6. Pre-Marital Sex and Social Control among the Rotumans
[Published in American Anthropologist 66:266-283, 1964]
THIS paper has a dual purpose; first, to describe the nature of social controls governing female sexual behavior prior to marriage among the Rotumans in Western Polynesia, and second, to illustrate the use to which demographic data can be put in testing hypotheses formulated from ethnographic evidence. 
SETTING AND SOCIAL ORGANIZATION
The island of Rotuma is located some 300 miles to the north of the Fiji group, on the western fringe of Polynesia. It is a volcanic island of about 17 square miles, and supports a population of approximately 3,000 persons. Racially the Rotumans are somewhat mixed, a consequence of the island's proximity to islands with Micronesian, Melanesian and Polynesian populations. There is also a Caucasian admixture introduced by renegade sailors early in the 19th century. The Rotuman language also reflects the island's anomalous position and its classification has recently been the focus of considerable discussion in the professional literature (cf. Grace 1959; Capell 1962). Culturally, however, Rotuma manifests a Western Polynesian base, Samoan and Tongan influences being most clearly evident.
Following discovery by Europeans in 1791  Rotuma was subjected to rather intensive acculturation from whalers, traders, and missionaries. The latter consisted of English Methodist ministers and French priests, who divided the island between them. In 1960 approximately two-thirds of the population were Methodist and one-third were Catholic. Politically Rotuma has been affiliated with Fiji since 1881, when it was ceded to Great Britain at the request of the chiefs.
Today Rotuma is still, despite over a century-and-a-half of contact with Europeans, a well-integrated, essentially conservative community. Western patterns have been adopted selectively, and the character of social life has not been seriously affected. In addition to the population of Rotuma, nearly 2,000 Rotumans now reside in Fiji, the bulk of these forming substantial enclaves in the capital city of Suva and at the gold mines in Vatukoula. Fiji offers opportunities for occupational advancement not otherwise available in Rotuma, and a sizable number of Rotumans have attained the status of teachers, assistant medical officers, health inspectors, and other positions of responsibility. Others have become skilled workers and command good salaries.
Since Rotumans in Fiji are exposed to far more intense acculturation than their kinsmen on the home island, they are, in general, much less conservative. Many persons, particularly some of the more successful ones, maintain only peripheral ties with the Rotuman community. Nevertheless, for the large majority of people, ties with fellow islanders remain intense, and the character of their social relations is unmistakably Rotuman. The relationship between Rotuma and the enclaves in Fiji is maintained by considerable mobility back and forth, and by indirect means of communicative and economic interaction (Howard 1961).
The Rotumans maintain a system of bilateral relationship, which includes recognition of the personal kindred (kainaga) as the basis for social relations. The degree to which kinship ties are actively maintained is variable, depending on several factors. Close kinsmen, genealogically to the third or fourth degree (with a degree counted as number of connecting links plus one; e.g. Mo equals first degree, MoBr equals second degree, MoBrDa equals third degree, etc.), are more recognized, but more distant ones are apt to be ignored unless it appears expedient to keep the ties alive. Expediency usually depends on residential proximity, which renders a person accessible in time of need, and on degree of wealth, which yields obvious advantages.
The term "kainaga" is also used to designate a bilateral descent group, which has significance for the system of land tenure. Most land is associated with a house-site (fuag ri), and ability to trace ancestry consanguineally to a particular house-site entitles a person to membership in the relevant kainaga. It is generaly asserted that each person has a claim in each of the fuag ri of his eight great-grandparents.  Kainaga land is managed by a pure, who is chosen by the members of the descent group to act as steward. The pure is obliged in theory to assist the members of the kainaga whose land he controls, should they require material aid (cf. Howard 1962).
Unquestionably, the most significant social unit in Rotuma is the individual household, which for the most part is economically self-sufficient. The average household contains approximately seven or eight persons, although many are considerably larger. Most consist of a nuclear family, with various bilateral extensions on either the husband's or wife's side. Social relations within the household are generally warm and cordial, and even the respect barriers traditionally prescribed between affines are apt to disappear with time. The men in the household do most of the farming and cut the copra, which is the main source of income. They also do some of the fishing, usually with lantern and spear, but sometimes with hook and line in the deep sea beyond the reef. The women weave mats, which are used domestically and for ceremonial exchange, and do most of the daylight reef fishing with nets. The division of labor is not rigid, however, and quite often all the members of a household work as a team at a particular task. In general, the status of women is very high, and the "good" husband does everything possible to make his wife happy.
Another social unit is the ho'aga, or work group. The historical evidence suggests that ho'aga were originally local kinship communities that lost their kinship unity as a consequence of acculturative pressures (cf. Gardiner 1898; Howard 1962). Today they consist of a group of households in the same locality, (though not necessarily adjacent) which work together on community projects under the direction of a sub-chief (fa es ho'aga). The functions of the ho'aga often overlap with those of the kainaga, particularly in ceremonial matters. Thus ho'aga members are obliged to assist one another in the preparation of weddings, funerals and other life-crisis ceremonies.
The largest social unit into which Rotuma can be divided is the district (itu), of which there are seven. These are primarily political units, under the direction of a paramount chief (gagaj es itu). He is the district's representative to the Rotuma Council of Chiefs, which meets regularly with the Government appointed District Officer to discuss affairs concerning the island.
Among the Rotumans in Fiji there are no social groupings which correspond directly to the ho'aga or itu, although the location of homes in Rotuma plays a significant part in the associations that are formed.
THE INTEGRITY CIRCLE
Of crucial significance for the purposes of this paper is the degree to which individuals are emotionally involved with one another, or to put it differently, the degree to which they manifest concern for one another's behavior. Those persons with whom one is actively concerned may be said to constitute an individual's integrity circle, the implication being that one's own integrity is tied to theirs. This may be considered to consist of two related attitudes, the first involving the degree to which one is shamed by the person's misdeeds, and the second involving the degree to which one is angered by an offense against the person. Thus if one's mother does something improper, he is more apt to experience shame than if a stranger were to behave in the same way; likewise, if someone injures a person's mother it is more likely to arouse anger.
An analysis of our data on interpersonal relations, consisting of personal observations and questionnaire material gathered from 30 subjects, revealed that the intensity of concern manifested for others, both among Rotumans in Fiji and on the home island, tends to be governed by two inter-related principles; namely, the degrees of genealogical and residential distance. To summarize our results, persons tended to manifest:
The integrity circle thus encompasses the personal kindred (in varying intensities). Since the criterion of inclusion is essentially based upon emotional commitment, neighbors and friends of long standing may also enter in, and sweethearts are certainly included. A person's integrity circle changes both in personnel and intensity through time. In the long run, alterations in personnel occur through such phenomena as births, marriages and deaths, while alterations in intensity are brought about by changes in status, age and residence. Thus, when a man marries, his wife is added to his integrity circle, but his change in status from single to married man may also result in a decrease in the degree to which he manifests concern for his sisters, particularly if he moves to his wife's residence. In the short run, a person's integrity circle may expand or contract situationally. For example, a serious misdeed by a Rotuman in Fiji would probably arouse some feelings of shame in all Rotumans simply because of his identification as a Rotuman; likewise, an unwarranted attack by a non-Rotuman on a Rotuman would arouse collective resentment.
As a theoretical concept, the integrity circle is of consequence for social control. To say that a person would be shamed by another's misdeeds is to imply that he has a stake in controlling that person's behavior; and on the other side, concern about offenses to the person implies an attempt to restrain others from such attacks, either directly or by the threat of retaliation. For the purposes of this paper, our interest in the integrity circle focuses upon the concern which various individuals manifest for the behavior of unmarried girls past puberty.
The Rotumans maintain a double standard with regard to pre-marital sex, encouraging the boys within their integrity circle to pursue sexual gratification, while discouraging the girls. Restrictions upon unmarried girls apparently date back to pre-contact times, as is revealed by descriptions of virginity tests at marriage (cf. Lesson 1839:435).  These restrictions were reinforced by the missionaries who, for obvious reasons, took a grim view of fornication. The aboriginal controls were social as opposed to moral, however, and the teachings of the missionaries have not seriously modified this emphasis. The rationale behind restraint in the pre-contact period was that virgins were more desirable as wives, since their husbands would not have to face the embarrassment of meeting other men who had taken sexual liberty with their spouses. Such restraint, by increasing a girl's desirability, also enhanced her chances of making a suitable match, to the benefit of her ho'aga (then constituting a kinship community), in some cases yielding social, economic and even political benefits.
The basis of the current Rotuman attitude toward pre-marital sex is similar. It is rooted in the conception that sexual intercourse involves a male taking license with a female. Whether or not she consents is not the issue; sexual license is considered justified only when the male accepts the corresponding responsibilities. Specifically, this means assuming economic support. To state this situation in terms of social economics: sexual license over women is a valued "commodity," and for a man to take it without paying the appropriate price (assuming economic support) is equivalent to stealing. In the aboriginal system this commodity was "owned" by the ho'aga, and it was they who benefited by a favorable marital transaction. Correspondingly, it was the members of a girl's ho'aga who had greatest interest in controlling her behavior, and by taking sexual liberty with an unmarried girl a boy was committing an offense against the whole group. In the current social system, with the ho'aga altered in character so that it is no longer a kinship unit, no clearly circumscribed group stands to gain in a comparable way by a "good" marriage. Yet pre-marital chastity is still valued for girls, and for a boy to take sexual liberties with a girl and not pay the price is still regarded as being more than just an offense against the girl; it is an offense against all those persons who are concerned with her integrity. Therefore, all those persons who include a girl within their integrity circle are concerned with controlling her behavior; correspondingly the more intense their concern, the greater is their stake in the efficacy of such controls. We are therefore led to the proposition that social control will be exercised over a girl to the extent that persons whose integrity is affected by her actions are in a position to restrict her behavior.
LOVE IN THE LIFE CYCLE: FROM INFANCY TO MARRIAGE
In Rotuma, as in most other Polynesian societies, children are normally greatly desired and are welcomed into every home. As a general rule infants and young children are a focal point within each household; they are indulged by everyone, often to the extreme. Until they are able to walk, children are continually in someone's arms, and enjoy a great deal of bodily contact. Demonstrations of affection by physical means continue in the form of fondling and caressinglong after the child has begun to walk, and children tend to seek physical contact with those whose affection they desire. Indulgence also takes the form of gift giving. The "good" parent is the one who grants his children's every reasonable request; if he can anticipate their desires so much the better. Toys and other presents are given to youngsters whenever there are a few spare shillings, and it is not unusual for parents to sacrifice household effects, which an outsider might regard as necessary, to buy their child a toy. In most households this pattern of indulgence extends to eating behavior; children are fed first and are given the best food while the elders content themselves with the leftovers. This association of affection with material giving is continually reinforced, and is reflected in the connotation of the word hanisi (the nearest Rotuman equivalent to our word "love"), which implies willingness to give tangibly rather than an emotional state.
Socialization focuses upon teaching children how to behave in the company of different people. They are taught whom they must respect and with whom they may act in an unrestrained manner. Particular emphasis is placed upon restraint in public, or in the company of high ranking persons. Rotuman etiquette forbids "showing off," and acting "proud" (fakman'ia) is the nearest Rotuman equivalent to a sin in the Western sense. The dominant disciplinary technique is ridicule, which can be sharply derisive at times. The effectiveness of ridicule as a social control lies in the degree to which love is withdrawn; and it is apparently successful, since the majority of Rotuman children are well behaved according to their own cultural standards.
These three features of child rearing--a high degree of bodily contact with physical demonstrations of affection, the association of affection with material indulgence and discipline by ridicule--play an important part in influencing the character of Rotuman courtship behavior. As an apparent consequence of early physical contact, adolescents are particularly responsive to similar contacts with peers. Motivations in this direction are doubtlessly increased by a progressive lessening of physical interaction within the family of orientation as the child matures. Contact between brothers and sisters is particularly likely to be restricted. Physical interaction between friends of the same sex provides some substitute satisfaction, this not being subject to implications of homosexuality under usual circumstances,  but it is mainly in the direction of heterosexual love, in the physical sense, that boys and girls are led.
The association of emotional commitment with material indulgence carries over into adolescence with two important ramifications. The first is that "giving" becomes instrumental for the gratification of emotional needs; the second is that a person's chances of being kept materially comfortable, if they are not economically independent, are increased to the extent that they command affection from others. A consequence of the first circumstance is that adolescent girls, being economically dependent, generally have control over only one highly valued "commodity," i.e. their sexual favors. They are therefore motivated to grant sexual license, having no other property with which to "express" love in a culturally suitable manner. The boy, in reciprocation, is expected to give presents of money, perfume and other items. He thereby not only demonstrates his love, but also permits the girl to balance her social economics, although her family would no doubt be dissatisfied short of total acceptance by the boy of her support (i.e. marriage). A girl who permitted a boy to take continual license with her without reciprocation would be considered a "larrikin" (the implication being that she liked sex for its own sake, and was not using it as a gift of love). If knowledge of the case spread, she would be regarded as fair game for any boy. The main consequence of the second circumstance, the association of material comfort and commanding affection, is that girls who are otherwise deprived have access to material resources by forming alliances with boy friends. A girl might thus be considered to be motivated to engage in pre-marital sex activities to the extent that she does not receive satisfaction of her material desires at home. 
The third factor, discipline by ridicule, has perhaps the most direct effect on patterns of courtship behavior. As a result of its dominance, children soon become extremely sensitive to the opinion of others, outside as well is within their integrity circle. This sensitivity leads most Rotumans to scrupulously avoid situations that might expose them to ridicule, and since courtship behavior is a prime target for teasing and banter, there is a strong tendency to avoid its public display. Overt displays of affection are particularly inhibited, since it is also regarded as a form of showing off. As one informant put it, "If a boy and girl showed love in front of other people it would be just as though they were saying that they were the only ones who had a lover." Even married couples take care not to express affection in front of others.
For these reasons courtship tends to be surreptitious, and careful efforts are made by lovers to conceal their alliances. As a corollary to this situation, unrelated boys and girls take care not to be seen talking to one another in private. If they are, it will be inferred that they are lovers, or are in the process of becoming lovers. Social pressures would then be brought to bear urging that the affair either be legitimized or terminated. Even at European style dances (mak fifis) couples are cautious not to speak with one another on the dance-floor, since the wrong (or the right) conclusions may be drawn. By operating at the trigger of such small incidents, social controls serve as a sanction to those who might be tempted to indulge in proscribed behavior (i.e. sexual alliance), as well as to those who are already having an affair. The sanctions are in theory, therefore, both a preventative and a cure.
The impact of these social pressures does not result in a thorough suppression of courtship behavior; it often, in fact, has the reverse effect. A boy and girl may have been seen talking innocently, but when the gossip and joking start they may decide to have an affair since everyone is talking about them anyway. Rather than suppress courtship, social pressures lead to an elaboration of the process, bringing into play all the ingenuity that boys and girls can muster. A game of intrigue is continually in operation, with notes being passed through trusted friends and clandestine meetings being arranged on the spur of the moment. Boys usually make the first. approach, although they may be reluctant until they have had some indication from the girl that their advances will be accepted. Often time is short, involving only a few seconds in which a boy can make his intentions known. In this precious time he must profess his love and attempt to arrange a meeting, but if the girl is willing--if she has been waiting for his approach--a great deal can be decided very rapidly. Meetings usually take place in the bush, where the chances of being caught are minimal. Girls may also arrange to sneak out of their house at night for a rendezvous in a nearby cooking house or other building; more rarely, since it involves maximum risk of discovery, a boy will sneak into a girl's house with the expectation of leaving before her parents or guardians awake.
All this is not to indicate that girls are readily willing to participate in such affairs. Even though they may like the boy who approaches them, most girls are very much concerned about their family's wishes and sincerely try to avoid casual involvements. A great deal of seduction is almost always necessary. The routine of seduction is for the boy to proclaim his love as intense and to promise to marry the girl as soon as the time is right. The girl responds with assertions that he is not really sincere, that he only wants to have intercourse with her and then will leave, making a fool of her. He may then offer presents to demonstrate his sincerity, and unless something interferes, the "negotiations" continue until the girl has extracted sufficiently intense promises to justify, in her own mind at least, her submission--or until the boy gives up. In addition to the near universal techniques of flattery and expression of intense desire (e.g. "Your eyes are more beautiful than the stars; I can not live without you," etc.) one other approach is especially likely to be successful--playing upon the emotion of compassion. Thus, whereas boasting and showing off generally arouse resentment, acting helpless and forlorn arouses sympathy and generosity. To aid those in want is a dominant Rotuman value, and the word hanisi connotes not only material giving, but pity as well. A boy is therefore much more likely to be successful in his seduction if he can convince the girl that he is a poor, unloved, mistreated fellow.
Since contraception is not practiced, except perhaps among a few of the most acculturated persons,  the risk of pregnancy is quite great. Abortion is practiced to some degree, usually through the use of harsh purgatives or by rough massage, but success is infrequent. In most cases, when a girl discovers her pregnancy, she confronts her lover with the news and reminds him of his promises. If he has been sincere and is willing to marry, they then take the appropriate steps, but in some instances he has never seriously considered marriage and terminates the relationship. The girl is then regarded as having been "tricked," and her family holds the boy in contempt.  If he formally apologizes to the girl's parents through ceremonial gifts of kava or specially prepared food, and if he accepts economic responsibility for the child's support, the situation is alleviated to some extent, but a young man who is a chronic philanderer becomes an object of scorn even to his own relatives, rather than a hero.
There are three principal types of marital arrangement. The first is the sok faeag, which is characterized by formal negotiations between the boy's and girl's families. The suitor, or a member of his family that desires the match, initiates the negotiations by sending a representative to the girl's family to speak in his behalf. Often it is his father, or a fluent uncle, but if it is desired to add weight to the proposal, a man of rank, the fa es ho'aga or even the district chief, might be implored to make the overture. The girl's parents can, although it is rare nowadays, make the decision without consulting her, and marriages may thus be arranged between parties who have never spoken to one another. If the offer is accepted, the two families engage in a series of exchanges which culminate in an elaborate wedding ceremony, after which the couple ideally reside uxorilocally. To have a marriage arranged by sok faeag is considered the proper way, and although it usually necessitates a large and expensive wedding, a girl's family generally prefers it since it brings no shame.
A second type of marital arrangement is the fu'u ("to stay"). It is perhaps the most common type, and occurs when a boy goes to the girl's house and indicates his intention of assuming the role of her husband. Fu'u marriages generally take place only between couples who have been having an affair, and when their alliance is made known to the girl's family the boy will almost certainly be accepted. If accepted, he remains in the house until a legal wedding is arranged. For a boy to go fu'u is a strategic move when a match is opposed by either his or the girl's family. In other cases a boy may be encouraged to go fu'u as a means of avoiding the expense of a sok faeag wedding. This is particularly likely if it is common knowledge that the couple have "known" each other, since under such circumstances the most significant feature of a sok faeag wedding is eliminated--the absence of shame. A fu'u wedding contains many of the elements of a proper wedding, but is performed on a smaller and less lavish scale.
The third type of marital arrangement is called taupiri ("to follow"), and consists of the boy bringing the girl to his home. It is the rarest type of union and is considered the most shameful. If indeed a legal wedding ensues, its spirit is expressed in the phrase, "Fitama a'ma'akia iria" (just to make them clean). Such a union is likely to take place only when a girl's family completely rejects her suitor, or when they have arranged a match for her in spite of her attachment to another boy. For a girl to go taupiri thus constitutes an act of outright defiance, for which she must formally apologize to regain her family's good graces.
All these features of child rearing and courtship are characteristic of Rotumans in Rotuma; in Fiji the situation is somewhat altered. For the majority, attitudes are similar in both places, but in Fiji the degree of personal freedom available to unmarried girls is considerably greater. Many of them work in commercial establishments and hence are in constant interaction with boys to whom they may speak without restraint. Even parents are likely to grant girls freedom, going so far as allowing them to date as a means of showing that they are "modern." Despite these concessions, however, a stigma is still attached to becoming an unwed mother, and girls are urged to consider their families and exercise caution.
ASSUMPTIONS AND HYPOTHESES
Social deviance can be considered from two alternative points of view. On the one hand is an approach based on the relative frequency of behavior patterns, on the other, an approach based on the degree of congruence between behavior and prescribed norms. Deviance in the first case consists of infrequent behavior; in the latter case it consists of failure to follow the prescribed rules. From the former point of view pre-marital sex indulgence by Rotuman girls can not be considered deviant behavior; at most it is a secondary mode. By adopting the second definition, however, an interesting research problem presents itself; namely, what factors, if any, are predictive of the degree to which Rotuman girls can be expected to participate in pre-marital sex relations (and thus to engage in proscribed behavior)? It is this question which we have tried to answer by using demographic data.
Our basic assumption is that the factors which are determinate for conformity to, or deviance from, prescribed norms can be considered as operating along two dimensions: (1) motivation and (2) controls. In other words, factors are regarded as determinate if they either affect motivation toward or away from proscribed behavior, or if they affect the controls that promote conformity to the norm. These dimensions are non-exclusive in that a single factor may have effects along both of them.
An example of a motivating factor toward pre-marital sex is the emphasis on physical contact which is generated by early childhood experience in Rotuma. This does not provide us with a workable hypothesis, however, since we do not have sufficient information to differentiate Rotuman girls on that basis. To do so would require a considerable knowledge of their individual childhood experience. We have therefore assumed this factor to be a constant within the Rotuman population.
A factor which would also appear to be motivating toward pre-marital sex, and which can be used to differentiate subjects, is age. The assumption is that the longer I girl remains unmarried, the greater will be her tendency to engage in pre-marital affairs. This, we reason, is related to an increasing need to have parental love replaced by heterosexual love. Also, as a girl's chances of marriage diminish, she has less to lose by using sex as a device for "trapping" a husband. Furthermore, age is related to controls in that the longer a girl remains un married the more opportunity she has for getting involved.
If our interpretation of the ethnographic data has been accurate, another motivating factor in the same direction is economic deprivation. Since love affairs generally bring material rewards, girls who are otherwise deprived may be regarded as motivated to establish a heterosexual alliance as a source of economic relief. Determining the degree of deprivation a girl endures presents some difficulties, but our evidence suggests it is related to household configuration. Any child is most likely to be indulged by lineal ascendants, or by persons who have requested an adoption; step-parents or other relatives forced of necessity to adopt a child are likely to be less indulgent. More specifically, since economic resources are controlled by men, co-residence with indulgent male figures is the most important consideration. Thus we are led to conclude that girls co-residing with their father or a willing foster-father are most likely to be economically secure, while those co-residing with other relatives, or non-relatives, are most likely to be deprived. Correspondingly, we assume the latter to be more motivated to violate pre-marital sex norms.
The factor which would appear to have most significance for controls is proximity to core members of the integrity circle. It is they who keep a girl under constant surveillance, thereby limiting her opportunities, and their proximity tends to increase her awareness of the shame she will bring if she were to become involved. Not all members of a girl's integrity circle have an equal effect on controlling her behavior, however. Perhaps because men are freer with ridicule about sexual matters than are women they are more likely to be ego-involved and make greater efforts to control the behavior of unmarried women within their integrity circle. Thus, a girl's father and brothers are the ones most concerned with her activities, although the efficiency of unmarried brothers may be mitigated by their sleeping away from home and a preoccupation with their own courtship activities. Hence maximum control lies with the girl's father, and it is our postulation that if a girl co-resides with her father, maximum control will be exercised over her sexual behavior. The plausibility of this postulate is enhanced by the fact that the father's presence generally correlates with an intact nuclear family, which allows the core members of a girl's integrity circle to exert concerted pressure on her.
Another factor which we suspected might affect controls was religious affiliation. Our impressions were that significant differences occurred in behavior between Catholics and Methodists (in non-sexual as well as sexual areas), and it seemed that a higher proportion of Catholic girls were known as "larikins" and got into trouble. Our reasoning to account for this was as follows: Catholic ideology, at least as it was presented to the Rotumans, appeared to weight obedience more heavily relative to personal responsibility than did Methodism. This ideological emphasis seemed to be reinforced by the structure of the Catholic Church on the island, which maintained a vast personnel gap between those in authority (priests and nuns) and the congregation, while the Methodist Church maintained an elaborately graded structure of lay preachers, catechists and church stewards in addition to ministers. In addition, the association of authority with maleness, as embodied in the priesthood, appeared more marked among the Catholics.  We therefore postulated that Catholic girls were more apt to be submissive, and hence more prone to submit to sexual advances by an assertive male.
Finally, since social controls in general appeared to be more relaxed in Fiji than on the home island, we postulated that girls living in Fiji would have greater opportunity and less reluctance to become involved in pre-marital affairs.
From these postulations the following hypotheses were formulated for testing:
In order to test these hypotheses we have used demographic data collected from two sources: living persons and the registers of births, deaths and marriages for the Colony of Fiji. Data collection began with a household census of Rotuma. In addition to household composition, an individual data sheet was made up for each adult, including information on family affiliations, residential history, religion and other materials of use for determining status in the community. At a later date information of a similar nature was ascertained from five Rotuman communities in Fiji (Suva, Levuka, Vatukoula, Tavua and Lautoka). In addition to these data, birth records dating back to 1904 were copied onto 3" X 5" slips of paper and alphabetized by mothers' names. Death and marriage records were also copied, and on the basis of this information the data collected in the field were corrected. It was then possible for us to tell whether a woman had borne a child outside of legal marriage, and to accurately calculate the time gap between legal marriages and the birth of first children. By checking residence patterns against the whereabouts of a girl's father, or his date of death, it was also possible to determine with reasonable confidence whether or not they were co-residing prior to her union.
Since the problems of working with the records became increasingly difficult as we went back in time, we decided to limit our universe to women born after 1920 for whom the data were satisfactory. Girls born after 1944 were also excluded. The group thus consists of all living Rotuman women between the ages of sixteen and forty (in 1960), except those for whom the information was insufficient or contradictory. The total numbered 545 women.
On the basis of the demographic evidence the dependent variable, premarital sex involvement, was operationally defined in the following way: A girl was considered to have engaged in pre-marital sex relations if she (a) gave birth to a child without having been legally married, or (b) gave birth to her first child in less than nine months subsequent to legal marriage.
Several assumptions underlie this definition. First it is presumed that since knowledge of contraception is non-existent, pregnancies regularly occur among girls engaged in pre-marital affairs. That not all girls who have such affairs get pregnant should only lead our data toward randomness (assuming conception is not affected by any of the independent variables), hence any correlations are likely to be more significant than they show. Abortion has been treated in the same way. Pregnancy as a criterion also implies degree of premarital sex participation, since chances of conception are enhanced by repeated intercourse. This may be viewed as an advantage rather than a hindrance, since a lesser degree of indulgence is theoretically dependent on the same factors that lead to total restraint. We recognize, of course, that a certain proportion of children are born prematurely and that our criterion of births in less than nine months after marriage may not always indicate pre-marital sex, but here, too, any error should lead toward randomness.
The independent variables are operationally defined in the following manner:
1. co-residence with father--A girl was scored as co-residing with her father if (a) they had been living in the same household from the time she was 15 years old until the present (1960) and if she had remained unmarried without evidence of pregnancy, or (b) if they had co-resided immediately prior to the date of her first marriage or conception (i.e. nine months before the birth of her first child). Only continuous co-residence was counted. If discontinuous co-residence was indicated (with the exception of short holidays) the case was put into the category of other residence.
Hypothesis 1: A higher percentage of unmarried girls not co-residing with their father will show evidence of pre-marital sex involvement than girls co-residing with their father.
Table I shows that nearly twice as many girls not co-residing with their fathers get pregnant as those who are. Our expectations were thus fulfilled and the hypothesis supported.
An examination of Table V, which categorizes the figures by all four factors, further strengthens our conclusion. This table permits 12 separate comparisons of the co-residence factor (controlling for age, religion and domicile).
In nine categories the percentage of pre-marital pregnancies is higher among the non-co-resident group, in two categories the percentage is higher among the co-resident group, and in one category (involving only four cases) the percentages are identical. It is interesting that the only two reversals occur among the youngest Catholic groups; the significance of this will be examined shortly.
Hypothesis 2: The percentage of girls showing evidence of pre-marital sex involvement will increase with advances in age.
Table II adequately illustrates the validity of our second hypothesis. The significance of the age factor can be further illuminated by reference to Table V, which in this case permits eight comparisons (with controls for co-residence, religion and domicile). In five categories our expectations were fulfilled, with increased percentages of pregnancies directly correlating with age; in two categories a marked increase takes place between the under 20 and 20-25 age groups, with a decrease in the over 25 group (which in both instances involves few cases); and in one category a decrease takes place in the 20-25 age group followed by an increase in the over 25 group. It should be noted, however, that in every category the percentage of pre-marital pregnancies is higher in the over 25 than the under 20 age group. 
Hypothesis 3: A higher percentage of Catholic girls will show evidence of premarital sex involvement than Methodist girls.
Table III supports the hypothesis but religion does not appear to be as significant a factor as co-residence or age on the basis of this table alone. An examination of Table V is quite revealing, however. Of the twelve comparisons possible (with controls for co-residence, age and domicile), in eight categories the percentage of Catholic girls with pre-marital pregnancies is higher, in three categories the Methodist percentage is higher, and in one category (involving few cases) the percentages are equal. Of particular interest is that all three reversals occur in the youngest age group, and that in each case the Catholic percentages increase dramatically in the 20-25 age group (from 9% to 57%, 8% to 47%, 11% to 30%), while the Methodist percentages increase only slightly in two categories (from 22% to 28%, 24%, to 31%) and actually decrease in one (from 18% to 8%).
The inferences we have drawn from these figures are that controls internalized as a consequence of Catholic religious training, considered independently from external (social) controls, are more effective deterrents for young Rotuman girls than controls internalized as a consequence of Methodist religious training, but that Catholic controls are more brittle and tend to disintegrate more rapidly and more completely. This conclusion is further supported by the fact that co-residence is not relevant for the youngest Catholic group (the only two reversals of the co-residence hypothesis occurred in these categories) while it is for the Methodist group.
In Rotuma, the Catholic Church does, in fact, emphasize the sinfulness of sex to a much greater degree. This emphasis on sex morality, which is apparently successful in restraining young girls, also tends to render sex an either-or proposition. That is, once a Rotuman Catholic girl has yielded to her sex drive, internal controls tend to give way completely. They are replaced with such rationalizations (frequently heard from Rotuman Catholics regarding many matters of sin) as, "After all, I am only human." The conflict between Rotuman sex needs, which are stimulated by the emphasis on physical contact, and Catholic doctrine is reflected in a high proportion of nuns and lay brothers who defect in favor of marriage after a period of service. The Methodists, on the other band, place a lesser emphasis on sex morality, and hence tend to transmit controls of a more flexible nature--less effective in the short run but more lasting.
We also believe our original reasoning to be supported by the data, i.e. that the tendency of Rotuman Catholic girls to be submissive to male authority plays a part in their higher rate of pre-marital pregnancy, although this factor apparently comes into play only after the initial restraints are broken.
Hypothesis 4: A higher percentage of Rotuman girls residing in Fiji will show evidence of pre-marital sex involvement than those residing in Rotuma.
Although Table IV appears to lend some support to the hypothesis, the proportions are not as great as we had anticipated. An examination of Table V throws further doubt on the validity of the hypothesis; of the 12 comparisons possible (with controls for co-residence, age and religion), in six categories the Fiji group shows a higher percentage of pre-marital pregnancies and in six categories the Rotuma group shows a higher percentage. Close examination of Table V suggests that the slightly higher percentage of pregnancies among the Fiji group as a whole is in fact a corollary of the higher proportion of non-co-residence in Fiji, as one might expect on a priori grounds.
It is possible, of course, that other factors related to urbanism, which we have not accounted for, contaminate our data. For example, it may be that knowledge of contraception is more widespread among the urbanites, or possibly it is easier to avoid recording "illegitimate" births in Fiji. Our field data support neither of these alternatives, however, and we are forced to reject the hypothesis, and to conclude that residence in Fiji does not decrease the efficiency of social controls that are effective in Rotuma.
In this paper we have attempted to illuminate the nature of Rotuman courtship behavior and its relationship to social controls by using two kinds of information: ordinary ethnographic information obtained by participant observation and questionnaires, and demographic data obtained from a field census and government records. On the basis of the ethnographic evidence alone our hypotheses might have been final generalizations, though perhaps we would have stated them in somewhat different terms. By testing these against the demographic data, we were able to ascertain further information, leading to a confirmation of two hypotheses (1 and 2), a modification of a third (3) and the rejection of a fourth (4). Use of the demographic material also permitted us to examine one factor, the religious one, which we would have been most reluctant to discuss without some means of objectifying it. Not the least of our reasons would have been the fear of arousing hostility and possibly misunderstanding without being able to cite concrete evidence. Allowing for the possibility of alternative explanations, we believe the demographic data speak for themselves and preserve a claim to objectivity in a delicate area. Perhaps our most rewarding finding was that one of our hypotheses proved false. It now seems evident that our impression of laxity in pre-marital sex conduct in Fiji was simply due to the lesser degree of surreptitiousness with which courtship is carried out there. What would be hidden in Rotuma is often open to view in Suva or Vatukoula. The discovery of this misconception has led us to reconsider several other aspects of Rotuman behavior in Fiji and to regard with caution generalizations that previously seemed quite plausible.
The use of demographic data in such a manner is not new. Sociologists have used such materials for many years, but anthropologists have been slow to exploit available possibilities. We grant that in many cases such data are either unavailable or difficult to ascertain. For many problems they are irrelevant. But if we, as anthropologists, are to defend the claim that our investigation is scientific, we must be prepared to test even our clearest impressions against the widest number of cases possible, within as well as between cultures. For some types of generalizations at least, demographic data of the kind used in this paper offer a valuable resource for testing.
 The data used in this paper were collected between October 1959 and June 1961 in Rotuma and among Rotuman enclaves in Fiji. The field work was supported by a grant from the National Institute of Mental Health to the senior author. Further support has been provided by the Human Ecology Fund, permitting us to systematize the data upon our return from the field. We would also like to thank the Government of Fiji for making their demographic records available to us, and Dr. Lindsay Verrier for introducing us to the research possibilities provided by such information. An earlier draft of the paper was read by Ralph Bulmer, Jeremy Beckett, Douglas Yamamura and Thomas Maretzki, each of whom have made valuable criticisms. The responsibility for the paper is entirely our own, however. [back to text]
 Discovery of Rotuma is credited to Captain Edwards of H.M.S. Pandora. [back to text]
 Rotumans are vague about which of one's great-grandparents' fuag ri should be selected, but in fact a situation never arises in which they have to make a choice. In litigation, any genealogical connection, amply demonstrated, is regarded as legitimate and a person may well justify rights in two fuag ri from one of his eight great-grandparents while not claiming any rights in fuag ri of some others. [back to text]
 We are using the term village somewhat loosely, meaning in effect a residential community whose members are in frequent interaction, but while there are some clearly circumscribed villages as such, it is not always possible to demarcate boundaries clearly. [back to text]
 This is not to imply that virginity tests were carried out in every case; no doubt tests were less important among people of low status. It is also probable that the blood of an animal was sometimes substituted to make every thing appear legitimate, but the custom nevertheless implies that virginity was valued. [back to text]
 Young men are often seen walking hand in hand, as are young women, but we found little evidence of homosexuality as such. The one exception was a female transvestite who was known to have had an affair with at least one girl, and to have attempted to seduce several others. [back to text]
 "Material desires" in this context include all those things which money can buy, including such items as tickets to a movie show. [back to text]
 Contraceptives are not available on Rotuma. [back to text]
 A girl's first pre-marital pregnancy is generally regarded as her suitor's responsibility. For the girl, the stigma is rather like that of having been duped into giving away a valuable family possession. She may be accused of having been foolish, but she is not held morally responsible. Subsequent pregnancies, however, are considered her own fault. [back to text]
 This conclusion was supported by a good deal of independent data, including the explicit pronouncements of both Catholic and Methodist leaders. We often heard it said that the Catholics are more respectful to chiefs, government officers, etc. Also in support of this conclusion was the fact that an overwhelming proportion of girls working in service occupations in Fiji (e.g., waitresses, laundresses, etc.) were Catholic, while Methodist girls tended to be clerks, stenographers, etc. [back to text]
 An assumption implicit in our testing of this hypothesis is that fertility is independent of age, which is probably not the case. It is presumed, on the other hand, that increases in fertility are somewhat offset by increased knowledge of techniques for avoiding and terminating pregnancies. [back to text]