Sorcery, Premeditated Murder, and the Canalization of Aggression in Sri Lanka

Gananath Obeyesekere, being an academic article [1] published initially in 1975 in the American Ethnologist, Vol. 14,  pp.1-23 …. and is also available in pamphlet form as No. 11 in the Studies in Society and Culture Series organised by the late Ananda Chittambalam and Michael Roberts … ISBN 955-9195-10-7 … while the illustrative snaps are impositions by the Editor, Thuppahi  and are intended to suggest the atmosphere called into being by the supplicants seeking support and/rt vengeance at the shrines

It is my intention in this paper to deal with a series of interrelated problems. First, we will consider the following specific questions and propositions: (a) How far can we make inferences about personality and social structure from official statistics computed by government agencies, in this case statistics on homicide and crimes of violence ?

Criminology as a discipline is especially concerned with this problem, and recent criminological studies in Sri Lanka have made social structural, cultural, and personality inferences from the statistical data (Wood 1961; Bloch 1960; Jayewardene and Ranasinghe 1963). At the outset, let me emphasize that I am not concerned with the conventional debate about the accuracy of governmental statistics. I agree with the criminologists who have worked on this problem that Sri Lanka’s official statistics on homicide and violent crimes are reasonably accurate, and on the face of it there is perfect justification for using these data for social analysis (see Wood 1961).

(b) Homicide and violent crimes are the expression of the hostile impulses and actions of people, and as such they pertain to the larger problems of how aggression is handled, controlled, and canalized in any human society. We cannot fully understand the problem of crimes of violence unless we analyze the more fundamental problem of the expression of aggression in society. Social scientists who have studied crime in Western societies will at once object, and state with some justification that violent crime is a special type of aggressive activity, radically different from, let us say, football or boxing, and one which merits analysis independently of other forms of aggressive activity. However, the position changes quite remarkably when we examine traditional and contemporary non-Western societies. It can be easily demonstrated that in many such societies there are modes of killing or hurting people other than by use of lethal weapons or poison-by those magical means known as sorcery and witchcraft. We are concerned with sorcery only, and we define it as a “technique of killing or harming someone, deliberately and intentionally, generally with homeopathic or contagious magic, accompanied by spells, charms, or incantations.” Thus sorcery can be compared with homicide and crimes of violence, if not in its actual effect, at least in its motive and intended effect.

(c) Sorcery is a deliberate and intentional act, for the person who consults a sorcerer does so after considerable planning and premeditation. In this sense sorcery is calculated; one might say it is a “rational crime,” using rational in the sense of “a careful consideration of means for achieving a given end.” The end in this case being murder or personal harm to a “victim.” The rational nature of sorcery is of crucial importance for the present paper, since it stands in clear contrast to the dominant pattern of the homicide and “grave crime” rate computed officially, and to the questionnaire data of the criminologists.

Criminologists who have worked in Sri Lanka unequivocally postulate that spontaneous and unpremeditated crime totally outweighs premeditated homicide and violence. Wood (1961:71) shows clearly that only 20 per cent of all cases of homicide are premeditated, 51 per cent of the cases are spontaneous but with previous enmity or dispute, and 29 per cent are spontaneous without previous enmity and dispute. Bloch (1960: 174) substantially agrees with him. A spontaneous homicidal act is one in which the time gap between stimulus and response is narrow, giving the individual little time to weigh rationally the implications of his action. In Sri Lanka, spontaneous homicide and violence may erupt under trivial provocation and result in brutal and instantaneous response (Bloch 1960: 485-486), which is probably true for many parts of the world.

In this paper we will show that the low frequency of premeditation in homicide and violence in the official record does not imply that the motive for premeditated crime is infrequent nor that its expression is infrequent. It only implies that the motives for such acts are canalized through functional alternatives. In Sri Lanka, as in many parts of the non-Western world, an alternative is the practice of sorcery.

If this is indeed the case, the official statistics show only one side of the picture. Crime in Sri Lanka is presented as spontaneous, not as rational and calculated, as it sometimes can be according to the evidence we shall give later. Furthermore, insofar as such data are one-sided and biased, the inferences made from the data on personality and social structure must also perforce be biased and one-sided, such as Bloch’s (1960: 42, 486) interpretation of spontaneous crime as a function of low frustration tolerance or a loose social structure.


The major problem of this paper leads us to consider the possibility of a functional alternative, where the same function may be served by substantively different structures. Thus the same set of motives may be canalized and expressed through different customs, and the same custom may be a vehicle for the expression of a different set of motives. This has serious implications for correlational analysis, in that totally misleading results may emerge if there is no control over the functional alternative (if and when it exists) of any custom or trait that is used as a variable. The problem is confounded further if one custom or set of practices is not only a functional alternative to another but also its structural alternative, i.e., if the two customs are isomorphic. This implies that the data gathered with respect to one custom or practice would be quite misleading if the data relevant to the alternative are ignored, since both types of data deal with substantively the same phenomenon. Undoubtedly it would be unusual to find a single custom or practice serving as both a structural and functional alternative to another, i.e., a nearly complete isomorph; but I submit that sorcery and premeditated murder and violent crime are structurally isomorphic and functionally similar in certain important dimensions or parameters. Let me specify these dimensions:

(a) The intention of the actor. In premeditated homicide and violence as well as in sorcery, the intention of the actor is the same, which is to kill or cause personal harm to the victim. In Sri Lanka, people can explicitly verbalize their hostile intentions, so that when people go to some of the more powerful sorcerers, the homicidal intention is clear. They want their enemies killed.

(b) The rational nature of the act. In both sorcery and premeditated murder there is a real time gap between stimulus and response. During the interim period the offender makes rational decisions regarding the kind of sorcery he should practice, the sorcerer he should consult, and the time, effort, expense, and risk involved. The sorcerer is, in effect, the equivalent of a hired killer.

(c) The traditional legal definition. Traditionally, both sorcery and murder were viewed as crimes and carried the identical punishment, which provides further evidence that the indigenous culture viewed these two practices as similar. Pieris (1956: 144) noted that in the time of the Kandyan kings (before 1815), “several persons were executed and their lands given to the injured party” for practicing sorcery.

(d) The techniques employed. Here we have the first noticeable substantive difference. In both the techniques are viewed as pragmatic and instrumental, but in one “poison, arson, or the knife” are employed, while the other uses magical practices which vary from one type of sorcery to another. In evaluating the difference between the two techniques employed, one must not make the mistake of seeing magical practices in sorcery from the Western intellectualist viewpoint as symbolic and expressive actions. Sinhalese persons are aware of the substantive differences between physical weapons and magical weapons, but the employment of both are viewed as pragmatic and instrumental techniques radically different from the expressive forms of worship in, let us say, a Buddhist temple. Moreover, they are aware that sorcery is not all of a piece, and that some forms of sorcery are more powerful than others. I am convinced that many Sinhalese persons would view the sorcery practiced by some of the well-known vas-kavi sorcerers (see below) as effective and deadly as hiring a paid assassin, or even more so. Also, the risks involved in premeditated murder or violence against the person must be weighed against the relative safety of sorcery: in modern law sorcery is no longer viewed as a crime, it is secret, and it is less messy.

The one truly significant difference between sorcery and the premeditated crime is in its relation to the consequence, for we know and can demonstrate that in fact people rarely die as a consequence of sorcery. But people believe that it is effective, and they do practice sorcery rather than commit a direct act of physical violence. In other words, the motive for premeditated murder can be channelled away from physical violence into sorcery. If so, the present data on crime statistics with their high ratio of spontaneous crimes to premeditated ones are highly biased and could be corrected by taking into account the factor of sorcery.

In order to do this, one has to have statistical data on the incidence of the practice of sorcery. I have emphasized the term practice, in order to distinguish it from imputations or accusations of sorcery. In some societies, like Navaho (Kluckhohn 1962), sorcery accusations or imputations are universal, but there is no real evidence regarding its actual practice. In other societies, like Sri Lanka, India, and the Philippines, people not only accuse others of practicing sorcery, but actually do practice it. Anthropologists have generally analyzed the pattern of sorcery accusations or imputations but not the practice, largely because its practice is generally veiled under a shroud of secrecy and its incidence is difficult to determine. In this paper we shall attempt to estimate the incidence of sorcery practice.

We encountered several initial difficulties. The most deadly form of sorcery practice in Sri Lanka is known as vas-kavi (vas meaning “evil,” “ritual danger,” “poison”; kavi meaning “songs”). Vas-kavi are songs constructed on the basis of highly elaborate astrological and mystical calculations; each song is designed for the specific needs of the client. These songs are then sung a prescribed number of times before the image of some dominant deity like the god Skanda (Kataragama). Ironically, many vas-kavi specialists are Buddhist monks, some of them having great reputations as sorcerers.[2] For example, it is believed that the moment a well-known vas-kavi specialist performs the prescribed ritual and steps out of the shrine the victim falls dead! Vas-kavi is very popular among politicians as a technique of getting rid of political enemies. The fees charged by the better known vaskavi sorcerers are also high, and only the affluent can generally afford their services. None of the monks who practiced as vas-kavi sorcerers were willing to talk about their clientele, and we could not get any useful information on the incidence of sorcery practice from this source.

Next to vas-kavi are the cruder forms of sorcery known as hūniyan. These are consistent with practices recorded in the cross-cultural literature and abound with homeopathic and contagious magic, e.g., pricking thorns into the wax image of the victim, burying charmed objects in the garden of the victim, and using the hair and other exuviae of the victim. Hūniyan is generally performed by ritual specialists known as kattaḍirāla, whose major ritual role, however, is the propitiation of demons who have afflicted individuals. Most of the kattadirālas we knew were performers of benevolent curative magic, but occasionally some of them would perform hūniyan in order to please an influential client or earn a large fee. The data we obtained from this source, although interesting, were inadequate for our purposes.

The only firm conclusion we could draw from our knowledge of vas-kavi and hūniyan specialists was that any person with enough money to pay the sorcerer could, in this culture, readily use vas-kavi or hūniyan with the intention of causing death or serious harm to an individual or group of individuals. However, it remains a conclusion that cannot be supported by statistical data.

A third type of sorcery is associated with shrines presided over by a powerful deity or deities. The client consults the ritual specialist of the shrine who enlists the aid of the deity in order to cause death or harm to the victim. These ritual centers were excellently suited for our purposes for the following reasons:

(a) These shrines were public places which did not conduct their rituals in secret.

(b) The fees charged by the specialists varied from one shrine to another and in relation to the nature of the client’s problem, but in general they were within the means of an average peasant.[3]

(c) Large numbers of clients visited the shrine, so that it was possible to get a sample large enough for our purpose.

(d) Many of these shrines were so famous nationally or regionally as centers for the practice of sorcery that they drew clients from various parts of the country or a large region. Therefore, they lacked the intimate character of village temples or shrines. The clients who visited the shrine did not know one another and could maintain anonymity.

The maintenance of anonymity is very important for the clients visiting these public sorcery shrines. The major technique for maintaining anonymity is through geographic distance. The incidence of its use is presented in Table I.

Sorcery must be practiced in secret; yet the sorcery shrines are public places. Clients can best ensure that they will not meet people known to them at the sorcery shrine by going to a distant one. Hence people who live in the vicinity of a shrine rarely visit it. In Table I only the cases in the first column risk being found out; the rest (80 per cent of the cases) are fairly certain of maintaining their anonymity at the sorcery shrines. Those distances are especially significant when we realize that Sri Lanka is a small country of only 24,000 square miles. The Muslim shrine has the largest proportion of persons within a five mile radius. This shrine is located in an almost exclusively Muslim town, hence Buddhists and Hindus from nearby villages could come there without much risk of recognition.


My original intention was to work with vas-kavi sorcerers, but for reasons mentioned earlier we had to work at the sorcery shrines instead. We picked three shrines, each representing one of the important religions of the country-Buddhist, Hindu, and Muslim.

The Buddhist shrine is located at Sinigama in the southwest coast of Sri Lanka, about 50 miles from Colombo. The presiding deity of the shrine is Devol Deviyo, viewed by the Buddhists of the area as a major deity in their pantheon. The Hindu shrine is located at Munneswaram in the northwest coast of Sri Lanka, 55 miles from Colombo. Munneswaram is one of the most famous of Hindu temples in Sri Lanka, sacred to Lord Siva. However, the practice of sorcery, which often involves the slaughter of a goat or chicken, takes place in the Kali temple located outside the main temple, and the ritual specialists are non-Brahmin. The scale of the sacrifice was consonant with the intention of the client-a chicken indicated physical harm to the victim (accident, illness), whereas a goat indicated death. Persons who were willing to take no chances sacrificed several goats to the deity. (These animals were subsequently sold to the local restaurants by the priests). The Muslim shrine is part of a mosque at Kahatapitiya, six miles from the University of Sri Lanka campus, on the Peradeniya-Gampola highway. The rituals are performed in a room where the body of a local Muslin saint is buried.

The shrines are not exclusive to the particular religions, but are open to all religions. In fact the largest number of clients in both the Muslim and Hindu shrines were Buddhist, demographically the largest segment of the country’s population. This again indicates that sorcery is viewed as an instrumental rather than expressive type of activity by the clients who patronize these shrines.

Table I Number of Clients Living Within Specified Distances from Sorcery Shrines Per Cent 100 100 100 100
Number 215 438 128 781
50 Miles and Over Per Cent 33.4 8.4 59.3 23.7
Number 72 37 76 185
10-50 Miles Per Cent 59.0 63.9 25.7 56.3
Number 127 280 33 440
5-10 Miles Per Cent 3.7 15.2 7.0 10.8
Number 8 67 9 84
5 Mile Radius Per Cent 3.7 12.3 7.8 9.2
Number 8 54 10 72
Hindu Shrine Muslim Shrine Buddhist Shrine Total


The research was designed in two stages. The initial stage was the collection of data from the three shrines: these data would be simple, lacking in the depth characteristic of case studies of individuals. Our intention in the second stage was to refine some of the conclusions that emerged from the first study, subject these hypotheses to a more rigorous test, and also get selected case studies in depth. For this purpose we picked a sorcery shrine in the Morawak Korale in the southern Province. This shrine was atop a huge mountain in a heavily forested region, and we felt that anyone who could undertake this backbreaking hike must be singularly steadfast in his intentions. However, this study was abandoned as the area was overrun by insurgents in the April 1971 insurrection.

The research design was extremely simple. Research assistants were asked to collect information for a period of 60 days in each shrine. Two research assistants were placed in Munneswaram, as this, we felt, was the most important shrine; one research assistant was placed in each of the other shrines. Since daily rituals were performed in both Munneswaram and Kahatapitiya Mosque, we had the necessary information collected, with minor breaks, within a period of slightly over two months. The Buddhist shrine had a different routine. In general the public belief was that Wednesdays and Saturdays were the special (kemmura) days for rituals for the gods; but the priest said that this applies for benevolent magic and propitiatory ritual only. His view was that sorcery rituals could be practiced every day. The upshot of this confusion was that most clients visited the shrine on Wednesdays and Saturdays, although a few came on other days, while on some days there were no clients. Thus, at the Buddhist shrine, 60 days of field work extended to a four-month period.

Owing to the sensitive nature of the subject matter, the research assistants were instructed to collect extremely simple information on the following:

(a) purpose of visit

(b) who was responsible for the harm done to the client (for provoking him to sorcery)

(c) was any action taken by the client before coming to the shrine

(d) was it the client’s first visit to the shrine

(e) number of persons in client’s group, their sex, and approximate ages

(f) wherever possible, information on race, religion, and civil status of client.

In fact our research assistant did not have to interview the clients directly for most of the above information in either Munneswaram or Sinigama, since the priests themselves in these temples elicit this routinely; so in these two places the interviewing was done for us by the priests! In fact some priests reframed their conventional questions so as to get unequivocal answers for our benefit. Thus the basic information was collected in as naturalistic a manner as possible without upsetting the social situation at the shrine premises. We had instructed our assistants to gather further sociological information from the clients after they had finished their rituals (puja) in the shrines. This was extremely difficult in Munneswaram owing to the large numbers that came here, but considerable information was nevertheless obtained. At Sinigama the numbers were smaller, the priest was most co-operative, and we were able to get extra sociological information which added depth to our study and permitted the testing of many hypotheses not anticipated at the time the research was originally formulated. The situation at the mosque was slightly different. Contrary to our expectations, the mosque had an extremely large number of clients. Furthermore, the priest did not ask the basic questions and our research assistant, posing as a mosque official, had to get the information himself. Thus it was virtually impossible to get any detailed information, except the very basic material. To sum up, the most detailed sociological information came from Sinigama, then Munneswaram, and lastly Kahatapitiya Mosque. However, in all three places we had sufficient information on the basic questions mentioned above.


While the crime statistics reveal that 72 per cent of the homicides and crimes of violence are spontaneous and 20 per cent premeditated, we hypothesize that a radically different pattern will be revealed when the data on sorcery are taken into account (see Table 2).

Table 2 shows a total of 803 cases of interviewed clients who come to the three shrines for sorcery during the 60 days of field work in each shrine, with an estimated yearly total of 7,441. The significance of these figures comes out dramatically when we compare the total of four categories of violent crimes for Sri Lanka for the year 1957 (see Wood 1961: 54), to use a convenient date. Wood (1961: 54) shows that 6,477 violent crimes (homicide, attempted homicide, grievous hurt, hurt by knife) occurred in that year; if we add arson (566) and riot (89) we have a total of 7,023. Thus the yearly figure for sorcery for just three shrines is more than the total figure for all violent crime in the year 1957! If we view sorcery practice as the equivalent of premediated violence, then this category of behaviour is more common than spontaneous violence. Such a view of human behaviour seems to us to be perfectly justified. In every human society people are taught to control two of man’s most imperious drives-sex and aggression. It is true that the degree of control and the nature of the inhibitory mechanisms vary from one society to another. Yet it seems to us that in every society there are those who, when confronted with provocative stimuli, act impulsively and aggressively, while others have learned to control or repress these impulses and bring them under the control of rational judgement. It can be presumed that in total the latter outnumber the former. Viewed in this light, the problem of premeditation in the perpetration of crime must be placed in a larger perspective than criminologists have hitherto done.

Table 2 Total Number of Clients Visiting Three Sorcery Shrines for a 60 Day Period, and an Estimated Yearly Total Estimated Yearly Total of Sorcery Cases 4,150 2,952 339 7,441
No. of Clients Who Came for other Purposes 80 97 15 192
Percentage of

Clients Who

Came for Sorcery

76 82 88 81
No. of Interviewed

Clients Who Came

for Sorcery

248 442 113 803
No. of Clients Interviewed 328 539 128 995
Total No.

of Clients

Visiting Shrine

915 600


128 1,643
Name of Shrine and Duration of Work Hindu Shrine

(two-month period)


Muslim Shrine

(two-month period)


Buddhist Shrine

(four-month period)



The problem of viewing crime as premeditated versus spontaneous clouds the larger issue. Namely, when people are confronted with stimuli that may provoke them to a homicidal or violent response, the following reactions probably take place: (i) the largest number of people repress or control the desire or drive for violence, and having controlled the drive, bring the whole problem under the control of the rational mind and the moral norms that govern their conscience, (2) a very small proportion of people confronted with identical stimuli act in a somewhat different, yet “rational” manner. They repress or control the impulse to act out immediately with violence, yet they may plan violence or murder in a manner that will bring about the desired end with the maximum efficiency and expediency, (3) a larger proportion of people act immediately and violently. This group would be very small when compared with the first, but large in comparison with the second. To sum up, many people when confronted with provocative stimuli are tempted to commit murder or violence against another, yet most people repress these wishes and bring them under the control of reason; while a small number commit what is known as premeditated murder, and a larger group practice spontaneous murder and violence.

These considerations bring us to another feature of the sorcery data that are presented in Table 2, that is, the yearly total for three shrines is a mere fraction of all the sorcery actually practiced in the country. This is obvious, as we have ignored the hundreds of specialists who practice the most deadly kind of sorcery, vas-kavi; the thousands of kattadirāla (priests of the demon cults) who practice hūniyan type sorcery at least occasionally; and furthermore probably over a dozen national and regional shrines similar to the ones we studied. It should also be noted that while the priests of the village devales (temples for gods) generally deal with benevolent magic, they also on occasion deal with sorcery. Insofar as there are thousands of such temples, this will also add to the picture of the incidence of sorcery we have presented. Thus we are confronted with a curious problem. If sorcery on the motivational level is in some sense an equivalent of premeditated crime, then this category totally outweighs the total number of cases of spontaneous crime, and the statistical relationship between premeditated and spontaneous crime noted by criminologists is reversed. But this again is implausible for, according to criminologists (and here we agree with them), the incidence of spontaneous violence in any country is larger than that of premeditated homicide and violence. We advance the following proposition to account for this disproportion: if any society provides culturally constituted mechanisims for the canalization of the aggressive impulses of individuals, then those individuals who would normally tend to repress these impulses and bring them under rational control would now be tempted to give them expression.

Several objections can be levelled against the preceding arguments. Some likely theoretical objections will be considered at the conclusion of this paper. Here we will consider an important substantive and factual objection which may be raised. In the case of premeditated (or for that matter, spontaneous) crime, the target of the attack is a known identifiable enemy. Is it the same for the sorcery cases collected from the three shrines? It is possible to argue that only insofar as the enemy could be identified could the two types of data be considered truly isomorphic. Table 3 gives the relevant information from the three shrines. In the Buddhist and Muslim shrines information relating to this problem is available from all the respondents; in the Hindu shrine it is available for go per cent of the total sample. I think it makes no difference psychologically whether one clearly identifies the enemy or guesses his identity (both could be treated in a single category of “enemy known”). Thus in 66 per cent of the cases in all three shrines the enemy is “known” to the client; in 34 per cent of the cases the enemy is unknown.

  Photo by Juliet Coombe

How does this information affect the two hypotheses stated earlier?

(a) At worst the number of sorcery cases presented in Table I must be reduced by 34 per cent so that we have a yearly total of only 4,911. This also means that the number of cases of sorcery is less than the yearly crime rate for I957, which in any case was an arbitrary baseline. However, our basic hypothe-

TABLE 3: Identity of the Enemy

 Percentages of the Cases for which Information is Available

N Per Cent Clearly Identified Per Cent Identity Guessed Per Cent



Munneswaram 223 55 16 29 100
Kahatapitiya 442 49 8 43 100
Sinigama 113 40 29 31 100
Total 778 48 18 34 100


sis is not affected, for there is nevertheless an extraordinarily large number of sorcery cases, and a radically different pattern still emerges even when those data reduced by 34 per cent are considered. Insofar as there are many other sorcery shrines and types of sorcery, this form of “rational crime” would totally and completely outshadow spontaneous crime.

(b) On the other hand, I would like to argue that the 34 per cent who cannot identify their enemy are relevant for the present study, for an important reason. I shall demonstrate later that a majority of these clients are persons not given to spontaneous emotional responses. Thus I believe that most of these clients simply do not proceed to find out who was responsible for causing them harm, for such a quest might eventually lead them into a personal confrontation which they are not psychologically equipped to handle. Thus potentially they are capable of identifying their enemies, and therefore there is no categorical distinction between them and the others in our sorcery sample.


According to Bloch (I960), Wood (1961), and Jayewardene (1963), disputes over land and sex are the main provocations underlying homicide; disputes over other types of property come a close third. Our analyses of the disputes underlying sorcery (using the Sinigama data) show that the overwhelmingly large reason for practicing sorcery is dispute over property, generally theft or damage done to property. For example, in Sinigama nine cases were scored for land dispute and nine for sex dispute, while 86 were scored for property disputes. According to Wood, sex and land disputes constitute respectively 17 and 16 per cent of the immediate causes of homicide (Wood 1961:71). One methodological reason for the disparity is obvious: theirs is a national sample, whereas ours deals with three shrines, and can give no accurate information on the nature of sorcery practice in Sri Lanka in general. However, a substantive reason for the difference, as far as theSinigama data are concerned, is that the majority of people who visit the shrine are not peasant farmers, but rather those who practice non-traditional occupations. It seems that, in our sample at least, those who practice sorcery tend to come from the more urbanized, probably more educated, and more enterprising segments of village society. The aspect of our data given in Table 4 has very interesting and important implications.


Occupations of Clients Visiting the Sinigama Shrine

1. Farmer 22 (17.0 per cent)
2. Other traditional occupation 3 ( 2.3 per cent)
3. Landed proprietor 16 (12.4 per cent)
4. Shopkeeper 12 ( 9.3 per cent)
5. Other non-traditional occupations 40 (31 .0 per cent)
6. No information 36 (27.9 per cent)
Total 129 (99.9 per cent)

Number 1 is the strictly traditional occupation of peasant farmers. Number 2 has two traditional occupations of Buddhist monk and devil dancer. Number 3 needs some explanation. These are landowners who though living in the village have emancipated themselves from the traditional occupation of peasant farmers, and have cultivated cash crops like tea, rubber, and coconut. They are generally wealthy and educated and do not consider themselves to be peasants. Number 4, consisting of shopkeepers, are obviously non-traditional. To get the full flavor of the non-traditional occupations in Number 5, it is well worth listing some of the occupations involved: mason, garage owner, automobile mechanic, proprietor of handloom shop, rice mill owner, beedi manufacturer, seamstress, taxi driver, village committee member, estate laborer, teacher, printing press owner, hospital attendant, government contractor, car salesman, pawnbroker, port worker, university graduate, tailor, salesman, storekeeper, typist, railway ticket collector, contractor, typewriter mechanic, and government clerk.

Thus one reason for the high incidence of disputes pertaining to property, other than land, is that the sorcery sample does not reflect the true occupational composition of the country, which is peasant farming, but tends to be biased in the “urban” direction. Even the category of people who are classed as landed proprietors rarely come to Sinigama because of land disputes, for it can be assumed that these wealthy people owning substantial areas of land would not have the problem of peasants, which in general is a problem of land shortage. Rather, landed proprietors practice sorcery because an enemy has destroyed their cash crops or stolen cattle or some other kind of property.

The data on social mobility permitted us to inquire whether our assumption that sorcery practice showed a bias in the “urban” or “middle class” direction was correct. Thirteen clients stated that they were outsiders in the village or town in which they resided, and as such were harassed by its citizens. This information was given voluntarily during the interview, and it is likely that more cases would have fallen into this category if we had had a specific question on the subject in our schedule. What is significant, however, is that all (except one for whom no information is available) who considered themselves as outsiders in the village come from the non-traditional occupations. Furthermore, the data indicated that these outsiders were confined to those groups-shopkeepers and other non-traditional occupations -that represent the most “urban” of the groups in our sample. Significantly, no one in the landed proprietor group stated they were outsiders, for as the author has shown elsewhere, typical landed proprietors are village residents who have made good through several generations and represent a village aristocracy and intelligentsia (Obeyesekere 1967:211-247).

That many of the clients are outsiders is also indirectly confirmed by the data presented in Table 5, which lists the clients’ enemies (victims) in terms of kin/nonkin distinction.

If 73 per cent of the enemies are nonkin, it is also likely that a fair proportion of the clients would also be outsiders to the village. Furthermore, the data clearly show that we are not dealing with traditional villages structured on a homogeneous kinship base, but rather with changed and highly acculturated communities.

TABLE 5:  Identity of Victims of Sorcery in Terms of Kin/Nonkin Criteria

Name of Shrine Per Cent


Per Cent


Hindu Shrine 20 80
Muslim Shrine 28 72
Buddhist Shrine 33 67
Total 27 73


The fact that the sorcery sample is more “urban” than rural has significant implications. The term urban is somewhat misleading and I must explain what I mean by it. Clients in general come from villages or from the small townships adjacent to a group of villages. But occupationally they constitute a new class in the village social structure, in general analogous to “lower middle class” in Western stratification terminology. Although not a city population, they constitute a highly acculturated segment of village society. What is at issue here is not the fact of living in an urban area but rather that Numbers 3, 4 and 5 in Table 4 reflect in general a class of people who are gainfully employed, economically well off, hardworking, and perhaps thrifty-the good middle class of the village or small town. Such people are generally law-abiding citizens. Sri Lanka’s official crime statistics also prove that homicide and attempted homicide are typically a village (rural) problem and that the rate is lowest in the metropolitan Colombo region (Bloch 1960, Wood 1961). It is true that the Colombo region has a-high rate for violence other than homicide but this in our opinion reflects a tendency of the urban proletariat, rather than the middle class. Our data clearly show that the motives for homicide and attempted homicide exist in the urbanized sections of Sri Lanka’s population, but these motives are canalized and expressed through the mechanism of sorcery. Since this is the case, it should again be realized that danger exists in drawing personality and social structural inferences from the official census data as Bloch and Wood have done. Our data furthermore show that middle class or occupationally “urban” people in Ceylon tend not to commit spontaneous homicide or violence, but rather tend to commit sorcery, which is a form of rational crime.


The foregoing analyses of sorcery data further suggest the validity of the following conclusion: the kind of people who practice sorcery, whether urban or peasant, have considerable inner control and are not given to spontaneous violence when confronted with provocative stimuli. This conclusion is borne out very clearly in the data for the three shrines. If our view is correct, the data we have should enable us to test a further hypothesis: the personalities of individuals practicing sorcery are such that when confronted with provocative stimuli they tend to control their impulses rather than resort to personal retaliation. The hypothesis is clearly confirmed by the data categorized in Table 6. In this table we have, for obvious reasons, eliminated cases where the enemy is not known to the client, or where the information was uncertain.

The data presented in Table 6 are very impressive. In only 10 per cent of the cases did the client confront his enemy personally, even though he knew who his enemy was, or believed he did. Moreover, in 6 per cent of the cases there was merely a face-to-face meeting in order to “persuade,” or to “explain.” In the Sinigama sample, where we have detailed case studies, these attempts at confronting the enemy were not always advantageous to the client, for in five instances the enemy in fact retaliated with physical or verbal abuse against the client. Only 4 per cent of the sample were truly aggressive: they not only confronted the enemy but also retaliated with verbal or physical abuse (mostly verbal, there being only two cases of physical retaliation in the whole sample). We feel that we are truly dealing with a type of individual who has the capacity to control his impulses and act with considerable restraint. This is further borne out in the Sinigama data. Here we have over a dozen cases where the clients have been pushed against a wall by their enemies through continual physical, verbal, and other forms of intimidation. Yet they have patiently borne the ignominy and have not immediately retaliated in kind; rather they have calculated their moves and postponed their retaliation through supernatural and magical means.

We are also impressed by another aspect of client behavior. A considerable number of people had taken some form of official action, such as informing the headman or police, or seeking court redress: 31 per cent at Munneswaram, 22 per cent at Kahatapitiya and 33 per cent at Sinigama. This is even more impressive since practically everyone who expressed opinions on the subject spoke of the futility of reporting to the police. Respondents were unanimous in their view that police officers were corrupt and under the control of “undesirables” and politicians. In any case, villagers are extremely afraid of the police, both for rational and non-rational reasons. In general we believe that villagers were quite justified in disassociating themselves from the police, for it is unlikely that the police would give them a sympathetic ear. At least one thing is clear, (whether the village views of the police are true or false) the very fact that people define the police as corrupt, brutal, and unreliable prevents more of them  from seeking police help. The fact that in spite of these views 29 per cent sought police and other official help suggests that we are dealing with individuals who seek rational solutions to their problems.

Sorcery and Justice

In almost all of our cases the client practiced sorcery as a retaliation to an actual wrong done to him by someone else. An examination of the interview schedules confirms, without any doubt, that this is simply not the kind of rationalization phenomenon one encounters in murderers. It is a well known fact that those who commit homicide tend to justify the rightness of their action. Our sorcery clients too felt that they were the victims of injustice and that their retaliatory sorcery action was justified. However, these subjectively felt justifications for their actions, we feel, had an objective validity in most instances; for example: a spouse had been seduced by another, property had been stolen or wantonly damaged, or a person had been harassed or bullied. In the few cases where the client practiced sorcery because someone else had practiced sorcery on him, the objective reality of the latter incident may be false. But at least the client subjectively felt he had been the victim of an unjust sorcery attack that demanded just retaliation in the same terms.

Furthermore, there were no mechanisms for resolving these conflicts on the local level. The traditional authority structure in the changing villages and small towns of Sri Lanka have almost totally disappeared, so that there are no institutional mechanisms for resolving conflicts and redressing wrongs done to a person. Institutions like the caste panchayats of India do not exist here, and the kinship groups, which traditionally may have helped resolve some of these conflicts, are rift by internal tension, largely as a result of land and inheritance disputes (see Obeyesekere 1967 for a discussion of details). Up to 1963, the villages had a headman, generally recruited from the local aristocracy, and these headmen sometimes acted as arbitrators in village disputes. But the headman system was abolished in 1963 and the new headman system (gramasevaka), where recruitment is on the basis of competitive examinations, was introduced. The loss of the older headman was the final blow to the traditional institutional structure. The new headmen were simply bureaucratic officers on the village level incapable of arbitrating or resolving local conflicts. The government, realizing

Table 6 Confrontation of the Enemy by the Client Total Number

of Cases

158 252 78 488
No  Confrontation 145 (91.7 per cent) 234 (92.8 per cent) 60 (77.0 per cent) 439 (90 per cent)
Physical or

Verbal Abuse

2 (1.2 per cent) 13 (5.1 per cent) 4 (5.1 per cent) 19 (4 per cent)


11 (69 per cent) 5 (1.9 per cent) 14 (18.0 per cent) 30 (6 per cent)
Munneswaran Kahatapitiya Sinigama Total

the seriousness of the problem, has introduced an institution known as a “Conciliation Committee.” However, it is obvious that these official Conciliation Committees have failed, for none of our clients admitted going to these committees to resolve his problems. The old authority structure has broken down, the problems faced by people have multiplied, and there exists no satisfactory alternative structure of authority. The police are distrusted and the courts are too impersonal and expensive for most villagers. Moreover, a new type of person has entered into the life of these changing villages-the “thug” or “local undesirable,” generally engaged in illegal activity, such as distilling illicit brews, peddling dope, and stealing cattle. Twenty-seven persons causing initial offense to our clients were described by the latter in one or more of the following terms: “thug,” “kassippu” (illicit liquor) dealer, cattle thief, or member of a clique of undesirables.

The above considerations lead us to another fascinating aspect of our sorcery data: the problem of meaning. In general, people who came to sorcery centers labor under a sense of felt injustice, and what they expect of the deity is that “justice be done.” If the local institutional structure cannot find means for redressing wrong, then one resorts to supernatural agents to correct wrongs and bring about rightness and justice in the world. Many who come to these shrines to solicit supernatural aid have heard from others that these deities will punish the wrongdoer. Let me quote from Case No. 4 (Sinigama):

The client had come once before over the loss of a cow. His prayers to Devol Deviyo were answered, for the god punished the people whom he had suspected of stealing it. Since then he had had unshakeable faith in the deity more than in the police ….

Some of the cases we have are quite pathetic and moving. I shall quote from Case No. 6 (Sinigama) to illustrate this aspect of the sorcery data.

Albert Singho looked a very worried and tormented man for his age. He was about sixty years old. He said for the last ten years his wife, who was younger than he by twenty-two years, had been carrying on with a man forty years old. Recently Albert performed a tovil ceremony called Puna Kapok to get rid of the bad times he was having. During this ceremony his wife’s paramour had come after liquor and threatened him, and in the ensuing fight his arm was broken by the man. Afterwards he had gone to the hospital and avoided the complications of going to the police by telling the doctor that he slipped and fell and broke his arm. He had done this purposely because he was not quite certain that his enemy would be punished properly by the police in a deserving manner and to his entire satisfaction.

He had therefore come to Sinigama after his arm was cured to get this man punished the way he deserved through Devol Deviyo, who had a reputation for being severe toward evil men, and to put a stop to this disturbance in his marital life as a result of his wife’s continued illicit love affair with this man. All that he wanted from the deity at Sinigama was to mete proper justice to his tormentor and to allow him to live a peaceful and happy life with his wife and children.

The deities at these shrines are arbiters of justice, who punish evil doers and redress wrongs. Indeed, the cultural view of these deities is that they will not fulfill the request of the client if his is an unjust one. The human need for justice and right cannot be resolved on the local level owing to the collapse of traditional authority. It is therefore resolved on the supernatural level by supernatural authority figures whom the human agents supplicate at the sorcery shrine.


The distinction between premeditated and spontaneous crime is only a heuristically useful one, in order to show the relevance or sorcery for understanding an aspect of comparative criminology. If, for example, one is interested in the social factors which cause crime this distinction may not be useful. The upshot of the present study is that the pattern as manifest in the official crime statistics of Asian and African societies is grossly inadequate for purposes of comparison, analysis, and inference, since they reflect (a) a bias towards spontaneous and unplanned attack, and (b) they ignore an important category of behavior-sorcery. If we are correct, then practically all criminological theories, assumptions, and prejudices relating to non-Western peoples have to be drastically revised or reoriented.

One of the theoretical objections that may be raised against the argument presented in this paper is whether the two practices, sorcery and premeditated murder, are indeed truly comparable, since a person is only symbolically killed in the former whereas in the latter actual death or injury occurs. It is indeed true that in the one case there is a corpse and in the other there is not (except when the self-fulfilling prophecy operates and an occasional death may result from sorcery). But the existence of the corpse is not relevant for the analysis, for we are interested in how the motive for premeditated violence is canalized through sorcery, rather than through planned murder. Insofar as the motive can be, and is, expressed through sorcery, it need not be expressed through overt practice of planned violence against the victim. Thus in societies where sorcery is practiced one would predict a relative absence of premeditated crime in the official records.

In sorcery practice two things are very clear. First, in almost every instance the client goes to the sorcery shrine with the explicit intention of killing or causing harm to someone, which in general he views as a just punishment. Second, he obviously believes that the magical techniques he uses are effective, or at least he believes that in some instances they are effective, for otherwise he would not practice them. If the intention is to harm someone, and this intention is put into effect in magical (sorcery) action which is believed to be effective, surely this implies that aggressive and hostile impulses of individuals are being channelled into sorcery. These impulses are characterized by hostility to a known or unknown enemy, but are not those in which the hostile rage is spontaneous, uncontrolled, and demanding immediate rather than delayed action. That sorcery is a mechanism for the canalization of aggressive impulses can hardly be denied; what could be denied is that the drive which would normally be expressed in premeditated murder and violence would now be canalized into sorcery. My answer is again, I submit, logically plausible, though I cannot prove it at this point in my research. A man who commits premeditated murder is thinking rationally, i.e., in terms of criteria of instrumental efficiency; and rationality in this sense involves not only the achievement of the goal, but achievement with minimal risks, costs, and other inconveniences. If cultures like Sri Lanka provide magical alternatives for achieving the goal with less risks involved, then sorcery rather than premeditated crime would be the rational choice for many individuals. It is, however, true that some persons may not believe in the efficacy of sorcery; others may believe that relative certainty in actual calculated murder is preferable to sorcery. These persons choose to ignore the risks and complications involved in actual killing and resort to murder and planned physical violence. But many would adopt what, from the cultural point of view, would be the more rational means of achieving the same goal, and it is this kind of decision-making by large numbers of people that accounts for the relative imbalance in the criminal statistics discussed in this paper. On the other hand, I am not suggesting that all these clients would practice murder if these shrines were not available, since I also affirm that the existence of a mechanism like sorcery motivates many people, not otherwise given to expressing their impulses in this manner, to resort to sorcery.

Another legitimate objection may be raised at this point. It can be argued that while the motive may be the same in the two instances, is the satisfaction of the motive or need, or the reduction of the tension the same? This is a complicated matter and I shall spell out my argument in some detail.

(a) It is important to remember that human beings can often anticipate the consequences of their actions, particularly if these actions are calculated. Thus while it is true that premeditated violence may satisfy the aforesaid motive, the individual may feel that it can produce unpleasant consequences. One tension is reduced but the actor may feel that others may be created anew.

(b) The cultural belief is strong that sorcery is effective; in addition, a personal factor may enhance the cultural belief. The individual who is pushed to the wall by unbearable circumstances wants to believe that it is effective. Thus the personal wish adds strength to the cultural belief in the efficacy of the action. It is the strength of this personal wish, reinforced and validated by the cultural belief, that impels some persons to continue to seek redress from other shrines when one shrine has failed to produce results.

(c) What is important therefore is that peoples’ motives are channelled away from premeditated crime into its (to us) symbolic counterpart-sorcery. Yet once the individual goes to the sorcery shrine, and indeed even before going, his motive for killing or hurting his enemy receives some form of gratification. Even before going to the sorcery shrine it is likely that an individual will fantasize the dire consequences that will befall his enemy as a result of his planned sorcery. This wish is immediately reinforced at the sorcery shrine, particularly in Sinigama and Munneswaram. Once the client goes to Munneswaram he is impressed by the gossip of other clients, shopkeepers in the vicinity, and priests or their assistants, who talk of the power and vindictiveness of the deity over evil doers. In Sinigama the priest often loudly and openly brags before his clientele about the power of the deity and the people who have recently died, in the most dramatic circumstances, as a result of his sorcery. Once inside the shrine the client practices cathartic techniques, both in Munneswaram and in Sinigama, and to a lesser extent in Kahatapitiya. Indeed one is amazed at the sheer sadism and vindictiveness of the curses uttered by the priest, and repeated by the client, against the enemy. It is a kind of action that will reduce the tension in the client and satisfy his immediate desire for vengeance.

(d) For most people steps (b) and (c) are probably all that matter, for after some time complex sociological factors may intervene and resolve on the local level the crisis that instigated the client’s action. For example, the courts may take over the problem and resolve it on that level; arbitrators and mediators may step in and conciliate the persons involved in the conflict; or once the individual has gone to the shrine, returned home, and immersed himself in the routine of village activity, he may give progressively less and less thought to the problem. Time probably is a good healer of social and interpersonal conflicts.

(e) Yet undoubtedly there are others who want concrete results from their magical actions. They may rationalize the failure in one shrine, and seek other and more powerful types of sorcery. For some the self-fulfilling prophecy operates, not simply in the rare dramatic manner as in “voodoo-death,” but in a more subtle and much commoner fashion. In village life, with its poor technological resources, people are constantly beset by personal vicissitudes, e.g., crop failure, illness, death in the family, or indebtedness. Hence it is possible for a client to interpret any personal misfortune that befalls his enemy, or a member of his family, not as an accident, coincidence, or chance (these terms have little significance in village life) but due to the client’s practice of malevolent magic. This is also true of those cases where the client cannot identify his enemy. In such an instance the misfortune that befalls almost any individual he does not like may be interpreted as due to sorcery, and that person himself may be retrospectively identified as the enemy. Thus the effectiveness of magic is constantly validated by the nature of human experience in small communities.

The data from the Kahatapitiya mosque permitted us to test the frequency of successful cases of sorcery. At Kahatapitiya the client observes a vow to offer a gift to the saint of the shrine if the sorcery has been successful. This meant that there were two categories of clients-supplicating clients (those who came to practice sorcery) and returning clients (whose sorcery had been successful). At Kahatapitiya during the 6o-day period, there were 24 returning clients as against 442 supplicating clients. Thus one in every eighteen cases of sorcery was “successful,” which perhaps is sufficient to reinforce the belief in the efficacy of the shrine.[4] However, in 50 per cent (twelve cases) of the successful cases, “successful” was interpreted as the recovery of a lost object or money (ten cases) or an apology or confession by the victim (two cases), without any physical punishment to the victim. Two of the clients interviewed felt dissatisfied with the outcome of sorcery, since the enemy did not suffer physical harm. In twelve of the successful cases the enemy was punished by the deity by death or personal injury. Thus one in 37 cases of sorcery resulted in the true fulfillment of the predicted consequence. We do know, however, that in at least three of these cases the client interpreted some incident that occurred in his village-death or injury to a person-as due to his sorcery, and the victim was retrospectively identified as his enemy. In another case the client voluntarily supplied the information that the prediction was fulfilled in that the enemy broke his leg in acar accident one year after his supplication at the sorcery shrine.

(f) The foregoing argument brings us to another important factor in the behavior of some clients. If a client strongly believes that sorcery is effective, he already imagines the death of his enemy sooner or later. His enemy is alive, but it is just a matter of time before he dies. Thus some individuals at least can indulge in a fantasy satisfaction of the wish and thereby delay the actual satisfaction of seeing the death of the enemy.

(g) Nevertheless, it is important to reiterate that the consequent satisfaction of the motive is not crucial to the point of view taken in this paper. What is crucial to the theory is that the intention or motive of the actor has been diverted into the practice of sorcery, rather than into the actual practice of planned physical violence.


What we have said of Sri Lanka is applicable to other parts of Asia and Africa for which we have evidence. In a symposium on African homicide and suicide (Bohannan i96o), several anthropologists noted the spontaneous nature of criminal violence. L. A. and M. C. Fallers (1960: 76) have this to say about the Busoga of Southeastern Uganda: “. .. in most of the cases in our sample the deed could not have been planned in advance, but rather was committed on a sudden impulse.” And A. W. Southall (1960: 222) says that among the Alur of Northwest Uganda “homicide never seems to be premeditated.” Jean LaFontaine (1960: 95) is even more specific for the Gisu of the eastern border of Uganda: “It is noticeable that in Gisu society killing is rarely the result of a premeditated act. . . Of the 99 cases studied here there are only four which can be said to be premeditated; this is less than five per cent of the total.” The view of human nature that emerges from these studies seems to me biased and completely implausible. It is highly probable, and the ethnography indicates this, that in these societies also the motive for premeditated murder and violence is being canalizedin terms of sorcery and other forms of supernatural action. To get a correct view of the problem anthropologists must move away from the study of sorcery accusations to that of sorcery practice. This is admittedly difficult, but it is an important and rewarding task.

***  ***

Gananath Obeyesekere was attached to the Department of Anthropology, University of California, San Diego, when he wrote this article.

This article was retyped for Thuppahi by Nadeeka  Paththuwaarachchii


[1] This research was aided by grant from the Foundations Fund for Research in Psychiatry (No. G 67-363), for which I thank them. Several colleagues and friends have made comments and criticisms on earlier drafts of this paper. In particular I wish to thank Roy D’Andrade, David Jordan, Milton Singer, and Melford E. Spiro. The writer also acknowledges a profound intellectual debt to A. I. Hallowell’s work on the Salteaux, notably his (1955) Aggression in Salteaux Society which has direct bearing on the theme of this paper. Though the Ojibwa do not, it seems, actually practice sorcery, nevertheless sorcery fears and accusations help express and canalize aggression in this society.

[2] The practice of sorcery by monks is viewed as irregular and unethical by both doctrinal Buddhism and popular standards. The public does not treat these sorcerer-monks as “true monks”; but they are feared and respected. Why some monks practice vas-kavi is not easy to explain. Perhaps the role immunizes them from too close a public scrutiny; or it may be that the monk role gives them the time, leisure, and opportunity to study the esoteric skills involved in vas-kavi. Psychologically viewed, these people are pathological murderers, and I suspect that some of them may actually administer poisons disguised as charms.

[3] Fees charged by the priests may be as highas 150 rupees in Sinigama for a “special” puja for the deity, and as low as Rs 5.00. In Munneswaram the standard fee paid to the priest was Rs. 21.00. We were puzzled by this figure until we realized that Rs 21.00 was at one time the standard equivalent for a guinea, which is the sum charged by London’s Harley Street medical specialists, and adopted by Sri Lanka’s medical doctors. The sorcery priests have taken over this norm from their Sinhalese colleagues, illustrating the business-like, rational orientation to their work!

[4] Assuming, of course, that the numbers of persons who come to the shrines are constant.


Bloch, H. A. 1960. Report on Homicide, Attempted Homicide and Crimes of Violence. Dittoed, Ceylon Police. Colombo.

Bohannan, P. J. I960. African Homicide and Suicide. Princeton.

Fallers, L. A., and M. E. Fallers. 1960 Homicide and Suicide in Busoga. African Homicide and Suicide, ed. P. Bohannan, pp 65-93. Princeton.

Hallowell, A. I. 1955. Culture and Experience. Philadelphia.

Jayawardene, H. S., and H. Ranasinghe. 1963. Criminal Homicide in the Southern Province. Colombo.

Kluckhohn, C. 1962. Navaho Witchcraft. Boston.

La Fontaine, J. 1960. Homicide and Suicide Among the Gisu. African Homicide and Suicide, ed. P. Bohannan, pp. 94-129. Princeton.

Obeyesekere, G. 1967. Land Tenure in Village Ceylon, A Sociological and Historical Study. Cambridge.

Pieris, R. 1956. Sinhalese Social Organization. Colombo.

Southhall, A. W. I960. Homicide and Suicide Among the Alur. African Homicide and Suicide, ed. P. Bohannan, pp. 214-229. Princeton.

Wood, A. L. 196I. Crime and Aggression in Changing Ceylon. Transactions of the American Philosophical Society. Philadelphia.



Filed under accountability, caste issues, cultural transmission, economic processes, female empowerment, heritage, historical interpretation, Indian traditions, landscape wondrous, life stories, modernity & modernization, performance, politIcal discourse, religiosity, security, self-reflexivity, sri lankan society, trauma, unusual people, vengeance

2 responses to “Sorcery, Premeditated Murder, and the Canalization of Aggression in Sri Lanka

  1. Pingback: Constitutional Melee: Dayan vs Rajan, Sanjana, Asanga and the Ethical Hosts | Thuppahi's Blog

  2. Pingback: Foreign Diplomats in Temple Trees – e-Con e-News

Leave a Reply