J .

STEPHEN
LANSING
University of Southern California

Balinese “Water Temples” and the Management

of Irrigation

Bali has jigured prominently in debates on the question of whether irrigation centralizes state
power. New evidence shows that irrigation is actually organized by networks of ‘‘water temples”
that constitute an institutional system separatejiom the state. Earlier attempts to identtb a dis-
crete system of irrigation management misconceived the problem. For most crops, irrigation simply
provides water for the plant’s roots. But in a Balinese rice terrace, water is used to construct a
complex, pulsed arttjhal ecosystem. Water temples manipulate the states of the system, at as-
cending levels in regional hierarchies. The pennanence of water temple networks contrasts sharply
with the instability of the traditional Balinese states. Since the water temples are real, perhaps it
is the Balinese “state” that is chimerical.

If one tries to erect a theory of power one will always be obliged to view it as emerging at a given
place and time and hence to deduce it, to reconstruct its genesis. But if power is in reality an
open, more-or-less coordinated (in the event, no doubt, ill-coordinated)cluster of relations, then
the only problem is to provide oneself with a grid of analysis which makes possible an analytic
of the relations of power.
-Michel Foucault, PowerlKnowledge

0 NE OF THE MOST ENDURING THEORIES OF THE STATE links the management of

hydraulic irrigation to the centralization of power. For Marx, “the prime necessity
of an economical and common use of water . . . necessitated in the Orient . . . the cen-
tralizing power ofGovernment” (Avinieri 1969:7). But Marx was faced with the difficulty
(pointed out to him by Engels) that his model of an “Asiatic mode of production” could
not apply to existing states, then under European colonial rule. Marx replied that an
“
intact example” could still be found on the Indonesian island of Bali (Avinieri
1969:456). A century later, Karl Wittfogel found himself in a similar bind: his model of
“Oriental Despotism” described early states, not those of the 20th century. Like Marx,
he suggested that a still-functioning example could be found on Bali (Wittfogel 1957:53-
54).
But despite a century of study, there is still no consensus as to whether irrigation is
really centrally organized in Bali. Indeed, most scholars have argued the reverse: that
Balinese irrigation is entirely in the hands of local-level farmer’s associations called su-
baks.’ In a recent cross-cultural comparison of irrigation systems, Hunt and Hunt con-
cluded that the evidence from Bali remains ambiguous (1976:394; see also Happe 1919;
van der Heijden 1924-25; Wirz 1927; Korn 1932; Millon 1962; Geertz 1972, 1980; Bir-
kelbach 1973; Hobart 1982).

J. STEPHENLANSING
is Associate Professor, Department of Anthropology, University of Southern California, Universip Park,
Los An&, C A 90089-0661.

326
Lansing] BAUNLBE TEMPLES”
“WATER 327

To put some flesh on the question, consider the irrigation systems of the Balinese dis-
trict of Badung (Figure 1). Badung is 115 km long and up to 40 km across at the widest
point. Several small rivers provide water for intricate networks of canals which irrigate
19,238 hectares (ha) of rice terraces. The largest subuk in Badung, however, controls only
328 ha of terraces. Altogether there are presently 151 subaks in Badung, ranging in size
from 328 down to 10.5 ha.
The question of whether these 151 subaks could function as completely autonomous
units brings us to the question of the role of water management in wet-rice agriculture.
For most crops, irrigation simply provides water for absorption by the plant’s roots. But
in a Balinese rice terrace, water is used to create an artificial ecosystem. The precise al-
ternation of wet and dry phases achieved through controlled irrigation governs the basic
biochemical processes of the terrace ecosystem. Water alters soil pH, induces a cycle of
aerobic and anaerobic conditions in the soil which determines the activity of microorga-
nisms, provides a range of mineral nutrients, fosters the growth of nitrogen-fixing algaes,
excludes weeds, stabilizes soil temperature, and over the long term governs the formation
of the plough pan and the maintenance of soil fertility (DeDatta 1981:297-298; Yoshida
1981). Phosphorus, for example, is increased from less than 0.05 ppm to about 0.6 ppm
by submergence (Yoshida 1981:147-151), while potassium depends largely on drainage.
Furthermore, the Balinese do not use storage devices, so that the success of irrigation
depends on accurate judgment of the (seasonal) flow of rivers and springs. Altogether, a
difference of a few centimeters in water depth, or a change of a few weeks in timing the
alternation of wet and dry cycles, can have a major effect on the terrace ecosystem, di-
rectly affecting crop yields.
The most recent controversy over Balinese irrigation centers on the possible role of
agricultural rituals as a scheduling mechanism. The points at issue in this dispute lead
directly into the results of my own research, concerning the role of regional networks of
“water temples” as managers of the terrace ecosystems.

Figure 1
Bdi.
328 AMERICAN
ANTHROPOLOGIST [89, 1987

Hobart’s Critique of Gee&

Over the past two decades, Clifford Geertz has developed the hypothesis that irrigation
is organized at the subak level by the timing of the rituals connected with the rice cult,
which “are symbolically linked to cultivation in a way that locks the pace of that culti-
vation into a firm, explicit rhythm.” Thus, according to Geertz, “a complex ecological
order was both reflected in and shaped by an equally complex ritual order, which at once
grew out of it and was imposed upon it” (1980:82). This argument was recently chal-
lenged by Mark Hobart, on the basis of Hobart’s study of a single subak in the district of
Gianyar, comprising less than 85 ha of terraces. Hobart’s critique was based on a com-
parison of the rituals of the rice cult to the physiological growth of the plant in the fields,
and the timing of agricultural labor. Hobart found that the intervals marked by the rice
rituals did not match the phases of agricultural labor, and perhaps more importantly,
that the rituals were very often quite out of phase with the stages of growth of the plants
(1982:73). Hobart concluded that ritual was therefore not a “master plan for cultiva-
tion.”
According to Hobart, the reason for the poor match between rituals, labor, and the
growth of the plants was that the rituals do not really follow the natural rhythm of plant
growth, but are instead pegged to the Balinese calendars. He observes that different
phases of the “rice cult” make use of both of the Balinese calendars, the “Hindu system
of 12 solar-lunar months . . . and the so-called Javanese-Balinese uku calendar,” which
is permutational and therefore “completely divorced from the flow of observable natural
events” (1982:57). In the subak studied by Hobart,
both calendrical systems intersect in the organization of the rice-cycle rites. So, for example, the
main temple festival (pi)odalan agung occurs in Purnama kedasa, or every full moon of the tenth
solar-lunar month. . . . Although Geertz has argued that the ceremonies are synchronized with
the stages of cultivation, in Tengah Padang at least, there is no simple correspondence, for the
rites follow a largely predetermined pattern. [ 1982:57]
As an example of the bewildering ritual complexity generated by following two calen-
dars at once, Hobart points to a particular ritual, nyungsung, “supposedly co-ordinated
with the ‘pregnancy’ of the rice, when the growing panicle causes the rice to swell.” To
“confuse neat theories of timing still further,” observes Hobart, “this ritual is supposed
to follow the solar-lunar calendar for the first planting, and the permutational uku cal-
endar for the second!” (1982:57).
By an odd coincidence, this particular piece of evidence (which Hobart uses to illus-
trate the detachment of Balinese rice ritual from cultivation practices) happens to be the
most critical piece of planning in the regional management of irrigation. The timing of
the nyungsung ceremony is one of the key pieces in the irrigation puzzle. But the signifi-
cance of this timing appears only at the regional level, as one of the instruments of terrace
management utilized by a regional network of water temples.
Thus Hobart notes elsewhere that the nyungsung ceremony for the first rice crop is timed
for the full moon of the tenth solar-lunar month, which coincides with the major subak
temple festival (1982:70). To understand the significance of this timing, we need to con-
sider the physiology of the rice plant, the climate of Bali, the Balinese calendrical system,
and the role of temples. Let us take them in order.
As Hobart notes, nyungsung is “supposedly co-ordinated with the ‘pregnancy’ of the
rice, when the growing panicle causes the rice to swell” (1982:57). The appearance of the
panicle primordia marks the beginning of the reproductive growth phase, a turning point
in the need of the plant for two crucial requirements: water and sunlight. The plant’s
need for water is critical through the development of the panicle, but declines thereafter
(DeDatta 1981:314). In contrast, the plant’s need for sunlight increases at this stage:
The solar radiation requirementsofa rice crop differ from one growth stage to another. Shading
during the vegetative stage only slightly affects yield and yield components. Shading during the
reproductive stage, however, has a pronounced effect on spikelet number. . . . Solar radiation at
Lansing] BALINESE
“WATERTEMPLES” 329

the reproductive stage has the greatest effect on grain yield; that at the ripening stage, the next
highest yield; and that at the vegetative stage, an extremely small overall effect. [Yoshida
1981:87]
Hobart noted that the ceremony marking the onset of the reproductive phase is timed
for the full moon of the tenth lunar-solar month. This date falls somewhere between the
end of February and the beginning of May on our Gregorian calendar. Figure 1 shows
rainfall patterns for the different regions of Bali, which differ markedly, with the greatest
amounts of rain falling in the mountains of south-central Bali. Table 1 indicates the
monthly mean annual rainfall by region.
Despite regional differences, it is clear that the rainy season tapers off abruptly in
March and April. The subak studied by Hobart is located in Zone 2: High Rainfall. Data
on sunshine for rice-growing areas are summarized in Table 2.
All other things being equal, the sunshine and water needs of the rice plant will ob-
viously be best satisfied if the reproductive phase begins just at the end of the rainy sea-
son, around the month of March on our calendar. The choice of the full moon of the tenth
Icaka month for the festival marking the appearance of the panicle, along with the major
local water temple festival, optimizes these factors. But in fact this is only the beginning
of the story. Not only Hobart’s subak but all of the subaks in the district of Gianyar, tra-
ditionally held the panicle-appearance ceremony on the full moon of the tenth Icaka
month. This meant that the whole district planted their first rice crop at the same time.
If they all planted the same variety of rice, they would therefore also harvest at the same
time: in mid-summer, when the danger of rainfall is minimal, and sunlight is assured for
ripening. The fields would be dry, facilitating harvest and drying. More importantly, a
regional fallow period would follow the harvest extending over approximately 15,000 con-
tiguous ha, thus effectively interrupting the food supply and/or life cycle of all the major
rice pests (insects, rodents, bacterial and viral diseases) (Oka 1979). Specifically, then,
timing the nyungsung ceremony for the full moon of the tenth month for an entire region

Table 1
Monthly mean rainfall by region (mm)

Zone Oct Nov Dec Jan Feb Mar Apr May Jun Ju1 Aug
I
Low rain 72 178 200 252 230 232 187 113 99 77 95 65
High rain 65 145 253 378 365 343 315 190 190 85 98 73
I1
Low rain 23 56 149 247 350 303 296 137 77 47 42 23
High rain 110 156 168 234 254 252 162 116 116 140 188 104
IV
Low rain 8 28 74 176 275 282 233 110 77 38 31 18
High rain 43 98 213 252 296 265 248 172 115 74 41 33

No irrigation in Zone 111.

Table 2
Seasonal sunshine by region.

Zone I Zone I1 Zone IV

Mean sunshine 60% 76% 50%
Minimum Jan (45%) Feb (58Y0) Jan (40%)
Maximum Apr (67%) Apr (89%) Aug (75%)
330 AMERICAN ANTHROPOLOGIST [89, 1987

has the following effects: (1) the plant’s needs for rainfall and sunshine are optimized; (2)
harvest is timed for the dry season; (3) pest populations are minimized.
But most Balinese farmers aim to get two rice harvests, not one. Traditionally, the first
crop is a particularly long-maturing variety, which allegedly compensates the farmers for
the extra time it takes to grow by being especially nutritious-good-tasting and life-sus-
taining. This rice, calledpadi tahun orpadi dcl, matures in about 200 days, but the Balinese
allow 210 days (six 35-day Balinese months) for the complete cycle. The second planting
depends, not on rainfall, but on organized irrigation in the dry season. Each subak may
select the variety of rice to be planted. These rices are generally termed cicih, and have an
average duration of about 120 days. Cicih are nonphotosensitive varieties, and so may be
planted at any time of year. But if two rice crops are to be harvested in one year, this
leaves no more than a month of leeway for a fallow period between plantings. It is critical
for pest control that this brief fallow period extend over a wide region, since otherwise
pests will simply migrate from field to field. Obtaining two crops of rice every year thus
requires very nice timing, particularly if dry-season irrigation requires a rotation of water
turns.
In summary, all three factors-optimization of water and sunlight for the crucial first
crop, provision for a brief but effective regional fallow period between crops, and coor-
dination to permit a possible second crop of fast-maturing rice-depend on timing the
first planting so that the reproductive phase (panicle appearance, marked by the nyung-
sung ceremony) occurs at the end of the rainy season, at the full moon of the tenth month.
However, it is clear from Tables 1 and 2 that there is at least a month’s leeway as to when
the rainy season actually ends. The reader is asked to bear this in mind, as we consider
the role of the “master water temple” Pura Ulun Danu Batur in setting the calendar each
year.
The “Master Water Temple” Pura Ulun Danu Batur
Along the rim of the crater of Lake Batur (see Figure l ) , well above the elevation where
rice can be grown, there is a temple called Pura Ulun Danu Batur, which is generally
recognized as the supreme water temple (or subak temple) of Bali. The Balinese name for
their religion is agama tirtha, the religion of holy water. Water is revered for its power to
make things grow, and to wash away impurities both physical and spiritual. Pura Ulun
Danu Batur includes shrines to 147 deities, foremost among them the supreme goddess
associated with the life-giving properties ofwater, Dewi Danu. Temple scribes keep a list
of the 204 subaks that constitute the primary congregation of the temple. Once a year,
around the end of the rainy season, the scribes write formal messages to each of these 204
subaks, inviting them to attend the major annual festival of the temple. These invitations,
written on palm-leaf lontar manuscripts in Balinese script, are then hand-carried to the
subaks by the temple messengers. No other temple in Bali regularly sends such invitations.
The 204 sub& which receive the invitations (about half of all Balinese subaks) are lo-
cated between the boundaries of four rivers, and are members of the temple’s congrega-
tion because the goddess of Lake Batur is believed to be responsible for the gift of the
waters that irrigate their fields. But the question is, why send formal written invitations
to a festival that has been held on the same date for centuries-especially since no other
Balinese temple issues such invitations? The text of the invitation itself provides a clue:
May there be peace.
Letter of invitation, to the head of village or subak X. My purpose is to inform you that we shall
hold the “Temple Festival ofthe Tenth Month” on the following date [date specified on the 210-
day uku calendar]. You are requested to bring ten coconuts, 1100 Chinese coins, ten measures of
rice, one duck and one pig.
Myself, the High Priest of Batur Temple
The request for offerings is formulaic, and nearly every subak brings more than is de-
manded. But the specification of the date provides real information, and (in my opinion)
Lansing] BAUNESE“WATER TEMPLES” 33 1

is the real reason for the use of these invitations. An explanation requires a brief digres-
sion into the workings of the Balinese calendar.
As noted by Hobart, the Balinese actually use two calendars. The first, called uku, is a
210-day calendar consisting of 30 7-day weeks. This calendar is independent of natural
or astronomical events, and simply repeats itself indefinitely in 210-day cycles. As we
shall see presently, the choice of a basic interval of 210 days seems related to the 2 10-day
growth cycle of first-crop Balinese rice.
The Balinese also keep track of lunar months and solar years. The lunar calendar is
accurate enough to stay in agreement with the actual phases of the moon, and so to check
the date according to this calendar, in principle one need only look at the sky. But cal-
culation of the solar year is much less straightforward. The Balinese use a version of the
Indian Icaka h i - s o l a r calendar, which adds an intercalary month every two or three
years, so that while over 10 or 100 years, the passage of solar years is quite accurately
counted, in any given year the marked “year” is either too short by 10 or 11 days, or too
long by about 19 or 20 days. Moreover, the relationship of luni-solar (Icaka) months to
the observed seasons fluctuates markedly. Thus, for example, the date called “Full Moon
of the Tenth Month” may fall anywhere between the end of February and the beginning
of May, and will change each year. One of the effects of this calendrical complexity is that
the man in the rice fields is seldom sure exactly which month it is (on the luni-solar Icaka
calendar), although he knows very well what day it is on the lunar month, and on the
210-day calendar. Keeping track of the insertion of intercalary months so as to get the
year right is the province of Brahmanical specialists. Moreover, in the days of the rajahs
each court had its own calendrical experts, whose systems often disagreed so that dates
often varied by as much as a month, from one principality to the next.
To be precise, the point of uncertainty is the question of which month (marked by the
phases of the moon) corresponds to which month (on the numbered, luni-solar Icaka
calendar). The Sanskrit word for “Ten” (dara), as in “Tenth Month,” is close to the Ba-
linese word dardar, which means “to break off, to end or finish.” Moreover, the “Tenth
Month” in the (originally Indian) luni-solar calendar falls around March-April, approx-
imately the end of the rainy season in Bali. So in each locality, the question of which
month is the tenth month is resolved by estimating when the rainy season is actually
about to end, as evidenced by natural signs. In one locality, the “tenth month” is marked
by the appearance of a particular moss on a venerable and authoritative old tree; in an-
other, by the appearance of “Dasa grass”; in another, by a change in the color of the sea.
Even today, despite the widespread use of a printed, standardized calendar all over the
island, it is still possible to hear someone say, “Well, it may be the tenth month down
there, but around here it’s still the ninth month!”
This does not mean that the Brahmanical experts lose track of the actual passage of
Icaka months. The passage of luni-solar years is accurately recorded, and anyone wishing
to know the “correct” date may consult the printed calendar (or get a second opinion
from a calendrical specialist). However, the printed calendar has only become widely
used in the last two decades, so that formerly each locality had its own view as to what
time it was.
This clearly posed problems for the coordination of irrigation over river systems that
encompassed several principalities, each engrossed in its own calendrical system. Hence
the need for the annual “invitations” from the master water temple to its member subaks,
which by pegging the “Full Moon of the Tenth Month” to a particular date (on the in-
variant uku calendar), established a sort of “Irrigation Year”-a common calendrical
framework for farmers, essential for scheduling water turns, cropping patterns, and es-
pecially for ensuring uniform fallow periods. As the priests of the temple explained to me,
this also enables the temple to bring the “Tenth Month” into conjunction with their es-
timate of the end of the rainy season (musim mokoh). Temple priests decide each year on
the likely duration of the rainy season and fix the date of the temple festival accordingly.
332 AMERICAN
ANTHROPOLOGIST [89, 1987

It is at this point that the second calendar-the 2 10-day uku permutational calendar-
reveals its uses. Since this calendar is independent of the seasons, it has no particular
starting or stopping date.’ Start on any given date, and the calendar helps mark out in-
tervals (useful for regulating water turns, phases of rice growth, etc.) for the next 210
days. As far as the farmers I’ve talked to are concerned, the reason for the 2 10-day length
of this calendar is obvious: it marks the passage of six 35-day “Balinese months,” the
growing cycle of first-crop rice. At the end of this 210-day cycle, the uku calendar is used
again, to mark the duration of the regional fallow period, and then to establish planting
dates and water rotations for the second crop. This explains Hobart’s second point: that
the panicle-appearance nyungsung ceremony is held on full moon of the tenth month for
the first planting, and fixed according to the uku calendar for the second planting.
So the uku calendar is annually pegged to the luni-solar calendar at the end of the rainy
season, and thenceforth enables the farmers to coordinate their labors, water turns, and
rice-cult rituals. The date of the second planting (called Gegadon or L‘opencycle”) is fixed
by the uku calendar, generally soon after the first harvest. If plantings are to be staggered,
this is arranged beforehand by meetings in the water temples, the dates of each water
turn specified on the uku calendar. Because the second crop is considered optional, the
rituals of the rice cult are generally reduced and abbreviated for the second planting.
Major water temple festivals are almost always held only once a year.

Regional Water Temple Networks

The titles and precise functions of regional water temples vary from district to district,
but the overall logic is reasonably clear. Looking at the system from the bottom up, each
farmer has a small shrine (beduguf) located at the spot where irrigation water first enters
his fields. Here, at the “upstream” end of his fields, he carries out the rituals of the “rice
cult,” from field preparation to harvest (Wirz 1927). Next, most (but not all) subaks pos-
sess temples called “Ulun Carik” (Head of the Fields), where they perform collective rites
during the first crop (kerta mma), and hold subak meetings. Next, at the place where a
major canal first enters a set of terraces there are temples called “Ulun Swi” (Head of
the Terraces), which usually coordinate irrigation for a collection of subaks. There may
also be temples called “Masceti,” which are regional water temples, sometimes equiva-
lent to “Ulun Swi,” sometimes superior or subordinate. Finally, each weir, spring, lake,
and the headwaters of each river have shrines or temples. Downstream, at the place
where important rivers reach the sea, there are major temples associated with defense
against pests and other malign influences believed to originate in the sea. Overall, then,
the water temples exemplify the basic Balinese cosmo-logic, whereby “upstream” is as-
sociated with purity and the uranic, while “downstream” is linked to chthonic dangers.
This logic, implicit in the physical layout of every temple, is imposed on every river sys-
tem and indeed on the whole island by the regional systems ofwater temples, which mark
out the paths traced by the waters of the goddess as they simultaneously cause growth
and bear away pollution.
To see how the system works in practice, let us consider a couple of regional temple
systems, and then return to the question of how the master water temple coordinates
whole river systems.
Sukawati
The village of Sukawati (see Figure 2) lies alongside the seacoast in the district of Gian-
yar (see Figure 1). It is downstream from Hobart’s subak but in the same district, in which
panicle appearance is timed for the full moon of the “tenth month,” according to the
“irrigation year” set by the master water temple at Lake Batur. There are 13 subaks in
the system, with a total of 402.86 ha of rice terraces. Water is obtained from weirs on two
rivers, the Wos and the Petanu. The subaks are divided into three groups, as follows:
Lansing] BALINESE“WATERTEMPLES” 333

Figure 2
Sukawati irrigation system.

Group 1: Subaks Palak, Sango, Sungguhan, and Bubun (subdivided into two tempeks)
Source of water: Petanu River
Group 2: Subaks Laud, Landep, Somi, Juwuk
Source of water: Wos River
Group 3: Subaks Lango, Babakan, Cau Duur, Cau Beten, Lebo
Source of water: Wos River
Traditionally, all subaks plant twice, in the first and seventh Icaka months (Sasih kasa
and kapitu). Group 1 has sufficient water to plant rice both times. Groups 2 and 3 have
less water, and so alternate rice with vegetables for their second crop. That is, in a given
year Group 2 plants a second crop of rice in August, while Group 3 plants vegetables
334 AMERICAN
ANTHROPOLOGIST [89, 1987

(and obtains water every fifth day). The following year, Group 3 plants rice and Group
2 plants vegetables. The fallow period extends for only 15 days, between the first and
second plantings, to maximize water use.
Figure 2 shows the location of these subaks, and of two types of local water temples:
Ulun Carik and Masceti. There are four Ulun Carik (Head of the Fields) temples, where
subak members perform the rites of the rice cult (Wirz 1927). Above these four temples is
the higher-level “Masceti” temple, Pura Air Jeruk. The practical functioning of the tem-
ple system is explained by the village head in this interview, transcribed unedited from a
tape:
Village Head: The Pura Air Jeruk is the largest temple hereabouts, that is, the temple whose
congregation includes all the farmers of the village of Sukawati. Now below this temple there are
also smaller temples, which are special places of worship for the subaks-each subak has its own.
There are 14 of these temples, 14 subaks, all of which meet together as one here. They meet a t
the Temple Pura Air Jeruk. Every decision, every rule concerning planting seasons and so forth,
is always discussed here. Then, after the meeting here, decisions are carried down to each subak.
The subaks each call all their members together: “In accord with the meetings we held a t the
Temple Pura Air Jeruk, we must fix our planting dates, beginning on day one through day ten.”
For example, first subak Sango plants, then subak Somi, beginning from day 10 through day 20.
Thus it is arranged, in accordance with water and “Padewasan”-that is, the best times to plant.
Because here time controls everything. If there are many rodents and we go ahead and plant
rice, obviously we’ll get a miserable harvest. So we organize things like this: when the rodent
population is large, we see to it that we don’t plant things they can eat, so that they will all die-
I mean, actually, that their numbers will be greatly reduced, pretty quickly.
Lansing: And this is all organized by meetings . . .?
Village Head: Here! [points at the temple Pura Air Jeruk]. The meetings held here, with the lead-
ers of each subak.
Lowing: Is there a fixed schedule of meetings?
ViflagcHead: Once a “year.” Each new planting season, there is a meeting. If the planting sched-
ule is not to be changed, there is no meeting. Of course, the ceremonies held here go on regard-
less-there are two temple festivals here, a one-day festival every six months, and a three-day
festival every year. . . . This place is the home of the spirits of those who have preceded us, who
built this t e m p l e 1 would call this temple the fortress of the farmers hereabouts.
Kedewatan
Here, seven subaks share water from a single large canal originating from a major weir
about 4 km upstream, in central Gianyar (see Figure 3). Where the water first enters the
terrace complex, there is a major temple called “Ulun Swi” (Head of the Terraces).
About 100 m downstream from this temple, the main canal splits in two, and there is a
“Masceti” temple alongside the upstream branch canal. A second Masceti temple is lo-
cated about a half kilometer downstream, where the second branch canal enters the sec-
ond set of terraces. The two Masceti temples form the congregation of the Ulun Swi tem-
ple. Each subak thus belongs to the congregation of the Ulun Swi, and to one or the other
of the Masceti temples (see Table 3).
The congregation of the Ulun Swi temple thus includes seven subaks with a total of
1,775 members, farming 558.04 ha of rice fields. All subak members share equally in the
responsibility to maintain the main canal and weir. During the rainy season, the whole
Ulun Swi unit plants the same variety of rice at the same time, ensuring a uniform fallow
period after harvest to control pests. Traditionally, this planting is timed so that panicle
development occurs at “full moon of the tenth month.” For the second planting, each
Masceti acts as a unit, choosing the crops to be planted, and assigning rotational irriga-
tion if needed. Each subak (or in the case of large subaks like Pacekan, each tempek unit)
takes turns in both maintenance of the irrigation works, and annual rituals at the Masceti
and Ulun Swi temples. Moreover, each Masceti (rather than each subak) sends a dele-
gation with offerings to the annual festival of the “master water temple” at the crater
lake, Pura Ulun Danu Batur.
Lansing] BALINESE
“ W A T E R TEMPLES” 335

Ulun Swl Templo-+

+l

r 8ub.k pacrkan k40

Figure 3
Kedewatan irrigation system.

Irrigation Functions of the “Master Temple” Ulun Danu Batur

Since space is limited, I will refrain from further exploration of variant regional water
temple systems, and proceed on to the question of whether some centralized authority
exists. We have already seen that the “master water temple” Pura Ulun Danu Batur
plays a significant organizational role by setting the “irrigation year” calendar for its 204
subuks. But does the temple play an active role in irrigation management, or is it primarily
a religious institution?
To answer this question, I wish to briefly recount three occasions in which I observed
the activities of the temple priests in irrigation management.
336 AMERICAN ANTHROPOLOGIST [89, 1987

Table 3
Ulun Swi Temple.

Masceti Temple 1 Masceti Temple 2

Subak Hectares Members Subak Hectares Members
Lungsiakan 54.82 190 Mas 38.74 130
Kibul Bebek 28.37 108 Sindhujiwa 94.59 317
(includes two
“tempeks”)
Pacekan 133.41 446 Mandi 126.26 367
(includesfour Tebungkan 81.85 217
“tempeks”)
Subtotals 216.60 744 341.44 1,031

Creation of a New Subak

In April 1983 I was summoned to the master water tern le by the two high priests in
P
order to film an unusual event: the creation of a new subak. A group of dry farmers from
a hamlet in the district of Gianyar had organized themselves into a work group, with the
. ~ terraced approximately 40 ha of hillsides (on two
intention of forming a ~ u b a kThey
adjacent slopes), and hired a traditional irrigation tunnel engineer to guide them in the
construction of tunnels and canals originating at a small spring about 4 km upstream.
Before beginning work on the first tunnel, a delegation consisting of the newly elected
head of the subak and half a dozen assistants made a pilgrimage to the master temple, to
obtain blessings and practical advice on what is required to create a subak. A temple priest
inspected the proposed tunnel and terrace sites.
On the morning I arrived at the temple, two large trucks carrying the entire member-
ship of the subak arrived at the temple and made offerings to the deities concerned with
irrigation, ofwhom the foremost is Dewi Danu (Goddess of the Lake). Afterwards, I ac-
companied eight priests from the temple and the subak on a journey by truck to the spring
from which they hoped to obtain water for irrigation. The head priest selected a site for
a small shrine, the “water-origin shrine” for offerings to the Goddess of the Lake, about
10 m from the spring itself. Ground-purifying offerings were made by the whole subak,
followed by a water-augmenting ceremony climaxed by the coaxing of a live duck to swim
away downstream, in the direction of the terraces.
The duck was soon followed downstream by a procession consisting of the subak, the
temple priests, anthropologist, and film crew, to the site of the new terraces. The high
priest, who is regarded as an irrigation expert, began by surveying the terraces and the
unfinished small tunnel which represented the last piece of the irrigation works. He crit-
icized the placement of the tunnel, on grounds that it was too far down the slope, and the
subak head agreed to change it. Next, he selected the site for the Ulun Swi (Head of the
Terraces) temple, and other priests actually measured out and marked off its dimensions
and explained the placement of shrines. These shrines included not only the Lake God-
dess of the master temple, but also a shrine to the principal deities of the local Masceti
(regional water temple).
Creation of a New Imgation Tunnel
Balinese irrigation tunnels often extend for hundreds of meters, and in some cases ap-
parently over a kilometer, through the volcanic rock. There are several teams of tradi-
tional engineers who are expert in the construction of these tunnels. I accompanied teams
of tunnel builders to the master temple for blessings, and observed the construction of
several new tunnels. According to the tunnel engineers and the priests, any new tunnels
Lansing] BALINESE
“WATER
TEMPLES” 337

require the sanction of the master water temple, since all water is considered a gift from
the goddess, and temple priests are responsible for mediating water rights of subaks shar-
ing the same rivers.
Pest Control
In 1979 the priests of the temple came to the conclusion that a plague of rodents threat-
ened to become widespread. Instructions were sent down to all member subaks to build a
special temporary shrine at all water inlets in every field, and perform a brief prayer and
offering every third day for 15 days. A widespread fallow period was also suggested, but
due to confusion resulting from the adoption of new high-yielding varieties of rice and
fertilizer, the fallow period was not uniformly observed. The small shrines duly ap-
peared-n time-by the thousand.
In order to provide a succinct impression of the role of the master temple in irrigation,
I have translated excerpts from an interview with the Jero Gde Alitan, one ofthe two high
priests of the temple, the acknowledged expert on irrigation mentioned above. Also pres-
ent were the chief Temple Scribe, a subak head who happened to be visiting, and a couple
of other temple priests.
On establishing a new weir:
High Priest: If someone wants to build a new weir, first he must come here to request a black
stone, to place underneath, as a symbol. The black stone is used as the foundation, and is posi-
tioned in a small ceremony. Then construction can begin-they can break ground.
Laming: Where is the ceremony held?
High Priest: There are two-here at the temple, and also at the weir. The ceremony is called
pmungkah mpelan-opening the weir. Afterwards, we build a shrine (Asaghan) at the weir, to wor-
ship the Deity who resides here at Ulun Danu, and also at Gunung Agung, and also at Masceti
temples. Every day before working on such a project, the builders must offer prayers . . .
On tunnel building:
High Priest: This was about two years ago. Here was the problem: about 12 km from here, in the
district of Bunutin (Bangli), there were two rivers. East of Bunutin there is a river, and also one
to the west. The people wanted to use the larger river. But when they invited me for the first
inspection, I saw that it wasn’t feasibleusing the larger river would be more difficult, and cost
a lot more. I figured out a way to use the smaller river, and they agreed to try it. And it worked-
they’re getting harvests already. Later we added another weir to the west of the first one. And
now water shortage is not a problem in Bunutin any more.
On settling a dispute over water rights:
High Priest: This was just downstream from Sekaan. There is a big spring there, called “Bulan.”
This spring had been used by local subaks, like Pejeng Aji. It wasn’t producing a lot of water-
about 100 liters per minute, I think. Now, downstream and off to one side there are a lot of
villages-Lukan, Jasan, Tegal Suci, Jati, Belong, Pisan . . .
Tmple Scribe: Batas, Tebuana . . .
High Priest: And they asked if some of the unused water could be brought up to them, like this
(refers to sketch). But the subaks that were already using the water didn’t want to permit this,
because it was their water. They were worried that they would get less. So they wouldn’t give
permission. This was debated until it reached the Governor. So finally I went. I said to them,
“Who created this water? Who decides if this spring is full, or dries up?’ And they had to answer!
I said, “DOyou understand that ifwe fight over this gift from the goddess, her spring might just
dry up? Completely vanish?” I brought them all up to the temple here, and when we had it
settled, work began. The new canal ran off below the spring, and took off quite a lot of water.
But-now this is the point-not 200 m further downstream, the flow was back to normal. In fact,
it actually increased after everything was finished! I tell you, that made quite an impression. The
head of that subak-what’s his n a m e h e was here just two days ago . . .
Temple Snibe: That was around 1968. But they all come-if we have to repair part of the temple,
or if we have a big ritual, they don’t fail to send a big delegation to help.
High Priest: They’re still afraid that the spring might dry up!
338 AMERICAN
ANTHROPOLOGIST [89, 1987

O n the plague of rodents in 1979 (see above):

Subak Head: It was all over [the region of] Gianyar! In my subak, for a long time the numbers of
rats had been growing . . .
Anotherpriest: Around Kemenuh, and all the downstream subaks . . .
Anotherpriest: Actually, from Payangan all the way to the sea!
Subak Head (speaking to the High Priest): And the Government helped out, right?
High Priest: No, it wasn’t a government matter-it was a subak problem. We sent instructions to
every subak. In truth, the government helped cause the problem, upsetting the fallow periods . . .

Conclusion
. . . dening ida amedalang ring tirtha iki sapunika ne kagaduh gaman ida, yan sira apreduha ring gaman sira,
nora wnang malih dwen ida carike
. . . because the goddess makes the waters flow, those who do not follow her laws may not possess
her rice terraces
-Rajapurana Ulun Danu Batur, Vol. II:24 28.b.l
Hitherto the debate on Balinese irrigation has focused on two alternatives: the “cen-
tralized” model, in which irrigation is managed by a state bureaucracy (Marx, Wittfo-
gel), and the “decentralized” model, in which each subak is an autonomous unit (Geertz,
Hobart). The evidence I have submitted suggests a third alternative: irrigation is cen-
trally organized by a system of water temples, separate from the state. The first question
a skeptic might ask is why the existence of the water temple system had not been reported
earlier. Part of the answer doubtless lies in the fact that previous studies concentrated on
individual villages rather than the regional patterns of irrigation-a problem that Hunt
(1976:398) and Coward (1980:8) suggest is common for anthropological studies of irri-
gation. But more fundamentally, the attempt to identify a discrete system of irrigation
management may have misconceived the problem. “Irrigation management” simply
means providing water for crops. But as we have seen, Balinese water temples do a great
deal more than this. Viewed from the standpoint of systems ecology, a rice terrace is a
complex artificial ecosystem. The water temples make decisions which manipulate the
states of the system, at ascending levels in regional hierarchies. For example, when the
regional water temple Masceti Air Jeruk institutes a 15-day fallow period over 500 ha to
control a pest outbreak, it is managing the regional terrace ecosystem, not just irrigation.
There are other reasons why the water temple networks should have remained nearly
invisible to outsiders. Consider the temples called “Sea Temples” (pura segara), usually
located on a beach or islet near a river outlet. Superficially, there is nothing to connect
these temples with irrigation, for they play no direct role in agricultural management.
But by marking the downstream terminus of a water temple network, they play an im-
portant part in the internal logic of the water temple system. The ability of water temple
networks to function as ecosystem regulators is predicated on fulfilling a role in Balinese
cosmology which places the instrumental logic of agricultural decisions in a wider reli-
gious context.
But is the system really separate from the state? I n response to this question, the priest
of the temple Masceti Pamos Apuh told the following story:
Around the year 1870, the neighboring princedoms of Bangli and Gianyar were at war with one
another. The irrigation systems which provide water for Gianyar originate in Bangli, and Bangli
is also the home of the “master temple” Ulun Danu Batur. Hostilities continued through the
rainy season, so farmers from Gianyar were afraid to pass through Bangli on their way to the
tenth-month festival at Ulun Danu Batur. Near the border of the two princedoms, where the
0 0 s river leaves Bangli and enters Gianyar, there is a regional water temple, Pura Masceti Pa-
mos Apuh. The farmers ofGianyar held special ceremonies at this, the most “upstream” of their
water temples, to transform it into a temporary “way-station” (Pcnyawangan) temple for the God-
dess of the Lake and her retinue. For as long as hostilities continued between Bangli and Gian-
yar, the temple functioned as a substitute “master temple” for the subaks of Gianyar.
Lansing] BALINESE“WATER TEMPLES” 339

The very detachment of the water temple system from the political order has undoubt-
edly helped to preserve it, even after the demise of the traditional Balinese states (Lansing
1983). But recently, the “invisibility” of the system has worked to its disadvantage. Be-
ginning in the 1970s, the Balinese were encouraged to increase rice production by adopt-
ing new high-yielding varieties of rice, along with new cropping patterns based on use of
commercial fertilizers and pesticides. Most recently, a series of studies by foreign con-
sultants have recommended changes in irrigation management. For these consultants,
the water temple system is indeed invisible. Irrigation development plans invariably as-
sume that the individual subaks are the highest-level “traditional” Balinese institutions
concerned with irrigation. Major changes in irrigation are under consideration:
The Bali Irrigation Project (B.I.P.) is the first large scale attempt in Bali island to improve the
irrigation systems. Past interventions by the Department of Public Works have been limited to
isolated improvements, with negligible external consequences. In contrast, the B.I.P. will inter-
vene in 130 subaks (about 10 percent of the total Bali subaks), many sharing the water from the
same river. The impact of the main improvements will concern:
-River water sharing and Subak coordination;
-New 0 & M rules;
-Programmed cropping patterns;
-Use of measurement systems;
--Changes in cropping techniques;
-Yield monitoring systems;
-Taxes and water charges.
In consequence the Subak may lose some ofits traditional facets, especially part ofits autonomy.
Feasibility Study, Part Two (1981)
Bali Irrigation Project
Some changes have already begun. Ten years ago, in the first flush of the “Green Rev-
olution,” Balinese farmers were instructed to ignore the temple-scheduling system, and
plant as often as possible so as to increase yields. Religious ceremonies continued in the
temples, but the system of region-wide fallow periods broke down. As a consequence, the
incidence of bacterial and viral diseases, together with insect and rat populations, began
to increase rapidly. Imported organochloride pesticides made some dents in the rising
pest populations, but also killed off eels, fish, and in some cases farmers in the rice fields5
By the early 198Os, most subaks were hotly debating a return to the regional water-temple
scheduling system in order to control pests, and reduce the need for pesticides6
Today, two irrigation-management institutions coexist on the island, institutions so
fundamentally dissimilar that they are all but invisible to each other. Downstream, for-
eign consultants dispatch airplanes to photograph Bali’s rivers from above, and draw
topographic maps of new irrigation systems. Upstream, a group of farmers drop frangi-
pani flowers in their canals before beginning a new ploughing. The new subak prepares
for the dedication of its Ulun Swi temple, two subaks arrive at the master water temple
for advice on dealing with the brown plant-hoppers which have destroyed half their crop,
and half a dozen men with picks and shovels shore up the sides of a field that has produced
two crops of rice each year for the past eight centuries.

Notes
Acknowlcdgmmts.This research was supported by a grant from the National Science Foundation,
XBNS-8210124. I wish to thank Gusti Ngurah Bagus, Harold Conklin, Janet Hoskins, Robert
Hunt, Kristina Melcher, Alexander Moore, Gary Seaman, and especially Therese de Vet.
’For descriptions of the subak system, see Birkelbach (1973); Geertz (1972, 1980); Grader
(1960[1938]); Korn (1932); Wirz (1927).
Various writers (e.g., Covarrubias 1937284) have described the festival of Galungan, which
occurs on the fourth day of the 11 th week (Dungulan) of the uku calendar as a “New Year’s Day,”
but in fact it is not. The New Year actually begins on the first day of the tenth month, eg., the day
after the new moon of Kasanga, following the new moon closest to the spring equinox.
340 AMERICAN
ANTHROPOLOGIST [89, 1987

3The film, The Temple of th Crater Lake, is available from Documentary Educational Resources,
5 Bridge St., Watertown, MA 02172.
4C.J. Grader briefly mentions the creation oftwo new subaks in North Bali in 1914 and 1931, but
describes only matters relating to taxation (Grader 1960[1938]:284-285).
5Hospital officials at R.S.U.P. Denpasar report several cases of apparent fatalities due to pesti-
cide poisoning.
6Three factors seem particularly salient: the fact that pesticides destroy the natural enemies of
pests, as well as the pests themselves; the appearance ofnew resistant biotypes ofpests such as the
brown plant-hopper and tungro virus; and the environmental degradation caused by pesticides.
Altogether, it would appear that there is a strong economic/ecological case to be made for reducing
pesticide use in favor of the tried-and-true Balinese water temple system of region-wide fallow pe-
riods (see Conway 1983 and Oka 1979).

References Cited
Avinieri, Shlomo, ed.
1969 Karl Marx on Colonialism and Modernization. New York: Anchor Books.
Bali Irrigation Project
1981 Bali Irrigation Project Feasibility Study. Jakarta: Ministry of Public Works, Directorate
General of Water Resources Development, Republic of Indonesia.
Birkelbach, Aubrey
1973 The Subak Association, Indonesia 16:153-169.
Conway, Gordon R., and D. S. McCauley
1983 Intensifying Tropical Agriculture: The Indonesian Experience. Nature 302:288-289.
Covarrubias, Miguel
1937 Island of Bali. New York: Alfred Knopf.
Coward, E. Walter, ed.
1980 Irrigation and Agricultural Development in Asia. Ithaca, NY: Cornell University Press.
DeDatta, Surajit K.
1981 Principles and Practices of Rice Production. New York: John Wiley & Sons.
Foucault, Michel
1980 Michel Foucault: Power/Knowledge, Selected Interviews and Other Writings. Colin
Gordon, ed. New York: Pantheon Books.
Geertz, Clifford
1972 The Wet and the Dry: Traditional Irrigation in Bali and Morocco. Human Ecology 1:23-
39.
1980 Negara: The Balinese Theater State in the Nineteenth Century. Princeton, NJ: Princeton
University Press.
Grader, C. J.
1960[1938] The Irrigation System in the Region ofJembrana. In Bali: Life, Thought, Ritual.
J. L. Swellengrebel, ed. Pp. 270-288. The Hague: W. van Hoeve.
Happt, P. L. E.
1919 Een Beschouwing over het Zuid-Balische Soebakwezen en zijn Verwording in Verband
met de Voorgenomen Vorming van Waterschappen in N.I. Indische Gids 41.183-200.
Hobart, Mark
1982 Padi, Puns and the Attribution of Responsibility. In Natural Symbols in Southeast Asia.
G. B. Milner, ed. Pp. 5 5 8 8 . London: School of Oriental and African Studies.
Hunt, Robert
1984 Measurement of Organization of Authority and Scale in Canal Irrigation Associations.
Unpublished ms.
Hunt, Robert, and Eva Hunt
1976 Canal Irrigation and Local Social Organization. Current Anthropology 17:38!4-411.
Korn, V. E.
1932 Het Adatrecht van Bali. 2nd edition. The Hague: G. Naeff.
Lansing, J. S.
1983 The Three Worlds of Bali. New York: Praeger.
Millon, Rene
1962 Variations in Social Responses to the Practice of Irrigation Culture. In Civilizations in
Desert Lands. Anthropology Paper #62. R. Woodbury, ed. Pp. 56-88. Salt Lake City: Uni-
versity of Utah.
Lansing] BALINESE
“WATERTEMPLES” 34 1

Oka, Ida Nyoman

1979 Cultural Control of the Brown Planthopper. In Brown Planthopper: Threat to Rice Pro-
duction in Asia. Pp. 357-369. Manila: The International Rice Research Institute.
van der Heijden, A. J.
1924-25 Het Waterschaps Wezen in het Voorrnalige Zuid-Balische Rijks. . . . Koloniale Stu-
dien 8:266-275; 9:425-438.
Wirz, Paul
1927 Der Reisbau und die Reisbaukulte auf Bali und Lornbok. Tijdschrift voor Indische
Taal-, Land-, en Volkenkunde (Batavia) 67:217-345.
Wittfogel, Karl A.
1957 Oriental Despotism: A Comparative Study of Total Power. New Haven, CT: Yale Uni-
versity Press.
Yoshida, Shouichi
1981 Fundamentals of Rice Crop Science. Manila: The International Rice Research Institute.