing land throughout the
period under discussion, although about one-third of agricultural taxes appear to
have been collected on lice grown on dry fields (gaga in Java, parlak in Bali)3

3 See Naerssen 1'.)41:46, Christie 1')82: 132ff. Other types of productive land mentioned in
inscript.ions included tgal (dry field, probably farmed by swiddening), sukat (fallowed or brush
land), po mahan (houseland with garden), keblVan (orchard land pl<Ult.ed with fruit. t.rees or other
long-lived perennials), padang (pastureland), paserehan (betel gm·den), patalesan (tm·o gm·den),
pakapasan. (cotton field), Zillah asinllll or pi/lVuyahan (saltem), alas (forest), renek (mm-sh), rlVang
(mountain slope'!), and tpitpi (river bank).
Chau .Tu-kua, a Chinese trade official writing in A.D. 1225, described .Tava as a 'broad and
level country well suited to agriculture' (Hirth and Rockllill 1911:77), which produced rice, hemp,

11
The Gift of Water

Irrigation of some sort was thus clearly a matter of importance to early

rulers on the islands, since much of their tax revenue was dependent upon the
enhanced production that it apparently brought. Sawah also had the effect of
anchoring populations and increasing their visibility, and of making the size of the
crop relatively stable and easy to calculate. Rulers also recognised the importance
of irrigated rice in settling certain areas of the countryside, concentrating
populations along waterways and thus reducing danger to travellers and trade.
Many of the grants issued during the late 9th and early 10th century on the mountain
slopes to the north of Magelang in Central Java, and in the Brantas Delta of East
Java were made on the understanding that forest and swidden land be converted to
sawah 4 (figure 2). There see ms no doubt, then , that some form of water
management was practised in these areas by the 9th century at the very latest, and
that this was encouraged by the rulers of the states involved.
One difference between the landscapes of the early state centres of Java and
those of Bali is both marked and significant in relation to the manner in which
ilTigation has been handled. While the Javanese landscape in the upland centres of
population is relatively nat or undulating, and the gradient of the slope is generally
gentle, the Balinese landscape is much more dissected. A comhination of steeper
inclines and differences in the lahar now has produced a situation in which even the
smallest trickles of water cut deep, steep-sided ravines which not only impede east-
west communication, but also create technical di ffic ulties in connection with the use
of water for inigation. Each of the three regions under discussion, then - the gently
sloping region of numerous small waterways in Central Java, East Java with its
combination of undulating uplands and massive river delta, and the sharply
dissected slopes of southern Bali - had its own unique water management
requirements, and one might expect to find these differences ret1ected in the
epigraphic record.
Chinese reports of the early second millennium A.D. men tion rice as the
major crop of Java but do not mention double cropping until the 15th century (Mills

millet, beans, t,u'o, gourds, betel nut . sug;u- cane, coconuts, bananas and other fruit for local
consumption, and pepper for export.

4 Grants of this kind were made ];u'gely on tile frin ges of tile settled terrilnries of bolh CeIllral and
East Java, especially along major road and waleI' conneclions wilh the coast. In Central Java
between A.D. 872 and 882 a series of at leasl len sima graIlls was made to several temples in tbe
mountains to the north of Magel,mg, along tile main route 10 Ihe north coast. In many cases these
gnmts, made at tile behest of local rakryans, bul wilh royal sanction, in volved rgill (swidden land)
or sukat (brush land) which was to be converled to sllwall (Boechm'i 1()R(i:E5, E6, E7, E9 , EIO,
E15, E18; Machi Suhadi and Sukm'lo 19R6: 2.7.1; 2.73 ; 2.7.4; 2.7.6). A fe w yem's later similm
grants were made by local rakrVlII/ along the route between Ihe Malang uphmds and tile coast of
East Java, tile staled reason being Ihe lhmger prescllIed 10 travellers by Ule {gal fi elds along tile road
(Sarkm' 1971:LV I & LVll). At Ihe begi nning of tile 10th century, during a period of exp,msion 1Uld
consolidation, ,mo ther granl was made in connection wilh foresl and smvilh hmd on tile slopes of
mounts Sumbing and SUIllloro to the nortil of Magelang for Ihe slaled purpose of prolec ling the
high road (Sm'km' 1971:LXX). At Ihe s,une lime in easl Java a gumt was made conceming forest
hmd which W,L~ 10 be convened 10 .I'{/Ivah because Ihe forest presented a d,mger 10 Iraders 1md cmL~tal
people (Ban'ert Jones 19):\4: 1):\1). In none of these cases did the rulers involve themselves direclly
in the process of land clear,mce or water conlroL They merely provided the tax incenti ves for
others to do so. .

12
 Water from the ancestors: inigation in early Java and Bali

1970:91). None of the Javanese or Balinese inscliptions written during the peliod
under discussion make any reference to double cropping, a fact which must be
borne in mind when assessing the importance of dry-season irrigation. We also
have no idea whether inigated rice duting the period was broadcast or transplanted.
However, in either case, the yields from a single long-season crop could well have
been sufficient to cover subsistence and create a substantial surplus, given the lack
of pressure on land at the time, although increasing emphasis upon rice as an export
commodity during the 11th and 12th centuries in East Java) may have led to the
adoption of early ripening rice strains in the delta region. Rice which ripened in
under a hundred days was available in South East Asia by the early 11th century. It
was at this time that the Chinese first imported short-season Champa rice from the
coast of Vietnam (Ho 1956-7), and it may well have been known in Java before this
date. However, the yield of Champa rice is said to have been far lower than that of
the later-ripening strains, and there is no evidence that short-season rice was of
much interest to the Javanese or Balinese before the 19th century. We are left,
then, with the probability that double cropping of lice occurred only in limited areas
before the 19th century. One of those limited areas must, however, have been the
Brantas delta region of East Java where Ma Huan reported its employment in the
15th century, and where rice was grown as an export commodity by the nth
century.
For a number of the reasons mentioned above, the delta region of East Java
probably provided the greatest opportunity for royal involvement in water
management: it was an area in which pioneering occurred late, and with royal
encouragement; the rivers utilized are larger than those in the uplands; and there
was pressure put upon the areas just behind the ports to produce rice for the
purposes of export by the llth century. The two other areas also provided
opp011unities for direct royal involvement in ilTigation: the pressure from the courts
to settle the route from southern Central Java to the ports on the north coast by
expanding the area of sawah cultivation, and the technical difficulties and need for
inter-village cooperation in irrigation on Bali. The responses of the various courts
to these opportunities, as reflected in the epigraphic record, reveal a good deal about
the political organisation of these states.

IRRIGATION IN EARLY JAVA AND BALI

Water management in Java: the epigraphic record

InsCliptions provide both direct and indirect information concerning early water
management. Most of the Old Javanese and Old Balinese language inscriptions
from the two islands are tax charters which record in some detail the economic
underpinnings of the villages concerned and the payments nonnally due to the court
and its vmious officials. Inigated lice cultivation provided the largest single source
of tax income, and one would therefore expect a good deal of attention to have been
devoted to irrigation in the charters. This is, however, not the case. Very little
space in most charters is dedicated to matters associated with irrigation and very
few charters indeed mention specific water control constructions. That said, water
management is alluded to indirectly in many charters, and in three different ways: in
references to officials ·connected with water, to technicians specializing in water

5 See the Mammjung inscription (Christie l'.182:504ff, Stutterheim 1928).

The Gift of Water

control, and to taxes in some way associated with irrigated agriculture (or more
specifically, to grants allowing villagers to extend their irrigated fields without
incurring additional taxes).
By far the most common reference to water control on both islands is found
in lists of officials. In inscriptions of the 9th and early 10th century, from both
Central and East Java, mention is made of several officials who appear to have had
some connection with water: the hulu airier (head of water), air warangan (water
'maniage'? official), I1wtmnwak (person in charge of dykes or dam s) , air haji (royal
water official) and jukut air (,greens and water'? official). Of these officials the
first three are found in lists of village officers. While the air warangan and
matamwak appear to occur only once each in the surviving corpus, the hulu air
appears regularly in lists of village officers found in charters issued during the 9th
and early 10th century. After this time, a reorganisation of the contents of the
charters tended to exclude these lists, but there is no reason to believe that the hulu
air or an official very similar did not continue to fulfil this type of village role
throughout the peliod under discussion. The hul/.( air appears to have been chosen
from amongst the village rama, that is, the core, land-owning heads of household
who comprised the village council (karaman). The office was one of moderate
importance, in the same class as such offices as the hulu wras (head of rice), hulu
alas (head of the forest), hul/.( bunt (head of hunting) and the hulu pkan (head of the
market). The number of hulu air within a single community appears to have
depended upon the size of its fanning population and the spatial distribution of its
fields in relation to different water sources. Up to eight hulu air have been recorded
for a single communityO, all apparently of equal status. The activities of these
officials seem to have taken place at a very local level, their authOlity being limited
to the community of which they were members, as was that of the far less
frequently mentioned air warangan and matamwak 7. All three offices apparently
dealt with the practicalities of water control and distlibution.
The air haji and jukut air were officials connected with the state
administration - both at a low level, judging by the epi thet~ attached to the names of
the office-holders. The jukut air is mentioned only once, early in the 10th centuryg,
but the air hqji, as a member of the class of l1utngilala cliwY(f haji tax collectors, is
mentioned regularly in charters from the 9th through the 13th century in Java. Both
the air haji and thejukut air appear, however, to have been religious officials rather
than members of the lay administration. The precise nature of their connection with
water is unclear.
A small number of charters from East Java, dating to the nth century and
later, mention an official called the panghulu bmw (head of water) in connection
with levies charged for use of water. Those paying the charge were religious
establishments; the collectors of the levies appear to have been members of the
village either adjacent to the religious premises or within the boundaties of which

6 See Machi Suhadi ;md Soekarto 1986:72ff.

7 airwarangan: Zoetmulder 1982:467, Ilwtlll/LI vak : Br;mdes 19l3:VL23.

8 jukut air orjukut er: Cohen Stumt 1875:1.2.4, Boechm-i 1986:E13.

14
 Water fro m the ancestors: inigation in early Java and Bali

the religious establishment was l o cated ~ . The single exception to this rule appears
to relate to waterworks in the immediate sUlToundings of the capital of Majapahit,
mentioned in the late 15th century inscription of Trailokyapuri (Brandes 1913:
XCIV and XCV), where the upkeep of the water system (largely for domestic water
supply?) may have been under more direct court controL Other mentions of water
levies paid by religious establishments to host farming communities occur in
different fOlms. The holy place of Sarwwadharmma in the Kediri area is stated in a
charter of A.D. 1269 (Brandes 1913 :LXXIX) to have 'purchased' (tuku) water
from a village, fo r instance. In areas not immediately adjacent to the capital's
metropolitan distlict, authOlity over water appears to have continued to rest with the
falming communities. The panghulu banu or hulu banu of the 13th century seems
to have held an office similar to that filled by the village-based and village-bounded
hulu air of earlier centUlies.

Water management in Bali: the epigraphic record

On the island of Bali, during this time, arrangements for regulating water seem to
have been rather different Here too the air haji appears regularly in the charters,
but only those issued in Old Javanese language after 993 A.D. As in Java, this
figure appears to have been a religious official operating within the state
administration, and here the post was normaJly held by a Saivite priest Another
representative of the court, the nayaka air (water chief), again a religious official, is
mentioned at least once on the Balinese chaners lO At the court level, then, there
may ha ve been some sim il arities in approach to water on the two islands.
However, below this level the differences are marked. No eq uivalent of the
Javanese village office of the hutu air appears in Balinese charters wlitten either in
Old Balinese or Old Javanese. Instead, from at least the 11th century onwards,
mention is made of an official called I1wkas er (Modern Balinese paka.l'eh), in
connection with an organisation called the kasuwakcm, apparently an early fonn of
the present subak (irrigation) society, which drew its members from all farmers
sharing a single water source for irrigation, and which crossed village boundaries
(S ukartoI986). The use of the term l17aka se r (chief, head) for the official in
position to allocate water, as in the Baturan inscription of 1022 A.D . (Go ris
1954:352.5b5), rather than the term klihan may indicate that the holder of the office
had somewhat more power in the past, but otherwise it appears, from the
fragmentary evidence available, that the organisation of this early .I'uwak group
(kasuwakan) diffe red little from the more recent .I'ubak society. The charter of
Udanapatya refers to 19 kasuwakan authorities to which farmers were attached
through an area stretching from Lake Batur in the north to Gianyar in the south
(Sukarto 1986:41) (figure 3). The kasuwakan names are different from the names
of the villages referred to in the same charter, so it does appear that their areas of
jurisdiction cross-cut those of the villages.

The technology and organisation of irrigation in Java and Bali

The contrast between Balinese and Javanese approaches to inigation is evident also
in connection with the techn ology associated with irrigation. No professional

9 See Bram]es ll}L~:LXXXIX, Pigeaud 1lJ60:III & 147.

10 nayaka air: Goris 1954:0J 356.l}.

15
The Gift of Water

irrigation technicians can be identified in the lengthy lists of village-based artisans

and builders mentioned in Javanese charters. In Balinese charters, however, the
undahagi pangarung (irrigation tunnel builder) appears in lists along with the
undahagi baul (stonemason), undahagi ka)'u (carpenter/builder) and undahagi
lancang (shipwright) ll. The technical difficulties of irrigation engineering on Bali
clearly called for greater professional expertise than did the construction of
irrigation works on Java, but there is no evidence that these professionals were
connected with the court.
The Balinese approach to taxation was somewhat different from that of
Javanese states, but here too there appears to have been no major irrigation tax
levied by the court. The single mention of a water tax called pa-er in a charter of the
11th century (Callenfels 1926:60) occurs in conjunction with a number of other
minor taxes levied upon such village enterprises as the snaring of animals. Other
charters grant tax relief upon the extension by a village of irrigation facilities in
order to increase the area of wet-rice cultivation, but in these cases the tax involved
was a tax on rice harvests rather than upon the water or the water-works.
The differences between the traditional approaches of the Javanese and
Balinese to water control recognised in the ethnographic literature have considerable
histOlical depth. The Balinese subak tradition is at least 900 years old. Java, on the
other hand, shows no sign of ever having had such an institution. The reason for
this divergence may be found, at least in part, in the different physical and political
landscapes of the two islands. On Java the territory held by a community was
relatively large, normally encompassing much of the water used by its members,
and the technical difficulties connected with the use of the water sources appear to
have been minor, at least in the upland regions. Javanese states were also rather
large, and few communities found thymselves on the borders with other states. On
Bali, however, water provided both barriers and technical problems. The ravines
cut by streams and rivers tended to become political boundaries, not only between
villages, but also between tht; numerous small states which at times co-existed on
the island. Thus, use of liver water for ilTigation could pose political problems. At
the same time, the depth of the water below the adjacent fields meant that elaborate
systems of tunnels and channels were often necessary to bring water (again across
political boundaries, in many cases) from points upstream. Under such
circumstances, the developm ent of such a non-political institution as the subak
society, which provided a framework for cross-border cooperation, comes as no
surprise. At the same time, the fact that no such institution appears in Javanese
records should also cause no surprise l2 . Irrigation societies occur elsewhere in
island South East Asia, as in the very steep landscape of the Bontok region of
northern Luzon, where heroic efforts are required to build and maintain the field
and irrigation systems (Brett 1985), but in most cases these associations operate
within the boundalies of a single village, as is and was the case on Java. There is
no reason to believe that the boundary-leaping irrigation society of Bali, with its

11 See for instance Goris 1')54:002 and 0.1 352.

12 Meer's (1')7'):42) assertioll that the vi llage n,une Subaki, found in a 15th century Javanese
inscriptioll, refers to a sltiJai.: is eJToneous. The n;une has a different derivation: slt-iJai.:i. No other
evidence for the ex istence of sulmi.:-type organisations in .lava appems in the epigraphic record, in
more recent histori es, or in ethnographic accounts.

16
 Water from the ancestors: inigation in early Java and Bali

high degree of disengagement even from local political authority, fonned a common
prehistoric heritage of Indonesian societies_ The tradition of handling inigation at
the village or sub-village level does, however, appear to be widespread in the
matitime region.

The role of the court

The suspicion that the Javanese and Balinese courts played no substantial role in the
practical organisation of irrigation is reinforced by examination of the very few
inscriptions which make direct references to specific structures for the control of
water. These few come from East Java and from Bali; none has so far been found
in Central Java. The East Javanese records fall into two groups: a collection of five
from the upland skirts of the mountains to the west of Malang, and a set of two
from the delta region. Of these, the upland charters span the longer period. They
comprise a series of three from Harinjing (Callenfels 1934) all dealing with a single
set of water works, dated 804, 921 and 927 A.D_ (Damais 1970:38); a charter from
Wulig (Bakalan) dated 935 A.D. (Brandes 1913:XLIV); and a much later charter
from Kusmala, dated 1350 A.D. (Callenfels 1919)_ The two charters from the delta
region were both issued during the reign of King Airlangga, that of Cane dating to
1021 A.D_ (Brandes 1913:LVIII) and that of Kamalagyan to 1037 A.D. (Brandes
1913:LXI).
The series of charters from HaIinjing concems a dam (mula-dawuhan) and a
channel (dharmma-kali) constructed at the behest of a religious figure, Bhagavanta
Dhali of Wulanggi, in or just before the year 804 A.D. The projects were carried
out privately , for the benefit of a religious establishment, and since they involved
the nooding of land belonging to others, land was given, as arranged, by the
Bhagavanta Dhati in compensation. The ruler was then approached in order to have
the taxes on the area transferred to the upkeep of the religious foundation and its
water works belonging to the family of Bhagavanta Dhati. The two later charters in
this collection confirm the benefits of this tax transfer upon the descendants of
Bhagavanta Dhari and list the specific provisions concerning the upkeep of the
Buddhist foundation and the associated water works_ The terms mula ('origin',
'ancestor' - a tenn applied to certain types of temple) which was linked with the
dam, and dharmma (religious foundation, pious act), connected with the channel,
indicate that the works were viewed as religious acts. None of the inscriptions
indicate the use to which the water was put, but it seems more than likely that they
were connected with a sacred bathing place (tirt/w) of the type which was common
in that region.
The edict of Wulig, dating to the mid 10th century, concerns the construction
of three dams at Kahulunan, Wuatan Wulas and Wuatan Tamya at the behest of
Queen Rakryan Binihaji, who was rakryoJ1 of MangibiL Although local villagers
were granted certain rights over the use of the water from the channels and were
given permission to fish in the waters, the purpose of the project was clearly to
benefit a religious foundation, for which a large tank (talaga) had been constructed.
The Saivite priest Punta Pakatappan was commanded to meditate upon the waters
controlled by the project, devoting his whole attention, day and night, to the Holy
Amblita (amrf({, 'the elixir of life, water of deliverance'). Here again the primary
object of the water control project appears to have been religious, any irrigation
benefits viewed as secondary.
The much later edict of Kusmala, from the same region , concerns a dam
associated with a Saivite temple built by the rakryaJ1 of Demung, and the provisions

17
The Gift of Water

made for its long-term upkeep. Like the other two projects on the slopes of the
mountains in this area, the construction appears to have been rather small, and to
have been undertaken at the behest of a person other than the ruler, for the benefit
of a religious foundation, although, as in the other cases, the ruler was approached
for tax transfers to support the pious construction.
The two 11th century charters of the delta region al so mention dams. The
inscription of Cane, from the early part of Airlangga's reign, mentions a dam only
incidentally. The main purpose of the tax concessions detailed in the edict was to
reward the members of the community of Cane for their loyalty and sacrifice of life
on behalf of Airlangga dllling his struggle to gain control of the state, following the
destructive attacks of 1016 A.D. The dam had already been in existence at the time
of the grant, fo rming part of a religious establishment, whose title, the holy
padadeng kadawuhan, indicates that it complised a sacred bathing place with spouts
in the form of breasts (dada) similar to those of the tirtha at Belahan on the nearby
mountain Penanggun gan (figure 2). This holy place was mentioned in the charter
as the focus for the continuity of the observance of the grant through the funding of
otferings made in the memory of Airlangga's generosity.
The charter of Kamalagyan, dated HB7 A.D., is unique in the surviving
corpus of Old Javanese language inscriptions in that it alone concerns a water
control project which involved direct royal action. It is also the only case on record
in which the primary purpose of the project was not religious. The circumstances
under which the project was carried out appear also to have been unique.
According to the inscription , a branch of the Brantas River running throu gh the
delta (probably the present Kali Pepe which discharges into the sea near Sidoarjo)
had, dmin g the noods following a rainy season, jumped its banks and settled into a
new channel to the south of the old . This sudden shift in the river caused great
hardship to those dependent upon its waters and to those unexpectedly n ooded,
including the farming communities along it) old banks, the religious establishments
which had used the river, and the major port of Hujung Galuh at its mouth. It is
stated that the local communities had tlied, more than once, to detlect the river back
into its old channel using their own labour resources, but that they had failed. At
this point they approached the ruler and asked for aid, calling upon him to use his
light to corvee labour to provide the needed manpower. This was done, and a dyke
was constructed at Walingin Sapta which det1ected the river northwards, back into
its old channel. Once the dyke was constructed, the king, won'ied about provision
for the long-term upkeep of the structure, which he apparently felt that individual
communities would not be able to financ e or for which there was inadequate
institutional basis for cooperation between so many communities, consigned the
dyke to the care of the nearby hermitage of Kamalagyan . The hermitage had its
own dam, built apparently for religious purposes, and therefore presumably had the
necessary experience. The ruler then transferred certain tax rights to the group at
Kamalagyan, adding to the grant already held by the religions foundation, in order
to compensate for the added financial and labour burdens connected with the care of
the new dyke.
In this instance also, the benefits of the dyke were refelTed to in religions terms:

This dyke was built in order to bring about benefits for the world and the revival
of all of the hol y religious foundations (along th e ri ver) ... This was brought
about throu gh the command of His Majesty, who has his capital al Kahuripan,
because he visibly showers the entire world with alllrta (i.e. the elixir of life)
which is his affection , causin g a rain of meri t.. By this construction he will

18
 Water from the ancestors: inigation in early Java and Bali

serve \.(l perfect all of the holy dlwnllIlllI foundations for the benefit of all of his
subjects ... The function of tile consu'uction is to enh;mce the splendour, which is
also hi s concern (in providing for) all of the holy dharlllfila foundatiolls ...
(BranLles 1913:LXI. 18-21).

Several points of interest arise from this inscription. The first is that the
circumstances of royal involvement in the construction of the dyke at Waringin
Sapta were apparently held to be so unusual as to call for lengthy explanation. The
fact that the local communi ties had exhausted their own resources before
approaching the ruler seems to indicate that local solutions to such problems were
the norm. Second, the fact that the structure, once built, had to be consigned to the
care of a religious institution that was probably located about ten kilometres away
and to be financed through the permanent alienation of a portion of the royal tax
base would seem to indicate that no regular administrative mechanism existed to
oversee such projects. Third, the ruler couched his explanation of his action in
religious tenns, emphasizing the benefits to the holy foundations and calling upon
the imagry of amrta, the water of life.
From Bali during this period we have only one inscription of the 10th
century which refers to the construction by a ruler of a tirtha (sacred bathing place)
at a temple (Goris 1954:2(5). Here the ruler was linked with religious water works
rather than with those used for more mundane purposes.
In sum, then, the tax charters of early Java and Bali point to a remarkably
low royal profile in respect to the pragmatic aspects of water management. Only
one charter in the surviving corpus records direct royal involvement in the
construction of a secular water control project, and this dyke was built in order to
maintain the accustomed , natural t10w of a river rather than to divert or directly
harness it. Use of this water for irrigation purposes was apparently a matter for
local initiative rather than for state direction. In the several cases recorded in which
tlle court actively encouraged the settling of certain strategic tenitOlies by converting
dry field systems to irrigated systems, this was accomplished indirectly, through
the granting of tax concessions in return for the extension of sawah cultivation. It
thus seems unlikely that the origins of these states ret1ect a perceived need for
central direction of large-scale, regional water control schemes. What coordination
was necessary - and, given the lack of pressure on land or water resources at the
time, this may have been relatively little - seems nonnally to have been provided
under the umbrella of a local agency. On Java this seems to have been the village
community with its large telTitory holdings, and on Bali it was, by the 11th century if
not before, an association ancestral to the modern subak.

Rulers, holy water, and the ancestors

One of the more interesting features of the charters surveyed above is the persistent
coupling of water and religion, a linkage which can be traced through the history of
the so-called Hindu-Buddhist religion on these two islands. On Bali, where this
religion has survived, it was until recently even called agama tirtha (the religion of
holy water) 13 The fact that the only royal official to be regularly connected with

13 See Hooykaas (1964: 139 and 148, 1968, IlJ69) OIl the central role play ed by water
(toya/tirtha=lIl1lrla) ill Balinese ritual: in tile enluUlcement of life tilHlugh purifyin g anll fertilizing
(pahersilwn) , alld as th e medium through which the soul is delivered to become an ancestor
(pallgluklllart). Note should also be taken of the association of the linggll sanc\u;u-ies Oil the

19
The Gift of Water

water in the early charters of the two islands was apparently a religious figure ,
probably a Saivite priest, may indicate that the focus upon the production of holy
water dates back at least to the 9th century. It does seem that the early Javanese and
Balinese rulers were more interested in the ritual than the practical aspects of water.
If one pursues the theme of water and religion through the archaeological
and epigraphic record, an interesting pattern emerges which will only be outlined
here. The earliest surviving inscriptions from the two islands - the Purnavarman
inscriptions of West Java, dating to the 5th century A.D. - initiate the theme. Two
of the five are inscribed on volcanic boulders sited in the beds of the tributaries of
the Cisadane River. A third , bearing the footprints of the god Indra's rain-bringing
elephant, is located at the connuence of the tributaries, just downstream from the
river-side megalithic site of Leuwiliang, which appears to have been connected in
some way with an ancestor cult. A fourth inscription stands on the banks of the
Cidanghyang (the 'holy river') a number of kilometres to the west. The fifth, the
Tugu inscription found near the north coast, records the digging of a channel named
after the holy Gomati River at a place said to have been connected with the ruler's
grandfather, who had apparently fonnerly dug a channel named after the sacred
river Candrabhaga. The description of the royal ancestor as guru and sage may link
him with the god Agastya, the holy-water bearing incarnation of Siva who was
widely popular in Java (Poerbatjaraka 1926). The link hetween water and religion
appears, in one way or another, in each of these five inscriptions.
The earliest epigraphic remains from Central Java have heen found on a
boulder beside the spring Tuk Mas on the slopes of the volcano Merhabu (figure 2).
This short inscription compares the spring to the source of the holy River Ganges.
During the 7th century a small Indian-style temple was estahlished heside a spling
at the foot of a hill near the north coast. - This was succeeded in the following
century hy the construction of a large religious complex amongst the hot splings on
the Dieng Plateau and a smaller group of temples by the sulphur springs on the
heights of Mount Ungaran (figure 2). In East Java, at the same time, the temple of
Sanggariti was established over a holy water source to the west of Malang in the
mountains, and the inscription of Dinoyo, placed by one of the sources of the
Brantas River, was dedicated to the holy water-bearing god Agastya (Bosch
1961 b) .
The temples of 9th century Central Java, and those of the succeeding
centUlies in East Java, clung to water, few of them being located far from the banks
of rivers or the edges of springs. Most of the smaller temples on the southern
slopes of Mount Merapi, as well as some of the larger temples in the Prambanan
region (figure 2), were explicitly desclibed in insCliptions not as Hindu or Buddhist
monuments (although that is the form that they took), but rather as ancestor
temples, the burial places of the ashes of dead kings whose spirits were invoked,
along with the spilits of the nam ed volcanoes, to protect the palace and to act upon a
curse. Sacred bathing places (tirtha) were increasingly ass ociated with these
temples from the 9th century, if not before, both in Central and in East Java. A mid

mountains of .lava and Bali (pm·ticuhu·ly between the rivers Petanu ;md Pakeris;m, mld see figure 3)
with ancestors (Honykaas 1964: 174); and that of sacred bathing places (tirrha) and the water of life
(amrta) with the gnd Siva (connected in Indian Hinduism with the Hi.malaY'L~ andlhe source of the
holy river G,mges) (Hnoykaas 1969); as well as the association of the lingga with sprinklers for
holy water (Hooykaas 1964: 148).

20
 Water from the ancestors: inigation in early Java and Bali

9th century inscription thought to describe the construction of Prambanan as a

funerary temple for King Pikatan, speaks of an associated tirtha wherein pilgrims
bathed in order to acquire protection (Casparis 1956:326). The remains of that
tirtha have recently been found on the southem side of the Prambanan complex. In
addition, a broad moat surrounding the nearby temple complex of Plaosan has
recently been traced, while the list of known tirtha remains in East Java lengthens
constantly. Many such remains have been found in the uplands around Malang , a
common theme being the association of the Sivalingga-cum-Mount Meru with amrta
(the water of life) through allusions to the stOlY of the chuming of the sea of milk in
the Adiparwa, in which Meru , noating on a turtle's back, was used by the gods to
chum tlle amrta from the sea. Other themes relate to the story of the capture of the
amrta by Garuda and the story of Bhima's search for immortality or deliverance
through amrta. The breasts of the lice goddess or Laksmi and lingga figures both
feature as spouts in so me of the tirtha complexes 14 The tirtha continued in use
after this period , their ancestor connections becoming more explicit by the 15th
century: inscriptions found at the 15th century temples of Sukuh and Ceto, high on
the slopes of the volcano Lawu (figure 2), mention tirtha.\· in co nnec tion with the
deliverance of souls 15, and the tirtha of Walandit near Malang, built sometime in the
10th century , is still associated by the Tenggerese wi th the ancestor cult focused on
the volcano Mount Bromo (figure 2).
The link between dead kings, ancestors, volcanoes, water and fertility is
even more marked in Bali, where most temples contain a tirtha or generate holy
water and venerate ancestors who are in turn associated with the dominant
volcanoes upstream . On a more practical level, regional cooperation in inigation in
present day Bali is articulated by a hierarchy of water temples, which mediate
between the separate .I'ubak associations (Lansing 1987). If regional cooperation
was required for inigation in early Java or Bali, it seems more likely, on the basis
of the epigraphic record, that this coordination was somehow dealt with by those
affected themselves, either directly or through temple associations, than it is that the
ruler was involved in the pragmatic operations. However, rulers had an intimate

14 Jalatunda and Belahan 011 the slopes of mount Penangg ungan me perhaps the best known of
these: tlle central spout at .Talalunda is in the form of a lingga-Meru amI one of those at Belahan is
tlle breasts of the goddess Sri or Laksmi . BOtll ti rllw .\· give prominence also to Gw-uda. The reliefs
at .T alatunda, in addition, form a genealogy, app'U·enlly linking the ruler buried at the temple with
the Pandawa line of the Malw/JIulrllto (Bosch 1961a). Other sacred bauling places in the Mal'Ulg
region include WahUldit , Tlngorejo, Torong, SumberawaJl, Biru 'Uld man y lesser-Jmown ones. The
well-k:l1owl1 rirrha of Panat'U·,m lies to the south-west on the slopes of Mount Kelud. Candi Jawi
nem Mahmg was moated, ,md Cmdi Tikus in the lowlands ne,H' Trawulan forms a rirrha about a
small temple structure resembling a linggu-Meru. The numerous stone and bronze holy water
vessels surviving from the Majapahit period in East Java link the religion of the time to the more
recent UgWlllI lirt!/.({ or Bali , as does th e persistant theme of wa ter/alll rlil linked to
deliverancelimmortalit y/ances tors in th e reliefs found on the fun ermy templ es of East Java,
matched by the th eme of fertilit y linked to the water spouts in the form of breas ts of th e rice
goddess and of lingga-mountains (for these !<Lst see not only .Talatunda mId Candi Tikus, but also
the spout·from Sirall Kencong now in ule National Museum in Jak,u·ta).

15 The Bhima and G'U·uda cult fi gures of C<mdi Sukuh both relate to the se;u·ch for IIlIlrta , the
water of deli veranceli mmortality (Stull erheim 1956), while at Candi Ceto, further up Mount
Lawu, symbols associated wi th the cult of the ancestors included such 'Unphibious or semi-aquatic
animals as crabs, frog s and turtles.

21
The Gift of Water

link with water through their role in religion. Living rulers built or encouraged the
building of holy places with water, and they alienated part of their tax base to
ensure the perpetual funding of these holy centres. Dead rulers inhabited them.
Water fOlmed a major link between the dead and the living, between the immortality
of the spirit and the fertility of the soil.

REFERENCES

Barrett Jones, A.M. (1984) Early tenth century Ja va fi'om the inscriptions
(Verhandelingen van het Koninklijk Instituut voor Taal- , Land- en Volkenkunde
107) Foris: Dordrecht.

Benda, H.I. (1962) 'The structure of Southeast Asian history ', Journal ()f
Southeast Asian History 3(1): 106-38.

Boechari, M. (1986) Prasasti Koleksi Mus eum Nasional, jilid 1, Museum

Nasional: Jakarta.

Bosch, F.D.K. (196 1a) 'The Old Javanese bathing-place lalatunda', in: F.D.K.
Bosch Selected studies in Indonesian archaeology, Martinus Nijhoff: The Hague,
pp.47-107.

Bosch, F.D.K. (196 1b) 'Guru, trident and spring', in F.D.K. Bosch Selec ted
studies in Indonesian archaeology, Martinlls Nijhoff: The Hague, pp. 153-170.

Brandes, J.L.A. (ed. N .I. Krom) (1913) Oud-Javaansch e Oorkonden,

Verhandelingen van het Bataviaasch Genootschap van Kunsten en Wetensbhappen
60: The Hague.

Brett, J. Prill (1985) 'Stone walls and waterfalls: irrigation and ritual regulation in
the Central Cordillera, northem Philippines', in: K.L. Hutterer, A.T. Rambo , and
G. Lovelace (e ds.) Cultu ral valu es and hum.an eco logy in Southeast Asia,
University of Michigan Papers on Southeast Asia 27, pp. 125- 155.

Callenfels, P.V. van Stein (1919) 'De inscriptie van Kandangan', Tijdschrift voor
lndische Taal-, Land en Volkenkunde 8: 339-47.

Callenfels, P.V. van Stein (1926) Epigraphia Bediea I (Verhandelingen van het
Bataviaasch Genootschap van Kunsten en Wetenschappen 66 III: 1-70).

Callenfels, P.V. van Stein (1934) 'De Inscriptie van Soekaboemi' , Mededeelingen
der Koninklijk Akademie van Wetenschappen, afdeeling Letterkunde 78 (serie
B):1l5-130.

Casparis J.G. de (1956) Prasasti Indonesia II: selected inscriptiol1s trom the 7th to

22
 Water fro m the ancestors: inigation in early Java and Bali

the 9th century A.D. Masa BarLl : Bandung.

Christie, J. Wisseman (1982) 'Patterns of trade in western Indonesia : ninth

through thirteenth centuries A.D.', PhD Thesis: University of London.

Coedes, G. (1968) The lndianized states or Southeast Asia, East-West Center

Press: Honol ulu.

Cohen Stuart, A.B. (1875) Kawl Oorkonden in facsimilie met inleiding en

transscriptie, Blill: Leiden.

Damais , L.-c. (l97()) Repertoire onol7la.l'tique de l'epigraphie Javwwise, Ecole

Fran<;aise d'Extreme-Olient: Palis.

Geertz, C. (1980) Negara. the theatre state in nineteenth- century Bali, Princeton
University Press: Princeton.

Goris , R. (1954) Prasasti Bali I: in scrip ties voor Anak Wung,I'u, Masa Barn:
Bandung.

Goris, R. (1965) Ancient history or Bali, Udayana University: Denpasar (revised

edition). . .

Heine-Geldern R. (1943) 'Conceptions of state and kingship in Southeast Asia',

Far Eastern Quarterly 2: 15-30.

Hill, R.D. (1977) Rice in Mala ya: {{ study in historical geog raphy, Oxford
University Press: Ku ala Lumpur. .

Hirth, F. and Rockhill, W.W . (Trans. and edits.) (1911) Chao Ju-kua: his work on
the Chinese and Arab trade in the twelfth and thirteenth centuries, entitled 'Ch u~f(m­
chi', Paragon replint: New York (1966).

Ho Ping-ti (1956-7) 'Early ripening rice in Chinese history' , Economic History

Review 9: 2()()-16.

Hooykaas. C. (1964) Agama Tirtha: five studies in Hindu-Balinese religion, N.V.

Noord-Hollandsche Uitgevers Maatschappi.i: Amsterdam.

Hooykaas. C. (1968) 'Brahman ritual in Bali', Madjalah llmu-ilmu Sastra

indonesia 4(3) : 187-99.

Hooykaas. C. (1969) 'An exorcistic litany from Bali', Bijdragen 125 (3): 356-70.

Hugo, G .J. , Hull, T.H., Hull , V.J. and Jones, G.W . (1987) Th e demographic
dimension in Indonesian development, Oxford University Press: Singapore.

Lansing, S. (1987) 'Balinese "water temples" and the management of iLTigation ',
American Anthropologist 82(2): 326-41.
The Gift of Water

Machi Suhadi and Soekarto, M.M. (1986) Laporan Penelitian Epigraphi Jawa
Tengah (Berita Penelitian Arkeologi 37) Departemen Pendidikan dan Kebudayaan:
Jakarta.

Meer, N.C. van Setten van der (1979) Sawah cultivation in ancient Java: aspects of
development during the Indo-Javanese period, 5th to 15th century, Australian
National University, Faculty of Asian Studies, Oriental Monograph Selies 22.

Menill, E.D. (1954) 'The botany of Cook's voyages', Chronica Botanica 4(5-6):
161-384.

Mills, lV.G. (trans. and edit.) (1970) Ma Huan: Ying-yai Sheng-Ian: the overall
survey of the ocean's shores (1433), Cambridge University Press for Hak1uyt
Society: Camblidge.

Naerssen, FH. van (1941) Oudj{{v{{ansche Oorkonden in Duitsche en Deensche

Verzamelingen., Proefschlift: Leiden.

Naerssen , FH. van (1977) 'The econo mic and administrative history of early
Indonesia' , in: F.H. van Naerssen and longh , R.C. de (edits.) Th e economic (u71l
administrative history of early Indonesia , Brill: Leiden, pp. J-84.

Nossin, J. and Voute, C. (1986) 'No tes on the geomorphology of the Borobudur
plain (Central Java, Indonesia) in an archaeological and historical context',
Proceedings of the symposium on remote sensing for resources development and
environmental management, Enscede, Netherlands, pp. 857-864.

Pigeaud, Th. G. Th. (1960) Java in the 14th century (th e Nagam-kertagama by
Rakawi Prapanca ofMajapahit, 1365 A.D.), 5 vols, Martinus Nijhoff: The Hague.

Poerba~jaraka (1926) Agastya in den Archipel, Proefschrift: University of Leiden.

Raft1es, T.S . (1817) Th e history of JUV({, 2 vols, Black, Parbury and Allen:
London.

Sarkar, H.B. (1971) Cmpus of the inscriptions of Java , 2 vols , Firma K.L.
Mukhopadhyay: Calcutta.

Steinmann, A. (1934) 'De op de Borobudur afgebee1de plantenwere1d', Tijdschr~ft

voor Indische T{wl-, Land- en Volkenkunde 74: 101-22.

Stutterheim. W.F. (1928) 'Transscriptie van een defecte Dorkonde op bronzen

platen uit het MaJangsche', Oudheidkundig Verslag: 105-8.

Stutterheim. W.F. (1956) 'An ancient Javanese Bhima cult', in: W.F. Stutterheim
(edit.) Studies in Indonesian archaeology, Martinus Nijhoff: The Hague, pp. 105-
143.

Sukarto K. Atmodjo, M.M. (1986) 'Some short notes on agricultural data from
ancient Balinese inscliptions', in: Sartono Kartodirdgjo (edit.) Papers of the Fourth

24
 Water from the ancestors: inigation in early lava and Bali

Indonesian-Dutch History Conference (Yogyakarta 24-29 July 1983) Volume 1:

Agrarian History, Gadjah Mada University Press: Yogyakarta, pp. 25-62.

Tichelman , F. (1980) The social evolution of Indonesia: the Asiatic mode ()f
production and its legacy, Martinus Nijhoff: The Hague.

Zoetmulder, P.l. (1982) Old Javanese-English Dictionary, 2 vols, Martinus

Nijhoff: 's-Gravenhage.

25
 THE GIFT OF WATER:
WATER MANAGEMENT, COSMOLOGY AND THE
STATE
IN SOUTH EAST ASIA

Edited by
Jonathan Rigg

SCHOOL OF ORIENTAL AND AFRICAN STUDIES

UNIVERSITY OF LONDON
1992

iii