1The term marriage refers to an institution, a ritual, and a conjugal status. The legal and festive aspects of marriage and the nature of conjugal relationships have changed in contrasting ways since the 1960s. How, then, can we capture the many dimensions of marriage at the turn of the 21st century? How can we establish a common history when each aspect of marriage is approached through different disciplines—the institution by legal scholars, the couple by sociologists, and the ritual by ethnologists—whose historical analyses are not always of equal depth? How can we draw up an overall picture of the population living in France when these varied approaches, based on different materials—ranging from exhaustive statistics to local ethnographic data—cannot all be generalized in the same way?
2A first opportunity to cross-analyse the three dimensions of marriage at individual level was provided by the Couple Formation survey (La formation des couples, or FC) conducted by INED in 1983–1984 on a representative sample of individuals aged below 45. Bozon (1992) observed a practically universal decline in the 1960s of the ‘traditional marriage’ model in which the conjugal union and the wedding were overseen by the family and the clergy. Marriage of this kind was celebrated with a religious ceremony and marked the start of conjugal life. In the 1980s, as marriage started to lose ground in favour of free or consensual union, weddings appeared to adopt a simpler form. Fifteen years later, however, the ethnographic studies by Segalen (1997, 1998) challenged the idea that elaborate weddings were on the way out, with descriptions of sumptuous receptions that revived past traditions of festivities spread sometimes over two or three days. How should this return to lavish display be interpreted? Is it a long-term trend reversal—as observed in other countries (Otnes and Pleck, 2003; Ingraham, 2008)—or a methodological bias linked to the ethnographic nature of the observations? Does the development of elaborate weddings, with each couple being free to invent their own form of celebration, reflect the individualization of society in the sense that there is ‘no longer one ritual, but many rituals’ interpreted in multiple ways, as suggested by Segalen (1997, p. 165)? Or is it another instance of the normalization of marriage rituals via the paradoxical injunction to personalize, as suggested, for example, by Déchaux (1998, 2010) with reference to funeral rites, which are likewise increasingly individualized?
3In France, a study conducted between 1999 and 2014 on the wedding preparations of around 50 couples in their 30s shows the importance attached to the reception and to the details of each moment (Maillochon, 2016). Similar observations have been made in other western countries (Boden, 2003; Otnes and Pleck, 2003; Carter and Duncan, 2017). Couples are under pressure to make the wedding into quite literally ‘the best day of their life’; not simply the first day of a long and happy marriage, but an exceptional festive event. The fact that marriage now follows rather than precedes cohabitation has changed the meaning of the ritual, which no longer celebrates the formation of a new couple but serves as a means to draw attention to and ‘represent’ an existing one (Maillochon, 2016). But do later marriages and second marriages—which Bozon (1992) found to be simpler and less elaborate in the 1980s—now also involve sumptuous wedding receptions liked those described in the survey of couples in their 30s?
4The aim of this article is not to establish a quantitative sociohistory of marriage (who marries, since when, and why?), but rather a quantitative ethnohistory. Following on from van Gennep (1909, 1943, 1946) who studied the sequential logic of marriage rites to identify their initiatory meaning, this study focuses solely on the structural aspects of marriage in terms of its three definitions (institution, couple, and ritual) and their relative positioning at the individual level.
5The study draws upon data from two surveys, the FC survey and the EPIC survey of individual and conjugal trajectories (Étude des parcours individuels et conjugaux; Appendix A) conducted in 2013–2014 by INED and INSEE in metropolitan France. Each survey makes use of a national sample representative of the population living in France. Contrary to ethnological studies, no population category is deliberately excluded. In fact, these surveys may be seen as two modules of a single observation exercise, with the second repeating some of the questions asked in the first. They thus satisfy the conditions required for a genuine historical comparison. Both provide data describing the respondent’s marriage to the person who is his/ her partner at the time of the survey, thus introducing a selection bias for the most lasting unions. This bias is acceptable, however, as there is little difference between the marriages concluded between 1964 and 1983 and recorded in the 2013–2014 EPIC survey and those described in the FC survey 30 years earlier (Appendix B, Table B.2). To reconstitute marriage histories over half a century, this study also uses long series of statistical data from civil records or population censuses. These exhaustive data provide a means to assess the reliability of information collected in EPIC (Appendix B, Table B.1) and to broaden the results to the entire French population when possible. However, contrary to purpose-built surveys such as FC and EPIC, they cannot be used to compare information at the individual level on civil and religious marriages, wedding receptions, and conjugal event histories.
6So by combining various quantitative sources, this article examines the fate of ‘traditional marriage’ which, in the 1960s, was structured around a ‘traditional couple’ (direct marriage without premarital cohabitation) and a ‘traditional ritual’ that necessarily involved a religious ceremony and its inevitable trappings (a wedding dress, a large reception). Are the waning appeal of the marriage institution and the development of cohabitation, which have caused the demise of the ‘traditional couple’, also affecting the ‘traditional’ wedding ritual? The first part of this article (Section I) describes the conjugal context of contemporary marriage and the disappearance of the ‘traditional couple’. While religious marriage is in steady decline (Section II), it maintains a specific status, not unrelated to the aestheticization of wedding celebrations. The history of non-religious rituals is less smooth (Section III). Certain components, considered as ‘traditional’ because long taken for granted, are imbued with new meaning. This is the case for engagement, as described by Pugeault (2010). At the same time, newly invented ‘traditions’ are also emerging, such as bachelorette parties and secular wedding ceremonies.
I – The demise of the ‘traditional couple’
1 – An end to the exclusivity of marriage
7In the 1960s, conjugal life began with marriage (Bozon, 1991) Today, a union is formed when a couple move in together, whether or not the relationship is officialized at a later date. According to EPIC data, 15% of people who married in the years 1964–1973 had lived together before marrying; the proportion was 84% in 1994–2003 and 2004–2013 (Table 1). Marriage no longer necessarily marks entry into a union but rather one of its possible forms of institutionalization, alongside, since 1999, the PACS civil partnership (pacte civil de solidarité) open to both same-sex and different-sex couples (Rault, 2009).
8After the 1960s, each new development in conjugal behaviours was described in terms of the marriage model and initially interpreted as an intermediate stage in the process rather than as a questioning of the institution itself. In the 1970s, the term cohabitation in French was associated with the adjective juvenile (young) because it was assumed to disappear with age (Roussel, 1975, 1978). Consensual union was seen more as a ‘test marriage’ than as a new, long-lasting living arrangement. In the mid-1980s, however, it became clear that cohabitation was becoming an alternative to marriage (Leridon and Villeneuve-Gokalp, 1988; Roussel, 1989; Villeneuve-Gokalp, 1990; Toulemon, 1996). The PACS was then presented as a new form of ‘engagement’ for heterosexual couples, a transitional stage before true officialization of the union, possible only through marriage. Yet, much like cohabitation which never became a ‘test marriage’, the PACS, it would appear, is not becoming a ‘trial run’ for officialization. For most couples, it is seen as a lasting conjugal contract, sufficient in itself, not necessarily leading to marriage. According to EPIC data, only 3% of couples who have married since 1999 were previously in a PACS union with the same person, of whom only 15% say that they would have married earlier had the PACS not existed. The occupations of married and PACSed couples (Costemalle, 2017) and their cultural values (Rault and Letrait, 2010; Bailly and Rault, 2013) are sufficiently different to suggest that the PACS is a truly alternative form of institutionalized union rather than simply a step leading to marriage. While take-up was low in the early days (22,000 in 2000, of whom a quarter were same-sex couples), the PACS has grown steadily in popularity, with 100,000 unions concluded in 2007 and 194,000 in 2017 (Papon and Beaumel, 2019), primarily by different-sex couples (95% of PACS unions since 2008). By 2017, the number of different-sex PACS unions registered in the year had almost caught up with the number of marriages (four PACS for five marriages). In just 18 years, the PACS has become a standard form of heterosexual union, competing strongly with marriage, although this major shift in behaviours is still barely visible in terms of the overall stock of couples (Buisson, 2017).
Characteristics of weddings celebrated in France between 1964 and 2013 by period and by conjugal, familial, and institutional characteristics of the union
Characteristics of weddings celebrated in France between 1964 and 2013 by period and by conjugal, familial, and institutional characteristics of the unionCoverage: Weddings between 1964 and 2013 (n = 3,533).
(r): For the respondent. (c): For the respondent and his/her partner. (a): As defined by Rault (2018), i.e. celebrated by couple alone.
(d): Weddings involving a ‘large reception’ or ‘several large parties’ by contrast with those celebrated with a small group of guests, as a couple or with no festivities.
Interpretation: Among the weddings celebrated between 1984 and 1993, 91% were a first marriage for the respondent; 68% were preceded by prenuptial cohabitation, and in 14% of cases one (or both) partners already had children; 71% of these weddings included a religious ceremony, 37% were preceded by an engagement (all types), and 4% of engagements were celebrated by the couple alone; for 73% of weddings, one or more large parties were held; and a wedding in 1984–1993 had 91 guests on average.
9Over the last half-century, marriage has been replaced by cohabitation as the marker of entry into a conjugal relationship and is facing competition from the PACS as the means to officially register a union. Having lost is role in initiating and officializing a union, marriage has also gradually lost its uniqueness: when it does take place, it is no longer necessarily a once-in-a-lifetime event. According to vital statistics data, the number of remarriages of divorcees started to increase in the mid-1970s and reached a peak in the mid-2000s (almost 20% of marriages in 2005), in parallel with the increase in divorce over the period.  The EPIC data confirm this trend: just 3% of people who married between 1974 and 1983 were remarrying, compared with 17% between 2004 and 2013 (Table 1).
2 – The new place of marriage in family and individual event histories
10In the years 1964–1973, people who had cohabited before marriage were barely older (five months on average) than those who married directly. The difference was three years five months in the years 2004–2013. This transformation of the entry into a union has doubtless contributed to marriage postponement (Prioux, 2003), alongside other often cited mechanisms, such as longer time spent in education and later labour market entry, that have also increased the age at first union (Rault and Régnier-Loilier, 2015).
11Indeed, vital statistics data show that men married at age 38.4 years and women at age 36.0, on average, in 2018, compared with 26.0 and 23.8 years, respectively, in 1970.  Mean age at marriage has risen by 12 years in the last half-century and that of first marriage by eight years.  While all the milestones of ‘entry into adulthood’ have been deferred over this period—age at completion of education has increased by around three years, and that of first labour market entry likewise—none have been pushed back as far as marriage.  Entry into first cohabiting union has been delayed by just one year over the same period (Rault and Régnier-Loilier, 2015).
12Compared to couples of their parents’ generation, today’s couples marry at a radically different age and stage of their life cycle. The family’s role in the decision to marry has also evolved. According to EPIC data, ‘making the parents happy’ is rarely given as a reason for marriage, either past or present, and is mentioned in fewer than 5% of cases. The desire to become independent of one’s parents is given as a reason by 35% of those who married between 1964 and 1973 but by just 6% of those who married between 2004 and 2013. Family formation is still the main reason for wishing to marry, given by two-thirds of respondents who married in both periods, while many of those who married in the more recent period also see marriage as a way to express their commitment (40% of cases in 2004–2013). 
13The decline of the marriage institution and marriage postponement at the individual level are reshaping the link between marriage and family. Since 2006, more than half of all children (more than 60% in 2018) have been born outside marriage.  The EPIC data show that the traditional sequence of ‘marriage then children’ is more and more frequently reversed. The number of couples with children (from the current or a previous union) when they marry has been increasing continuously over the last 50 years, rising from 1% of all marriages concluded in 1964–1973 to 28% of those concluded in 2004–2013 (Table 1). The presence of children (from the current or a previous union) is more frequent for remarriages than first marriages.
II – The spiritual and cultural roots of religious marriage
14The decline in Catholic marriage—Catholicism being the main religion in France —became discernible from 1960 (Dittgen, 1994) and has been confirmed over the last 30 years. According to the French Conference of Catholic Bishops (Conférence des évêques de France),  there were 147,146 Catholic weddings in 1990 and fewer than half that number (70,369) in 2012. The decline in Catholic weddings is greater than that of weddings as a whole: 51% of all marriages (and 60% of first marriages) registered in 1990 were celebrated in church  versus just 29% (respectively 36%) in 2012.  This trend cannot be attributed solely to demographic changes (fewer marriages and a growing relative share of remarriages of divorcees that cannot be consecrated by the Catholic Church). It is also linked to the decline in religious practice, changes in attitudes to religion, and competition from other religious denominations (Dittgen, 2003).
1 – Contrasting levels of decline
15EPIC data provide information on the principal religions represented in France and on the religiosity of respondents in a union.  They reveal a substantial weakening of religious marriage over the last 50 years; 91% of all weddings between 1964 and 1973 included a religious ceremony compared with 51% between 2004 and 2013 (Table 1). Whereas religion was an omnipresent feature of the nuptial ceremony half a century ago, just half of contemporary weddings now include a religious dimension.
16Paradoxically, the proportion of religious weddings (all religions) has increased for remarriages—from 8% in 1964–1973 to 23% in 2004–2013—but it has decreased for first marriages, from 87% to 57% over the same period.  This rise can be explained not only by the increasing number of Muslim and mixed-faith couples but also by the ambiguity of the term ‘religious’ which, in the case of Catholics, may signify the sacrament of marriage (which, by definition, is unique), a blessing, or a moment of prayer led by a priest. The increasing proportion of religious second marriages does not make up for the decline among first marriages, but it reflects a change in the place of religious faith in conjugal unions, their officialization, and their ritualization. While a religious wedding is of great spiritual importance for a devout minority, it may also be instrumentalized for its ceremonial attributes.
2 – A religious wedding by conviction
17There are contrasting reasons for wanting a religious wedding, linked both to religious faith (a spiritual value) and tradition (a cultural, social, or family heritage), two dimensions that are not necessarily incompatible. When asked why they had a religious wedding (Figure 1), respondents who married for the first time in 2004–2013 mentioned their ‘religious convictions’ (44%) and ‘respect for tradition’ (33%) much more frequently than a desire for a beautiful ceremony (10%) or to make the family happy (6%). Tradition and family become less important in second marriages, behind religious convictions and the desire for a beautiful ceremony.
18While the proportion of religious marriages has fallen over the last 50 years, the share of couples who married religiously and who did so for reasons of religious faith has remained steady, at 41% in the years 1964–1973 and 43% in 2004–2013. The partners’ religion(s) influence(s) the reasons given for a religious marriage (Figure 2).  Couples of the same religion more often marry religiously for reasons of faith (notably Muslim couples) than mixed-faith couples and, especially, non-believers,  who do so by tradition (39%) and to make the family happy (36%).
3 – The growing importance of aesthetic criteria
19Religious marriage is also chosen for more pragmatic reasons, namely the desire for an eye-catching ceremony. While couples who decide to marry religiously because they want a beautiful ceremony are still a minority (11% in 2004–2013), their proportion has increased rapidly in the last 50 years. They were extremely rare in 1964–1973 (1% of weddings) (Figure 1). The choice of a religious wedding for aesthetic reasons is more frequently reported by nonbeliever couples (13%), whose number has risen sharply in the last 50 years, than by those with a religion (Figure 2). It also varies by the degree of importance attached to religion. In the years 2004–2013, 11% of couples with no religion and 18% of those attaching no importance to their religion nonetheless married religiously ‘for the ceremony’. This choice is increasingly explicit not only among non-believers and those with weak religious attachment but also Catholic couples (an increase from 1% to 11% between the 1964–1973 and 2004–2013 periods) as well as Jewish, Buddhist, and mixed-faith couples (from 3% to 12%), but not at all among Muslim couples (1% in both periods).
Change in reasons for wanting a religious ceremony by date of marriage
Reasons for wanting a religious ceremony by both partners’ religion
Reasons for wanting a religious ceremony by both partners’ religionCoverage: Religious marriages among first marriages concluded between 1964 and 2013 (n = 2,339).
20Religious marriage for aesthetic reasons reflects the change in individual attitudes to religion, notably a subjectivation of religiosity (Hervieu-Léger, 1999, 2003; Raison du Cleuziou, 2014) but also the aestheticization of wedding festivities. The wedding is now a celebration of love (Daumas, 2004) and a consecration of the couple’s life together. For many future spouses, the civil ceremony is too austere to symbolize their romantic commitment; it is (by definition) too administrative to express any feeling, too hastily expedited to mark the solemnity of marriage. Even if they are non-practising non-believers, some couples turn to religion, especially the Catholic Church, for a sacrament or a simple blessing to give meaning and, above all, solemnity to their commitment. In this respect, a Catholic wedding celebrated in a church or cathedral is clearly a more spectacular event than a Muslim marriage conducted in the family setting (Collet and Santelli, 2012).
4 – The ‘secular ceremony’: a new stage of the marriage ritual?
21According to EPIC data, the proportion of couples who married religiously while reporting no attachment to religion reached a peak in the 1994–2003 period, against a backdrop of social secularization, individualization of religiosity, and the aestheticization of weddings. While the aesthetic appeal of religious rites remained strong, a competing ritual emerged, that of the ‘secular ceremony’ which, after the civil ceremony (obligatory in France), provides an opportunity for bride and groom to reflect on their marriage vows in a carefully designed setting and to restore a form of spirituality to the wedding festivities. Such ceremonies make up for the shortcomings of the civil wedding in terms of symbolism and display while avoiding any reference to religious rites that may be meaningless for the couples concerned. Planned by the couple in accordance with their tastes and their personal and conjugal history, the secular ceremony celebrates love and commitment while combining both words (readings, exchange of vows, etc.) and rituals (exchange of rings or jewellery, lighting of candles, cord braiding, thread weaving, etc.). For bride and groom, it is a new invention rather than a tradition, corresponding neither to the reworking of a religious model nor to a reproduction of family conventions (Maillochon, 2016). Yet far from escaping social norms, the secular wedding as presented in social media, with its standardized formats and its professional celebrants, is in turn becoming progressively institutionalized. The religious conformism dictated by tradition (religious marriage was inescapable, whatever one’s individual beliefs) is now being replaced by a social and commercial conformism that imposes the duty of a spectacular wedding.
5 – The end of ‘traditional marriage’
22In the early 1980s, the ‘traditional marriage’ combining a ‘traditional couple’ (no cohabitation) and a ‘traditional ritual’, notably a religious ceremony, was still a widespread social reality. According to the Couple Formation survey, they accounted for 59% of first marriages celebrated between 1973 and 1983 (Table 2).  However, this type of marriage is practically non-existent today. According to EPIC data, only 13% of first marriages between 2004 and 2013 (and barely more second marriages) followed this pattern. The normalization of premarital cohabitation and the secularization of society have gradually put an end to ‘traditional marriage’. This change took place in several stages. It began with the demise of the traditional couple over the period 1984–1993 (by which time just 34% of first marriages were direct marriages). At that time, a majority of couples married religiously, but after a period of cohabitation (48% of first marriages). This model then lost ground over the 2004–2013 period, becoming barely more frequent than civil marriage (44% versus 39%). Civil marriage after cohabitation, which will soon represent more than half of all marriages, is thus the complete opposite of the traditional marriage of the 1970s in which the religious ritual marked the beginning of conjugal life.
Change in types of first marriage since 1973, Couple Formation (FC) vs EPIC surveys (%)
Change in types of first marriage since 1973, Couple Formation (FC) vs EPIC surveys (%)* Designations according to Bozon, 1992.
Coverage: First marriages.
23The proportion of couples who marry directly with a civil wedding has changed only a little over the last 50 years. However, these are generally not ‘deviant traditional marriages’ as designated by Michel Bozon (Table 2), but rather couples who may be geographically separated—for work-related reasons, for example—or who may have a preference for ‘living apart together’, i.e. being in a relationship without living under the same roof. Such marriages, paradoxically more frequent among people with religious beliefs, also include cases of mixed-faith unions which may be difficult to celebrate religiously.
24The diffusion of premarital cohabitation has sounded the death knell of the so-called traditional couple, but its impact on the choice of a religious ceremony has been limited. We will now look at the ways in which traditional rituals (festivities, betrothal, white wedding dress, etc.) have been transformed by the demise of the traditional couple and the relative decline of religious marriage.
III – The new rituals of tradition
25The data from the FC survey show that at the end of the 1970s, traditional marriages (no premarital cohabitation and religious ceremony) were the most ritualized form of matrimony, often including a formal engagement and a sumptuous wedding reception (lavish meal, more than 50 guests, etc.). Other types of wedding were simpler in form and scale. Marriage thus became less showy between 1960 and 1984, reflecting couples’ lesser compliance with religious and family norms (Bozon, 1992). More than 30 years later, the EPIC survey reveals the historical singularity of this simplification of traditional ritual and, above all, the extent to which the current individualization of practices is a relative phenomenon. Despite the disappearance of the ‘traditional couple’ over the last 50 years, the festive traditions associated with marriage have not been lost (Table 1), although their history is sometimes troubled. Certain components of the traditional marriage ritual (smart clothes, the wedding dress especially, and large wedding receptions) are still very much alive and even gaining new vigour (Table 1), while other components, such as engagement, are taking new forms. Entirely ‘new traditions’ are being invented, such bachelorette parties and marriage proposals.
1 – The growing importance of display
26‘Small weddings’ are unusual. In 1970, as in 2010, just 5% of weddings brought together fewer than 10 people. It is customary to celebrate one’s wedding day in good and numerous company, and this norm has been reinforced over the last 50 years. The mean number of guests has increased sharply in recent decades, from 77 on average in 1964–1973 to 108 in 2004–2013.  While the proportion of people who report having had a ‘big wedding’ has fallen slightly since the years 1984–1993 (Table 1), this is because the criteria for qualifying as such have shifted upwards: those reporting a big wedding in 1964–1973 had 92 guests versus 128 in 2004–2013. And the number of guests at the largest 25% of weddings has increased from 100 to 130 (from 200 to 350 for the largest 5%).
27The increase in guest numbers illustrates the changing approach to the definition of ‘friends and family’ that reflects the transformation of intergenerational relations since the 1970s. The bride and groom, older on average and more financially independent than their parents when they married, are also more closely involved in the wedding preparations, of which the guest list is a key component. Most couples not attracted to the idea of a traditional wedding refuse to surrender to family pressure in this respect. Yet weddings are still a family celebration (Maillochon, 2002, 2009). According to EPIC data, for 90.3% of weddings celebrated between 2004 and 2013, there was a strong family presence: in 27% of cases, all guests were family members, and in 63.3% they were invited alongside the couple’s friends. The proportion of weddings celebrated only with friends has increased substantially over the last 50 years but remains modest, rising from just 2.4% between 1964 and 1973 to 9.6% between 2004 and 2013. While couples more frequently ‘select’ their guests, including within their own respective families (notably because many are stepfamilies), their choice is still governed by numerous social constraints (Maillochon, 2011). The rules of politeness and reciprocity are respected at a conscious level, those of social homogamy much less so, although they play an equal role in the final choice of guests. Invited friends are often from the same social group as the couple or belong to the social space in which the couple positions itself (Maillochon, 2009).
28Weddings with a religious ceremony have the largest number of guests (112, on average, versus 58 for civil weddings). Weddings of people who already have children tend to be smaller affairs (62 guests compared with 102 for childless couples), as are second marriages (54 guests versus 99 for first marriages). Since the late 1970s, guest numbers have increased substantially for all wedding types. The probability of having a wedding with more than 50 guests was 5 times higher in the years 2004–2013 than in 1974–1983, all other things being equal (Table 3). However, it is second marriages and those with children that have seen the biggest increase in guest numbers over the study period (Figure 3). While breaching the canonical codes of marriage as a life-long commitment and as the setting for founding a family, these weddings follow its standard outward trappings (notably religious) and bring together a large gathering of people.
29Alongside the longer guest list, the extreme importance attached to the marriage setting (search for an ideal location, choice of wedding theme, coordinated decoration of the various venues, ‘colour code’ or dress code for the guests, etc.) is part and parcel of the contemporary wedding and has lengthened the preparation time (Maillochon, 2016). The EPIC survey data provide no information on wedding arrangements but do shed light on one key aspect of decorum, that of the wedding dress. 
30In the 21st century, 81% of women still marry in a white dress. Far from disappearing, this tradition has grown stronger since the 1990s despite the demise of the traditional couple that the white colour was intended to symbolize. Whatever the type of ceremony, the bride wears a wedding dress. A vital accessory for church weddings (worn for 95% of such unions between 2003 and 2014), a wedding dress is also worn by two-thirds of brides who have a civil wedding only. While rarely featuring in marriages preceded by births (21%) and for remarriages (34%) in 1964–1973, it is now a common choice (67% and 59%, respectively). A wedding dress is still more commonly worn in religious marriages than civil ones (Table 3), but it is in the celebrations from which it was previously banished because the virginity it symbolized was manifestly not respected—unions with children and second marriages—that it has gained the most ground.
31For an equivalent type of marriage, the probability of wearing a wedding dress rather than another outfit was 4 times higher in 2004–2013 than in 1974–1983 (Table 3). White is no longer the symbol of virginity (Currie, 1993; Bouchet, 1999) but of specialness for brides, who see no reason to abandon the splendour and solemnity of the event simply because they have been married before or have already experienced an(other) happy event (motherhood). For the rest, the wedding-dress market and the brides-to-be can always circumvent the problem of ‘pure white’ as a symbol of purity by replacing it with dresses that are ivory, cream, pearly white, or alabaster. Young women (under 25) who, in the 1970s, could at last contest the old traditions and modernize the image of the bride in her white wedding gown have now become the very artisans of its comeback. 
2 – A return to engagement
32In the mid-1980s, Bozon (1992) observed a sharp decline in engagements, which preceded fewer than half of all first marriages at that time (Appendix B, Table B.2). With the increase in age at marriage and families’ weaker involvement in partner choices and wedding decisions, their purely ritualistic version (a meal bringing together selected members of each family) seemed to be losing ground. Rarer still among couples who lived together before marriage or who married civilly, the tradition seemed to have little future. The EPIC data confirm its decline in 1974–1983 but also its stabilization beginning from 1984–1993 (to around one-third of marriages, Table 1) despite the simultaneous increase in civil marriages, remarriages, and marriages with children. Engagement was more frequent in the period 2004–2013 than in 1974–1983, for a constant marriage structure (Table 3). It is for civil marriages and, above all, remarriages that engagement has gained most in popularity (+35% and +200%, respectively, between 1974 and 2013); it has declined for religious marriages (−16%), suggesting a change in practices concealed under a name that remains unchanged (Figure 3).
Factors associated with nuptial and prenuptial rituals
Factors associated with nuptial and prenuptial ritualsCoverage: Marriages between 1974 and 2013 (n = 3,159).
Note: The odds ratios in bold are significant (p < 0.05).
(r): For the respondent.
33The French term fiançailles (betrothal or engagement) has many meanings, and its contours have been poorly defined since the 1990s. Pugeault (2010) shows that the term has become problematic for couples, who do not always know what it actually refers to (moral or religious commitment, preparation for marriage, or specific celebration). In recent decades, engagement rituals have evolved, moving beyond the family circle (which nonetheless predominates) to become an event celebrated by the couple alone or with friends (Pugeault, 2010). With the growing importance couples attach to the ‘announcement’ of their wedding—which may take multiple forms—this trend is becoming stronger. Since the 2000s, young couples in their early 30s now announce their wedding not just once but multiple times, with a ‘marriage proposal’ ceremony, a party with friends, an announcement to each family, or perhaps a relatively simple shared family party organized increasingly in the future spouses’ own home (Maillochon, 2016). The word fiançailles can designate one or other of these events, and sometimes all of them.
Change in frequency of rituals by respondents’ age and their conjugal and marriage history
Change in frequency of rituals by respondents’ age and their conjugal and marriage historyInterpretation: The proportion of marriages preceded by an engagement fell by 29% between 1974 and 2013 and by 44% among people marrying over age 30. By contrast, the proportion increased by 200% among those marrying for a second time and by 35% among couples who had a civil marriage only.
Coverage: Marriages between 1974 and 2013 (n = 3,159); marriages between 1984 and 2013 (n = 2,436) for bachelor/bachelorette parties which were too infrequent before 1984.
34The EPIC questionnaire does not provide enough detail to study premarital rituals and their multiple forms. However, the data do show how the relational structure of the engagement celebration has evolved.  In his analysis of the EPIC survey, Rault (2018) defines three types of engagement, each chosen with equal frequency in the marriages celebrated since 1996: ‘private’ engagements limited to the couple itself and with no special celebration; ‘family’ engagements; and ‘open’ engagements celebrated with friends. Describing the evolution of these different forms of celebration sheds light on the semantic difficulties still posed by the term. While the family-only celebration seems to be losing ground (68% of engagements in the years 1964–1973 versus 27% in 2004–2013), engagement is still very much a family affair, with family members included alongside the couple’s friends in ‘open’ engagement parties (Figure 4), the modal form of engagement over the period 2004–2013 (35% of cases). However, there may be cases of multiple celebrations in which the family-only (or other) type of party is considered secondary to the party with friends in respondents’ answers to the EPIC questionnaire.
Type of engagement celebration by date of marriage
Type of engagement celebration by date of marriageCoverage: Marriages between 1974 and 2013 that included an engagement celebration (n = 1,241).
35The share of private engagements—celebrated ‘by the couple alone’ as specified in the questionnaire—has increased fourfold in 40 years, from 6% in 1964–1983 to 22% in 2004–2013. They have become very popular among young people. They concern twice as many under-30s as older people and appear to comply with the new canons of the ‘marriage proposal’. Such proposals are no longer addressed to the bride’s family, as was obligatorily the case in France until 1933 (Hongre de Verdilhac, 2007), but follow an American model (Schweingruber et al., 2004) which has spread via social media. They are now a conjugal event, often in a romantic setting, in which the man officially asks his partner to marry him. This new version of the ‘marriage proposal’ releases the bride from the male domination of her father, while transferring it, symbolically at least, to her future husband. Women who ask for their partner’s hand in marriage are still seen as counterexamples, indirectly exposing the gender norms subjacent to this new practice. It is a practice that confirms the gendered division of conjugal roles, in contrast with the egalitarian ideals defended by most married couples. While engagement has become a ritual of commitment to the future wedding, not to the future couple, it still complies with the hierarchy of class (Rault, 2018) and gender relations that underpinned its traditional form.
3 – Inventing new traditions: bachelor and bachelorette parties
36Bachelor and bachelorette parties (or stag and hen parties in the United Kingdom) are a recent phenomenon in France. Studies by Segalen (2005) and by Monjaret and Pugeault (2012) were the first to describe these festivities at the turn of the century. The EPIC survey, which provides entirely new statistics on the question (Table 1), shows that these events have gradually gained in popularity over the last 50 years. They were rare in the 1960s (8% of weddings between 1964 and 1973) and mainly concerned men but gradually spread to women and are now equally popular among both sexes (43% of marriages concluded between 2004 and 2013). Bachelor and bachelorette parties involve a wide variety of activities, ranging from simple pyjama parties to workshops (kitesurfing, ikebana, cupcake making, etc.), local pub crawls, or a weekend in a European capital. Unlike engagement, there is no semantic ambiguity here, despite the wide range of practices covered by the term. The French term enterrement de vie de célibataire (literally “burial of singlehood”) even has its own abbreviated forms, EVG for men and EVJF for women, that have become normalized through their use by social media and commercial party-planning websites.
37Bachelor and bachelorette parties are especially popular among the youngest couples, notably those who marry between ages 25 and 30 (Table 1 and Figure 3). It is also in this age group that their growth has been most spectacular, with a 261% increase between 1984 and 2013 (Figure 3), to the point where bachelor and bachelorette parties are now seen by these generations as a ‘tradition’ that necessarily precedes the wedding.
38The growth in popularity of bachelor and bachelorette parties in France has occurred in parallel with the increase in premarital cohabitation. They no longer mark the transition from singlehood to married life (or ‘a last goodbye’ to freedom) but from an informal couple to an institutionalized one. In fact, couples who have already lived together hold bachelor and bachelorette parties much more often than those who marry directly (55% of cohabiters versus 24% of non-cohabiters for first marriages in 2004–2013). The party serves to mark the end of singlehood as a civil status rather than as a living arrangement. It is not a preconjugal ritual (a commitment to life as a couple) but a prenuptial one (an announcement of the future wedding). Its growth in popularity mirrors the increasing opulence of the other dimensions of contemporary weddings, of which it forms a sort of initiatory ritual. Bachelor and bachelorette parties bring together friends who may be from different social circles, enabling them to meet each other, become acquainted, and even make shared preparations for the big day.
39Monjaret and Pugeault (2012) demonstrate the three-pronged family, generational, and gendered order upon which these festivities are founded. While women’s bachelorette parties are now just as frequent as men’s bachelor parties, the two events are highly gender-specific and even sexist in nature, as confirmed by the observation of such parties from 1995 to 2014 (Maillochon, 2016). They generally take place independently for the man and woman, and in single-sex groups. The activities are organized around gender stereotypes, celebrating supposedly masculine and feminine characteristics, and the corresponding social roles: spa, beauty treatments, cookery classes, and cake tasting to symbolize women’s femininity, beauty, and love of food; bungee jumping, go-karting, and wine tasting to test men’s bravery, strength, and skill. Both sexes may indulge in heavy drinking and drug taking, and in activities with sexual undertones (chatting up strangers, strip-tease shows, etc.) that are more frequent and more extreme among groups of men than women. In all these respects, the bachelor and bachelorette parties popularized by the young generations constitute a rite of conformity not only in conjugal terms (already enacted in most cases) but also in marital terms, based on the gender division of domestic labour. Their festive exterior conceals a powerful injunction to comply with the gender order.
40Marriage no longer marks the birth of the couple but its potential for fulfilment. Yet the accompanying ritual, now bereft of meaning, has by no means lost its lustre; on the contrary, it has thrived and strengthened over the last 30 years. The EPIC data reveal the extent to which the components lending grandeur and panache to wedding festivities (large wedding party, wedding dress, etc.) are all alive and well or have regained new vigour after waning in the 1970s and 1980s. While premarital cohabitation led to the demise of the ‘traditional couple’ in the 1980s, it only partially and temporarily challenged the ‘traditional ritual’ (family engagement ceremony and religious sacrament), reworking the model while maintaining its trappings of splendour (engagement party with friends, marriage proposal, secular ceremony, etc.). By complying with this norm, cohabiting couples have revitalized the marriage rituals that were contested by their elders in the previous generation. Couples with children and divorcees, who once married more simply and discreetly, still have less lavish weddings than the others, but it is among this group that marriage practices have changed most radically. Far from challenging the rituals from which they were previously excluded, the less traditional couples are driving a resurgence of elaborate weddings, the strength of which reflects their increasing relative weight in the marriage structure.
41The hypothesis that ‘traditional marriage’ (religious ceremony to initiate life as a couple) has disappeared over the last 50 years is thus contradicted by the EPIC data. While the ‘traditional couple’ (who married directly without first living together) was challenged from the start of the period, the changes in the marriage ritual observed over the last 30 years point to a counter-revolution. The weakening hold of family and religion has not eliminated the other social constraints imposed upon the couple—especially economic constraints but also gender norms—that still influence marriage rituals. The growing popularity of large receptions that must be perfect in every way, the revival or emergence of customs founded on gender roles within the couple (marriage proposal by the man to the woman, sexual dimorphism of wedding clothes, sexism of bachelor and bachelorette parties) confirm that the wedding ritual is a powerful reminder of the economic and gendered order for all those involved.
Appendix A. Marriage data in the EPIC survey
42The EPIC survey of individual and conjugal trajectories (Étude des parcours individuels et conjugaux, 2013–2014) was conducted in 2013–2014 by the French Institute for Demographic Studies (INED) and the National Institute of Statistics and Economic Studies (INSEE) and headed by Arnaud Régnier-Loilier and Wilfried Rault.  It provides an update of knowledge on conjugal life in France and offers insights into the changes that have occurred by age and by cohort. To ensure that the survey was representative of the population aged 26–65 in metropolitan France, 7,825 people (from a sample of almost 14,500 dwellings drawn at random from the 2012 census survey) answered a questionnaire administered by INSEE interviewers.
43The EPIC survey gathered detailed information on the marriage ceremonies of people in an intact marriage at the time of the survey, so these data concern ‘surviving’ couples only. Among the 5,607 respondents in a union at the time of the survey, 3,533 were married and answered questions on the context, organization, and events of their marriage. The following topics were covered, in order (those analysed in this article are in italics): reasons for marriage; engagement celebration and types of guests; civil wedding (day, venue); religious wedding (day, venue, reasons for not marrying religiously, reasons for marrying religiously); wedding reception (day, number of guests, types of guests); wedding organization and payment; wedding clothes of bride and groom.
Appendix B. Analysing the history of marriage via data from the Couple Formation survey (1983–1984) and the EPIC survey (2013–2014)
44Thirty years after INED’s Couple Formation survey, the EPIC survey repeated some of the same questions posed in 1983–1984, thus extending the data series on marriages celebrated between 1960 and 1983 to the period 1964–2013. In these two studies, only the weddings of couples that were intact at the time of the survey are described. This methodological choice was made in order to obtain survey data that were representative of contemporary couples and not simply of wedding rituals. The questionnaires thus gave priority to an exhaustive description of the respondent’s relationship at the time of the survey and its officialization, if any, rather than his/her first relationship (and first officialization, which would have led to the omission of second marriages). The data obtained were thus representative of intact marriages at the time of the surveys but with a lower probability of including short-lived unions. For the older generations, we thus observe the most long-lasting unions, and the impact of this bias must be taken into account when interpreting the results. How can we take advantage of the data in these surveys, unique in terms of both their structure and historical depth, without overlooking their biases, liable to distort the analysis of change? Two procedures are used to appreciate the distortion between the entire set of marriages concluded over the period 1964–2013 and the ‘surviving’ marriages at the time of the EPIC survey.
Coverage bias of weddings in EPIC: an age-based approach
45For each marriage cohort, the mean age at marriage of EPIC respondents was compared with the mean age of people who registered a marriage (exhaustive data) over the same period, the only variable common to both observation methods. Comparisons of the means were made at the p = 0.05 level.
46The age at first marriage of EPIC respondents who married between 1984 and 2013 is no different from the national average (Table B.1); there is no obvious selection bias for these unions. We can thus assume that the first marriages between 1984 and 2013 recorded in EPIC are fairly representative of all first marriages concluded over the period.
47For first marriages before 1984, the mean age of EPIC respondents is below the national mean age, so they may be affected by selection bias. This bias is probably structural and stems from the survey scope (respondents aged 26–65); only the youngest people who married in the 1960s and 1970s were interviewed in EPIC. The survey could not include older couples who married in the 1960s and 1970s (whose numbers were small at the time) and who would have been older than 65 in 2013–2014. For this reason, this article relies mainly on less biased data from the FC study to describe older unions.
Comparison of ages at marriage in EPIC and civil records by sex and marriage order of the individual
Comparison of ages at marriage in EPIC and civil records by sex and marriage order of the individualNote: The differences significant at p < 0.05 are shown in bold.
Coverage: Individuals married between 1964 and 2013 whose union is still intact (EPIC); all weddings celebrated in metropolitan France between 1964 and 2013 (civil records).
48From 1984, the remarried EPIC respondents have a higher mean age than the national cohort of remarriages. The over-selection of late second marriages in EPIC is difficult to interpret. It concerns a small number of cases only briefly mentioned in the article.
Consequences of selection bias on the variables of interest
49Nothing proves that the wedding rituals of the most lasting marriages were any different from those of more short-lived ones. To nurture the discussion, the weddings of the two oldest marriage cohorts described in EPIC are compared with those of the same two cohorts surveyed 30 years earlier in FC. We assume that due to the more recent scope of the FC survey (23 years at most versus 40 years for EPIC) and the lower divorce rate in 1983 than in 2013, the cross-sectional information is less biased in FC than in EPIC.
50The descriptions of first marriages of never-married individuals between 1964 and 1983 in FC and in EPIC are quite similar (Table B.2). The numbers of guests and the frequency of religious marriage are identical in both. Engagement appears to be relatively less frequent among intact marriages in 2013, perhaps because traditional marriages are more lasting, although this hypothesis is not really supported by the figures for religious marriage. These differences may also stem from the different sampling protocols used in the two surveys: a quota method was used for FC and random sampling for EPIC.
51We do not have sufficient evidence to confirm a clear relation between marriage stability and wedding rituals. The approximation whereby the cross-sectional EPIC data over the period 1974–2013 and FC over the previous period are interpreted in historical terms is, in our view, acceptable, with all the necessary reserves.
First marriages celebrated over the period 1960–1983, FC vs EPIC
First marriages celebrated over the period 1960–1983, FC vs EPIC* 1964–1972.
** Word-for-word comparison of the reasons for marrying religiously is not possible. The FC survey only gave three response categories (other than don’t know and other): religious or moral convictions; to have a beautiful ceremony; and to make the family happy. The EPIC survey removed the ‘moral’ dimension and added an additional category (out of tradition) which was visibly chosen by those who married for reasons of family conformism and ‘moral’ conviction.
Coverage: First marriage still intact at time of survey (1983–1984 for FC and 2013–2014 for EPIC).
Calculated using data from the INSEE website: https://www.insee.fr/fr/statistiques/2546230?sommaire=2546239&q=mariage.
Available at https://www.insee.fr/fr/statistiques/2381500.
The mean age at first marriage in metropolitan France was 24.7 years for men and 22.6 years for women in 1970, and 33.0 years and 31.4 years, respectively, in 2017. INSEE’s 2018 Demographic Balance Sheet is available at: https://www.insee.fr/en/statistiques/2382603?sommaire=2382613.
INSEE, France, Portrait social 2017, p. 161; Couples et familles, 2015, p. 89.
The reasons for marriage are not mutually exclusive in the EPIC questionnaire, so the sum of responses may exceed 100%.
For INSEE data on non-marital births, see: https://www.insee.fr/fr/statistiques/2381394.
In 2013–2014, 58.9% of EPIC respondents reported being close to the Catholic faith, 8.2% to Islam, 1.9% to Protestantism, 0.6% to Buddhism, 0.2% to Hinduism, 0.5% to Judaism, 1.4% to another religion, 26.5% to no religion, and 1.6% did not answer. The orders of magnitude are similar to those observed in the 2005 ERFI–CGS survey (Régnier-Loilier and Prioux, 2009) and, for Catholics, are consistent with the estimates obtained from opinion polls for Bayard and La Croix (Raison du Cleuziou, 2014).
Statistics are available on the episcopate website: https://eglise.catholique.fr/conference-deseveques-de-france/guide-de-leglise/leglise-catholique-en-france-et-en-chiffres/371402-statistiquesde-leglise-catholique-en-france-guide-2017/.
In France, civil marriage is the only form of marriage recognized by law; a religious marriage must be preceded or followed by a civil marriage.
Calculations are based on statistics of the Conférence des évêques and vital statistics data (INSEE).
The EPIC questionnaire asks each respondent about the strength of his/her (and his/her partner’s) attachment to religion.
The EPIC figures are similar to those of the FC survey: 91% of weddings between two never-married partners were religious in 1960–1972 and 83% in 1973–1983 (Appendix B).
The EPIC questionnaire records the religion of the respondents and their partner, but not the type of religious wedding celebrated, so for couples who are non-believers or who have different religions, we do not know under what religion they married. In what follows, we thus refer to marriage of Catholics and not Catholic marriage, marriage of Muslims and not Muslim marriage, etc.
While not strictly accurate, the term ‘non-believer’ is used, for the sake of simplicity, when respondents report not being close to any religion or refuse to answer (28.1% of married respondents).
The EPIC results for couples who married between 1973 and 1983 and whose union was still intact in 2013–2014 are very similar to those obtained in 1983–1984 in the FC survey (Table 2).
Guest numbers may vary at the various wedding events (pre-reception champagne, wedding breakfast, informal lunch the following day, etc.), and depending on the length of festivities (the civil wedding may take place on a different day from the religious wedding and reception). The EPIC questionnaire asks respondents to estimate the numbers of guests at most of their wedding event(s).
In this article, a ‘wedding dress’ is understood as a purpose-designed outfit, usually a long white gown.
The universality of the white dress in France, and across the world more generally, is linked partly to the development of an international wedding market (Howard, 2006; Ingraham, 2008).
Or at least the most important celebration, from the couple’s viewpoint, when there are several.
See also the survey description in this issue of Population (Rault and Régnier-Loilier, 2019).