During the Middle Ages emerged, in western Europe, what John Hajnal (1965) has called the Western European Marriage Pattern. This model was characterized by an increase in the age at when people first married, as well as an increase in the rate of celibacy, especially female celibacy, which increased from 1% to more than 10% (for a comparison of European and Asian locations in the pre-transitional period, see Dribe et al., 2014). Authors such as Malthus (1798–1992) have indicated these mechanisms (increase in the age of marriage and in the rate of celibacy) as among the reasons for the decrease of fertility and mortality throughout Europe. The increase in the age of marriage, as well as the high proportion of those remaining unmarried, would have effectively reduced the fertility rate. Thus, the overpopulation that led to violent outbreaks of disease associated with overcrowding became avoidable. The increase in numbers of those who remained single did not arise from any social pact for controlling population excesses, but rather – comparable to the age of marriage – was determined by a set of social norms that governed the marriage market (Dribe and Lundh, 2014). These social norms were related to factors such as social and family position, health, the environment and social respectability. Although we cannot measure all these determinants, we will approach them through different indicators (individual, familial, economic, epidemiological, nutritional and physical). Thus, the aim of this study is to define the determinants of celibacy at the age of 45 for both men and women born between 1785 and 1965. By this method we will attempt to answer the question of whether rural societies shared the same determinants of celibacy as other areas of the European continent. As the bibliography shows, Spain did not fully adopt the European Marriage Pattern, although the average age at first marriage and the rate of remaining single during these centuries did increase slightly (see Reher, 1991). This article aims to provide new evidence on remaining single in Spain, and to analyze those factors that gave rise to this increase.

We will use the terms the single state and celibacy without distinction. When using them we will be referring to reaching the age of 45 without having married, since the possibilities for people of this age to start a family and to bear children will have been considerably reduced. We will examine the determinants of the single state in nine rural villages in the province of Zaragoza. Based on available data, we will consider only the determinants of celibacy in native individuals who did not migrate. Devos et al. (2016), in a recent publication, have highlighted the paucity of existing work on the single state and its determinants. This shortage of research is even more palpable for the Spanish case.

Some of the variables used in this study are being applied for the first time in the study of celibacy, while others have been used in different geographical or temporal contexts. Among these variables is male height in the twentieth century. They serve as a high-quality proxy for health and nutritional status. In this context, the Demographic Transition that began in the area in the early twentieth century assumed a revolution at the family and economic levels (Marco-Gracia, in press). Because of this, the analysis has been carried out for two separate periods: pre-transitional (1785–1899), and transitional (1900–1965).

The study area includes nine rural villages in the northeast of Spain: Alfamén, Botorrita, Jaulín, Longares, Mezalocha, Mozota, Muel, Tosos and Villanueva de Huerva (see Fig. 1). In municipal terms they occupy an area of 500km2 around the middle valley of the Huerva River. The database ‘Alfamén & Middle Huerva Database’ (AMHDB) was constructed based on individual information from the parish records of baptisms, marriages and deaths for the period between 1473 and 1950. The archives for this period are of a very high quality. The central inherent problem of the Spanish Catholic archives comprises the under-registration of the birth of females; in this period (1785–1965) the births of 106.7 boys per 100 girls are registered. For the period from 1951 until 2012 the database was completed by means of individual interviews centred on the same information guidelines. Individual data were pooled following the family reconstruction method developed by Fleury and Henry (1956), and included 95,817 individuals. This area had a total of 5,486 inhabitants in 1780, 5,884 in 1860, 8,086 in 1950 and 5,444 in 2012. Rural–urban migrations have seriously affected some municipalities during the course of the twentieth century.

Fig. 1.

Study area ubication.

(0,07MB).

Source: Own elaboration.

The study area is located on the Ebro Valley at the foot of the Algairén Mountains. The distance to Zaragoza varies from 19.75 to 40km. Its inhabitants are traditionally agricultural, managing chiefly cereal fields and vineyards. The area also has sheep grazing pasture. Data on occupations and literacy have been extracted from population censuses (1857, 1860), electoral rolls (1890, 1894, 1900, 1910, 1920, 1930, 1934, 1945, 1951 and 1955), the population register of 1824, and more rarely, from church registers and interviews. Information on professions and literacy was linked to the life patterns for each individual. Professions have been categorized into five groups. The first group, ‘day-labourers’, includes landless agricultural labourers, small landowners and other workers, such as unskilled labourers. Ordinarily, day-labourers combined agricultural tasks with small jobs in times of lesser agricultural activity. The second category, ‘farmers,’ includes all landowners who could support themselves exclusively from their farms. The third group, ‘shepherds’, includes all herder of sheep and goats, regardless of whether or not they are owners of livestock. The fourth group, ‘artisans’, consists of non-agricultural workers engaged in manual production (potters, bakers, blacksmiths, glaziers, etc.). The fifth category, ‘highly qualified’, includes non-manual professions of high social prestige in rural areas (doctors, veterinarians, teachers, bankers, etc.), all of which require a high level of education. The rest of the professions (military, public servants, drivers, etc.) have been included in the ‘other’ category, along with individuals who we do not aware of their occupation.

In a large part of Aragon, including this area, the inheritance was divided equally between sons and daughters. This transmission of family wealth was made at the time the descendant contracted their marriage, through the dowry, or at the end of their life, through the will.

For the analysis of the determinants of celibacy, we have added to the database two complementary variables from the military conscription details of each man: height adjusted according to age, and reasons for not entering military service.1 The analysis only includes those native individuals whose birth dates are known. We calculate the time in days from the measurement to attaining 21 years of age. Sanz-Gimeno and Ramiro-Fariñas, 2002 indicate that growth rates can continue up until 21 years, when the maximum height is reached. In order to adjust the height of all individuals to this maximum, the recorded height was compensated for by adding 0.0273785mm for each day until the age of 21 (an adjustment of 1cm per year). This is clearly an imperfect adjustment, since different people grow at different rates depending on their genetic and biological conditions. No compensation was made if the individual was measured after the 21 years. Extreme cases of less than 1.300m and greater than 2m were withdrawn. After adjusting the height of each individual, the adjusted height was compared with the average height of those born during the same quinquennial (beginning in 1900–1904 and ending in the extended period 1960–1965). From this comparison, the distance in millimetres from the five-year average was obtained by using the methodology proposed by Reher and Ortega (2004). These distances were arranged into five groups, with a relatively homogeneous number of cases: 1 – very low, more than 45mm below average; 2 – low, between 45mm and 15 below average; 3 – common, between 15mm below average and 8 above; 4 – high, between 8mm and 35 above the average; and 5 – very high, more than 35mm above the average. Regarding the reasons for not entering military service, these were grouped into 7 categories: 1 – temporary physical illness or problems, but of sufficient gravity to be grounds for exemption; 2 – moderate physical problems (such as sight or hearing problems); 3 – severe physical problems (such as blindness or serious physical incapacity); 4 – physical problems poorly defined; 5 – family poverty (mainly due to incapacity of the father or the widow of the mother); 6 – having brothers serving in the army; 7 – no motive put forward. These categories follow the proposal of Ayuda and Puche (2014) for the Valencian records.

In order to study the influence of the economic context during the years of birth and the following years, a variable has been included on the fluctuation of wheat prices in the city of Zaragoza for the pre-transitional period (the series has been extracted from: Peiró, 1987). We have taken the development of prices, once the trend of the wheat prices was removed through a logarithmic basis. We have considered the possibility of a subsistence crisis when prices were at least 10% higher than the average for the period.2 This criterion has been used in many international articles, as it is a threshold that affects the standard of living (for more information on its justification and use, see Bengtsson et al., 2004). We wish to examine whether having been born in the year of a subsistence crisis (or one or two years before it) affected the chances of individuals marrying before the age of 45.

We can observe in Fig. 2 that the rates of remaining single for men and women over 45 years have been similar from the end of the eighteenth century to the quinquennium 1860–1864. From that moment onwards, the male rates began to differ from the female rates, although it was not until the quinquennium 1915–1920 when the graph curves for men and women separated indefinitely. This trend is explained to a large extent by the rates of female migration, which have been higher than the male rates since the second half of the nineteenth century. Cities had become centres of population attraction, and some of these families required maids, mainly from the surrounding villages. Thus, while in Spain as a whole the rate for females remaining single was higher than that for men (Reher, 1991: 15), in those municipalities analyzed we find the opposite effect. During the 1970s the greatest rate of female outmigration was heralded by the arrival of new ideas linked to the Second Demographic Transition, which not only changed attitudes towards marriage and the single state, but also encouraged females to take up new urban jobs. The figures for those remaining single in the study area during the nineteenth century were low, close to the Spanish average of around 7.3% for men and 10.9% for women in 1887 (Reher, 1991), and below that of other European locations between 1716 and 1913, such as Casalguidi in Italy with 17.1% men and 10.3% women, Sart in Belgium with 18.2% men and 14% women, or some localities in the south of Sweden, with 12.8% men and 20.5% women (Dribe et al., 2014).

Fig. 2.

Percentage of singles who reach 45 years, according with quinquennial of birth (1785–1964).

(0,18MB).

Source: AMHDB.

The raw material for this study comprises unmarried people over 45-years-old, born between 1785 and 1965, and who did not dedicate their adult life to a religious vocation. The selection criteria, only individuals over 45-years-old, could have underestimated the coefficients of some of the variables by reducing the number of individuals in the sample. Among those individuals, and in order to identify celibates we take all those individuals whose date of birth is known to us, namely the natives. We must also consider that they either continued to reside in one of the sample villages or, if deceased, that their dates of death are known. We therefore omit all those who migrated. We know that day-labourers and the daughters of day-labourers migrated in higher numbers, since they were less tied to the land through land ownership (Marco-Gracia, 2017). Our interest is in knowing the marital status of natives at the age of 45, so that those who died before reaching that age are excluded from the sample. In order to know the effect of parental longevity on single people, given the relevant role played in the Netherlands (Kok and Mandemakers, 2016), if we simply do not know the exact age of death, then we accept the age recorded by the parish priest as an approximation. In order to complete the remaining variables on the family context (numbers of unmarried siblings, numbers of deceased siblings, parity, etc.) we need parents who have remained in the locality, and so we only take into account those individuals of whom we have information on their parents.

In this article we only analyze contextual, socio-economic and physical variables because other relevant variables are unavailable, such as those related to the character of individuals. It is also of great interest to understand how the context influences the potential of each individual to remain celibate. Our approach to these determinants will employ logistic regressions. We will take as a dependent variable the fact of being or not being single at 45 years of age. It requires a ‘zero’ value to have married, and a ‘one’ to remain celibate. In other words, the dependent variable analyses marital success. To better understand the effect of the different variables, we will group the regressions (in groups of six) into hierarchical models by gender and period. In each regression, we will introduce new variables to interact with those previously studied. Four groups of regressions have been developed: men 1785–1899 (regressions 1–6), women 1785–1899 (regressions 7–12), men 1900–1965 (regressions 13–18), and women 1900–1965 (regressions 19–24). The regressions include the following variables: Table 1.

We are going to examine heights and the reasons for avoiding military service separately because some authors have suggested that shorter individuals were more likely to be prone to diseases and other physical problems (e.g. Hacker, 2008; Sohn, 2015). However, as can be seen from Table 2, this relationship is not evident in this study area. Shorter individuals who reached 45 years old did not report any additional physical problems or serious illnesses than other groups. In fact, they were found to be average.

We will now analyze the most significant results of the hierarchical models. To do this we will divide the results by the year of birth, 1785–1899 or 1900–1965, depending on gender. We will focus on the variables that appear significant in different logistic regressions, locating similarities and differences between genders and periods. In the analysis of the results, and in each of the following subsections, a summary of the regression results with the most important and significant variables has been introduced (Tables 3–6). The complete regressions are located in the appendices. The bulk of the analysis refers to a society in which the only acceptable marriage was Catholic, heterosexual and monogamous, and to succeed in the local marriage market it was necessary to have an even ratio of the number of possible spouses (Abramitzky et al., 2011).

This analysis has been performed with individuals who remained in their hometowns. The results are consistent with those obtained for other parts of Europe, mentioned above. Among the individual characteristics two variables stand out: socio-economic status and female literacy. The results, although not significant, seem to show a greater tendency for farmers to marry, especially the daughters of farmers during the pre-transitional period, whose dowries included lands. This is possibly because their agrarian properties were an insurance against hunger in times of poverty. Literate women during the pre-transition period were the ones who most commonly remained single. Therefore, female literacy was not an essential factor among those men who sought a wife.

The variable for the number of unmarried siblings within the family context is highly significant in most regressions, regardless of gender or period, although its intensity is greater in the case of men and for the twentieth century. This variable concerns the existence of other unmarried siblings and could indicate two situations. The first is the difficulty for some families in enabling their adult sons and daughters to achieve success in the marriage market. Second, it could indicate the existence of a more favourable environment for remaining single, because unmarried brothers are a guarantee of company as well as an economic guarantee in low-income contexts. Regarding the presence of family members at home, for single men who were born during the pre-transition period (1785–1899) it was important to assess the potential for remaining celibate, because the longevity of the mother and the longevity of the father favoured bachelorhood. Similar results have been found for the Netherlands during the nineteenth century (Kok and Mandemakers, 2016).

The order of birth (parity) may also encourage or discourage the option of remaining single. During the pre-transition period, both men and women married less when occupying intermediate positions. Amongst men born in the twentieth century, this effect is not significant. However, being born last would have favoured the option of men remaining single during the twentieth century. During the pre-transition period women born into families that had experienced zero infant mortality were more likely to remain celibate. This could be linked to small dowries, because they could have many surviving siblings. During the twentieth century, women born into families with a high infant mortality rate or born into small families were the ones who most commonly remained single. Additionally, women who, during the Spanish Civil War, reached the average age for first time marriage remained single in a greater proportion, possibly because of the unexpected sex imbalance in the marriage market. Another variable concerning imbalances in the marriage market is the sex ratio. In men born during the entire period studied a greater proportion of male children favoured a higher rate of remaining single, especially during the pre-transitional period. In the case of women, the results are not significant for the pre-transitional period, and counterintuitive for the twentieth century. In the 1900–1965 period, high sex ratio rates were linked to higher rates of females remaining single, perhaps because the marriage markets were not so dependent on the local marriage market especially in recent decades. In any case, female rates for remaining single were much lower than the male rates throughout the twentieth century.

The results indicate that among men who were born during the pre-transitional period, the economic context of the year of birth had some influence on their further development. Men born in a year of very high living costs, potential subsistence crises, were linked to a lower ratio of marrying before reaching 45 years old. For men born during the twentieth century we have a reliable indicator for physical health problems: conscriptions linked to health problems. Serious health problems were also linked to a lower marriage ratio. Height has proven to be a useful indicator variable regarding physical appearance and the nutritional context. As in other European countries, men who deviated from the average height, both short and tall, remained celibate in greater proportion. In the case of women, and in the absence of individual data, we have estimated the height of parents and siblings. The daughters of tall fathers and the sisters of tall brothers, who presumably were taller than the average, were linked to remain single at 45 years old.

This article has allowed us to better understand the factors that determined remaining single in rural Spain in the very long-term using microdata for thousands of individuals. Therefore, we can conclude that multiple and analysable nutritional factors, familial, economic and personal variables affected the potential for men and women concerning marriage during the last centuries. The results have shown that there were differences depending on the period of birth of the individual and also according to gender.