Between Economy and Social Coercion: Nuptiality in Transition. The Case of the Don Army Territory (Southern Russia),1867

Abstract

The nuptiality transition in the Don Army Territory 1867–1916 depended on residence, economy, and religion. In rural areas, population growth and the scarcity of good land undermined the Orthodox tradition, though it was defended by the Church; rural Armenian-Gregorians persisted in early marriage, unlike Old Believers, Buddhists, and Lutherans. In cities, Armenian-Gregorians and Jews, unlike Catholics, adopted late marriage. Age at marriage among Orthodox was slightly later in cities and their hinterlands, against the tradition that Church conservatism favored. The different trajectories of nuptiality reveal a struggle between religion, economics, and urban life, making nuptiality a scene of contradictory influences.

Keywords

nuptiality transition age at marriage time series age difference between bride and groom religions

In most traditional societies, marriage was inextricably linked to economics and religion. On the one hand, religion could impose marriage schedules and doctrines on sexuality and fertility;¹ on the other hand, in rural areas, access to cultivable land conditioned the establishment of a couple and thus age at marriage.² This was eminently the case in the southern Russian governorate called the Don Army Territory (named after the fact that it provided an irregular army in the service of the Tsar; it was mostly agrarian with a few towns clustered in the south and turned to trade—see below—). Until the First World War, marriage was the central social institution that sanctioned procreation and governed inheritance and access to resources. Early marriage was made possible by the close involvement of parents in selecting partners for their offspring.³

For rural Orthodox, the decline in mortality associated with the hygiene transition, including notably water cleansing⁴ starting in the last quarter of the nineteenth century, resulted in an increase in marriageable youth. But the concomitant ecological crisis led to a shortage of good quality land, and therefore to difficulties in settling. At the same time, the powerful Orthodox Church was engaged in a conservative policy that, even without explicit regulation of age at marriage, favored the Cossack order, for which weddings were an important moment of social recognition and age at marriage was traditionally early. How did this system respond to the increasing total number of young people coming of age and demanding resources?

Our hypothesis is that the possible contradiction between economy and social coercion should be reflected in nuptiality, which combines intimate, religious, social, and economic dimensions. The longitudinal follow-up that we made possible for the Don Army Territory during the demographic transition at the turn of the twentieth century will inform us about the course of this tension.

In rural areas also resided Old Believers, who are schismatic Orthodox and who in the rest of Russia were ostracized at that time. In the Don Army Territory, they contributed to the irregular army as Cossacks, but by definition, were not subject to the authority of the Holy Synod. Geographically and anthropologically close, did their nuptiality resemble that of rural Orthodox, which would fuel the idea that rurality and denominational proximity would maintain early marriage? We shall see that this was not the case. Was the maintenance of early marriage then the prerogative of Orthodox? The case of rural Armenian-Gregorians will show that it was not. The explanatory picture of early marriage thus becomes not simple. In rural areas, there were also Lutherans, of German origin, who may have been torn between a Western pattern of later age of marriage and the possibility of aligning themselves with the pattern of earlier age of marriage practiced by their rural Orthodox neighbors.

As life in the city offered other opportunities than agricultural work, its economy could change the rules in favor of later marriage, against the resistance shaped by Church conservatism. The influence of religions in the city is also nuanced, which the cases of Catholics, Jews, Armenian-Gregorians, and to a lesser extent, Muslims will allow us to document. Starting from the question of the relationship of nuptiality to economy and religions, we shall find a relative diversity between and within the same denominations. We shall see that nuptiality, as expected, remained dependent on the total numbers of marriageable persons by age and sex that structured the matrimonial market. These numbers themselves depended on the mortality and fertility conditions that had recently prevailed for each group.

The intersection of geography, urban or rural residence, and religious obedience, controlling for the numbers of marriageable persons and for price fluctuations, will help us to evaluate the respective contributions of economic and religious factors. In particular, in the same cities, did denominations respond in the same way to urbanization and the transformation of production, or was there a religious component that could explain differences in marriage patterns?

The coexistence in this governorate of eight denominations, representative of those present in Russia at that time,⁵ its rurality and its cities, the diversity of its settlement—including peasants, Cossacks, Kalmyk Cossacks, Germans, Jews, Muslims, Armenians—allows for comparing nuptiality associated with different types of economy—the commons system, the city economy (centered on trade)—and very different denominations. This is a boon, coupled with the chance to have detailed data that have proven to be of high quality.

We thus document the transition of nuptiality, based on high-quality archives, for a region east of the St. Petersburg-Trieste line where marriage was both early and universal according to the matrimonial typology defined by Hajnal.⁶ Some studies of marriage patterns in Russia have focused on the early nineteenth century, prior to the abolition of serfdom in 1861;⁷ others have relied on the single pre-revolutionary census of 1897 for the whole of Russia;⁸ still, others have relied on disparate historical sources rather than on time series evidence.⁹

After having recalled the context and defended the quality of the data (section “Context and data”), recalled (classical) indicators of nuptiality, clarified their obtaining based on the population reconstruction,¹⁰ presented the time series of these indicators of nuptiality from 1867 to 1916 for the longest by denomination-district (section “Determinants”), we propose a vector auto-regression (VAR) linking the short-term fluctuations of these indicators to those of grain price, redemption payments, and total numbers of marriageable men and women, by denomination × geographic ensemble (sections “Vector Autoregressive Regression” and “Results”). These allow the specificity of each denomination-ensemble to be respected, and thus comparison between groups, which is what we are looking for. The regressions for each group are linked together, which will allow us to obtain better estimations of relationships and to assess geographic or denominational covariations between denominations-ensembles.

Nuptiality time series during the demographic transition in the Don Army Territory finally show that between denominational and economic constraints, marriage reflects a field of often contradictory forces (section “Discussion”). Economic necessity did not always prevail, and different patterns can be of the same obedience. From 1894 onwards, the delay in marriage made possible by the decline in mortality was adopted only by certain religious groups. Nuptiality trajectories then draw a landscape going towards more contrast between conservatism and new practices of later marriage or more definitive celibacy. By addressing groups residing side by side but with such diverse trajectories, this case study helps to show the subtlety and historicity of the nuptiality transition.

Context and Data

Population The Don Army Territory, a southern region (oblast) of European Russia (Figure 1), was characterized by sustained high fertility and a gradual mortality decline between 1867 and 1916. It was one of Russia's foremost grain-growing regions,¹¹ but agrarian and industrial resources barely kept pace with demand from a rapidly growing population.¹² While the population of Russia as a whole grew sharply between 1860 and 1914 (from 74 million to 164 million¹³), that of the Don Army Territory grew even faster, rising from 0.9 million to 3.4 million over the same period. Most of this growth came from the combination of high fertility, steady mortality decline, and substantial immigration in the wake of the 1868 law allowing outlanders to buy land and buildings. The cities of Rostov-on-Don and Taganrog were incorporated in 1888. In a region that originally was primarily rural,¹⁴ cities developed thanks to the presence of rich coal fields around the city of Alexandrov-Gruchevskii and to the growth of commerce and light industry connected with agriculture.¹⁵ These cities attracted a number of ethnic groups, each with its own religion, which made different models of nuptiality stand side by side.

Figure 1.

Cities and districts of the Don Territory in 1900. In the corner, the Don Army Territory (152,700 km² ) in 1900 European Russia. Source: M. Gilbert, The Dent Atlas of Russian History (2^nd edition, London 1993). Cities (indicated by dots) and districts of the Don Army Territory.

Denominations Most of the denominations present in Russia at the time were to be found in the Don Army Territory. Religion conditioned the social status of the populations and in particular the question of land and property rights. Although different denominations shared the same spatial setting, there was little contact between them.¹⁶ Most lived as self-governing communities. The case of the Don Army Territory is not exceptional: In the governorate of Ekaterinburg around 1900, Glavatskaya et al.¹⁷ on individual data also found few religious intermarriages. The largest denominational group remained the Orthodox (90% of the Don Army Territory's population), especially in the countryside. Jews, Muslims, and Catholics lived in urban areas. One district (Sal'skii) was populated predominantly by Buddhist Kalmyks. As an institution, the Orthodox Church was closely identified with the Russian monarchy and had achieved hegemony among the Don Cossacks.

The quality and detail of its statistics, notably marriage time-series by religion, district, and age of the spouses that begins in 1867 and ends at the 1917 Revolution allows the comparison of the economics and demographics of marriage across almost all denominations present in the Tsarist Empire. Marriages, deaths by age, births, and population counts were recorded by the Statistical Committee until the law of December 18, 1917, when the Soviet state took over.¹⁸ The population reconstruction¹⁹ by marital status is based on annual civil registration of deaths by age, births, marriages, and migrants, and the censuses of 1873 (which was conducted for the Don Army Territory only) and 1897, and this for each denomination-district. The initial age pyramid, fertility indices, migration rates, and life expectancies parameterizing Lederman life tables work as controls in a large-dimension stochastic minimization of the distance between reconstructed and recorded deaths by age, births, marriages, migrants, and censuses, where the dynamic is Lotka-McKendrick. We shall work with the reconstructed age distributions, where the fit for Lotka-McKendrick, by construction, corrects age heaping, if any, in the original data. The sensitivity analysis on simulated data shows satisfactory fit.²⁰ It allows for flawed or missing data, overcomes identifiability issues, and avoids replacing unknown quantities with assumed values.²¹ It finally yields nuptiality rates by sex, denomination, district, and city, from 1867 for the earliest districts to 1916 for the latest. The detailed statistics of marriage between the civil status of bride and groom (both single, single and widowed, both widowed) allow for estimating rates of first nuptiality.²² We do not have the room for detailing the reconstruction method, and it was done elsewhere at length.²³

During the period under consideration, the Don Army Territory was subdivided into districts (okrug) represented in Figure 1; the two districts of Rostovskii and Taganrogskii²⁴ were gained in 1888. Cities were identified in the 1897 census as Rostov-on-Don, Novocherkassk, Taganrog, Alexandrov-Gruchevski, and Nakhichevan’.²⁵ In rural areas, in addition to their own land allotments, which could be bought, sold, or transmitted by inheritance, Cossacks had proprietorial rights over all natural resources of their villages.²⁶ Most lived by working or letting their lands, but they could hold any position, including in the Church and commerce. The data we exploit cover all inhabitants, Cossacks and non-Cossacks, of the region. Cossacks were Orthodox, (Orthodox) Old Believers (and Coreligionists), or Buddhists, but we also consider non-Cossack Orthodox, and also exploit data on Jews, Catholics, Lutherans, Armenian-Gregorians, and Muslims, none of whom were Cossack.

Quality of the data Up to the 1917 Revolution, the Tsar was the head of the Russian Orthodox Church. Official practice was that all Russians with the exception of Old Believers and sectarians were automatically Orthodox from birth.²⁷ The Orthodox Church's responsibility for recording births, deaths, and marriages is stipulated in the Compendium of Laws of the Russian Empire (Svod Zakonov Rossiskoj imperii) Volume X, article № 25–33. The Church alone could register marriage, civil marriage being non-existent. The Church also had the task of recording statistics on vital events and transmitting them to the Statistical Committee (Compendium of Laws, Volume X, article № 28, 34, 35, 78). Questions of marriage and divorce were dealt with by religious courts (Compendium of Laws, Volume X, article № 24, 33, 38, 73). Russian statistics may have a poor reputation, but the double recording system, civilian (gathering church records) and military, and the two censuses, 1873 and 1897, instead of the single 1897 census for Russia, give first guarantees on the quality of the data from the Don Army Territory. The excellent adjustment of the reconstruction based on Lotka-McKendrick equations on a long enough period (1863–1916)²⁸ for groups whose birth and death series are coherent²⁹ and their level high enough brings further security about the quality of the data. Moreover, times series of fertility, mortality, migration, and nuptiality, reconstructed independently from a denomination-district to another, are coherent in terms of levels and fluctuations.

Measurements The appropriate measure of age at marriage that is virtually independent of the age structure is the singulate mean age at first marriage $SMAM = 15 + \frac{\sum_{a = 15}^{49} S_{a} - 35 S_{50}}{1 - S_{50}},$ where $S_{a}$ is the proportion of women remaining single at (yearly) age a in a fictitious cohort exposed to current nuptiality,³⁰ and $S_{50}$ is the proportion never marrying in that cohort at age 50.³¹ The SMAM captures the conditions of the current year in terms of nuptiality, just as period life expectancy or the total fertility rate capture the conditions of the current year in terms of mortality or fertility. By contrast, the mean female age at first marriage $A_{m} (t) = \frac{\sum_{a = 15}^{49} a p_{f}^{m} (t, a)}{\sum_{a = 15}^{49} p_{f}^{m} (t, a)}$ , where $p_{f}^{m} (t, a)$ is the total number of women first marrying at age a at year t, depends on the age structure of the population. The age difference between bride and groom gives a hint at man–woman differences. The dispersion of nuptiality over age is measured by the proportion of never-married women between the ages of 20 and 29. This measure is more interesting than the proportion of ever married women, whose fluctuations are of too small magnitude for rural groups. At last, Coale's standardized proportion married³² $I_{m} (t) = \frac{\sum_{a = 15}^{49} p_{f}^{m} (t, a) h (a)}{\sum_{a = 15}^{49} p_{f} (t, a) h (a)}$ measures the contribution of married women in a maximum fertility pattern, where $h (a)$ is the fertility rate at age a of the Hutterites and $p_{f} (t, a)$ the total number of women of age a.

Inter-Denominational Marriages

We do not have direct statistics on inter-denominational marriages. We do, however, have the distribution by district and sex from 1867 to 1916 of the numbers of conversions to Orthodoxy, without mention of age. These numbers give upper bounds of the numbers of conversions for marriage during the year of marriage. The ratios of these upper bounds to the numbers of marriageable by denomination for the whole Territory are very low: almost zero for Buddhists and Armenian-Gregorians, between 0.3% and 1.0% between 1891 and 1916 for Old Believers. The ratios of these upper bounds to the numbers of marriageable Orthodox range on average around 0.08% (maximum 1%) for men and 0.03% for women for Upper Don, nil for Lower Don, around 0.03% (maximum 0.5%) for men and 0.01% (maximum 0.2%) for women in cities. In any case, conversions are considered as migration in the reconstruction. Inter-denominational marriages therefore do not affect the calculation of marriage indicators.

Descriptive Analysis

To situate, Table 1 provides an overview of the diversity of nuptiality indicators by denomination-district around the year 1900. Coale³³ indicated $I_{m}$ = 0.55 as the value separating West-European from non West-European marriage patterns. This is firm evidence that the Don Army Territory belonged to Hajnal's eastern pattern of universal and early marriage, especially for Orthodox, Old Believers, Lutherans, and Buddhists, whose $I_{m}$ remained persistently over 0.55. However, the Don Army Territory is different from the rest of Russia, if only by the diversity of religions and its economy based on the commons (there is no single “Russian” marriage³⁴), and that is why it deserves a specific study.

Table 1.

Nuptiality indicators in the Don Army Territory, by district and religious denomination, around 1900 (mean 1897–1903).

District	Denomination	Life expectancy at birth, women	Singulate mean age at first marriage (women)	Standardized proportion married, women, $I_{m}$	Age difference bride-groom (S.D.)	Proportion of women never married at 20–29	Proportion of women marrying before 20	Proportion of men marrying before 20
First Don	Orthodox	35.7	18.9	0.81	2.2 (2.4)	0.10	0.74	0.52
First Don	Old Believer	38.4	24.0	0.70	2.1 (2.3)	0.11	0.78	0.48
Second Don	Orthodox	38.7	18.3	0.83	2.2 (2.3)	0.04	0.82	0.59
Donetzkii	Orthodox	37.9	18.1	0.86	2.1 (1.9)	0.03	0.82	0.56
	Old Believer	43.6	23.6	0.69	2.5 (2.8)	0.24	0.74	0.52
	Lutheran	47.5	17.7	0.56	3.9 (3.0)	0.03	0.70	0.28
Ust-Medevskii	Orthodox	37.9	18.7	0.83	2.2 (2.3)	0.08	0.84	0.63
Ust-Medevskii	Old Believer	35.5	18.2	0.78	2.6 (2.9)	0.24	0.74	0.52
Khopyorskii	Orthodox	35.2	17.8	0.84	2.0 (2.0)	0.03	0.87	0.67
Cherkasskii	Orthodox	36.5	19.7	0.75	4.1 (4.2)	0.13	0.68	0.33
Sal'skii	Buddhist	37.0	21.3	0.75	1.6 (1.9)	0.12	0.74	0.59
Rostovskii rural	Arm.-Greg.	35.5	19.3	0.60	7.4 (3.7)	0.12	0.78	0.07
	Orthodox	34.0	19.5	0.69	2.6 (3.2)	0.12	0.64	0.33
	Lutheran	44.5	19.8	0.87	1.6 (2.2)	0.03	0.66	0.30
Taganrogskii rural	Orthodox	34.2	19.1	0.71	3.3 (3.5)	0.10	0.74	0.35
Cities: Rostov-on-Don	Orthodox	24.3	21.1	0.63	5.1 (4.3)	0.20	0.44	0.08
	Catholic	39.9	19.3	0.68	5.5 (3.7)	0.12	0.55	0.15
	Jew	57.2	21.8	0.75	3.3 (3.1)	0.20	0.45	0.08
	Muslim	39.9	22.8	0.68	6.1 (3.3)	0.16	0.49	0.10
Nakhichevan’	Arm.-Greg.	47.6	21.2	0.59	7.9 (4.4)	0.21	0.61	0.03
Novocherkassk	Orthodox	33.1	23.1	0.63	4.2 (4.1)	0.30	0.41	0.09
Alex.-Gruch.	Orthodox	37.1	21.8	0.61	5.1 (4.0)	0.17	0.67	0.19
Taganrog	Orthodox	33.7	20.4	0.60	5.1 (4.2)	0.19	0.47	0.08
Taganrog	Jew	53.1	23.6	0.56	5.3 (4.8)	0.32	0.44	0.12

The column on the female life expectancy at birth gives a hint at mortality conditions and its variations across denominations: rural Lutherans and Jews had a life expectancy at birth substantially higher than other denominations; Orthodox and Old Believers had the lowest. Table 1 shows that women substantially married younger in the upper Don (districts of First and Second Don, Ust-Medevskii, Khopyorskii, Donetzkii) than in the rural lower Don (Cherkasskii, Taganrogskii, Rostovskii, Sal'skii) and in cities.

Figure 2 presents the times series of the SMAM, Figure 3 the proportion of women never married between 20 and 29, Figure 4 those of $I_{m}$ , and Figure 5 the mean age difference between bride and groom. For the sake of readability, we have regrouped districts into three ensembles for Orthodox: Lower Don, Upper Don, and cities. There is no reason why the short-term fluctuations of these time series should be smooth. Regular time series are the apanage of populous ensembles, such as France as a whole or England as a whole. Life expectancy, marriage, and fertility time series for a group of parishes in seventeenth century French Pays de Caux (around 6,000 people on average) look similarly noisy,³⁵ for example. One reviewer suggests converting the curves into trends. However, time variability matters, to avoid letting one believe that the past was smooth, which should lead to a smooth marriage market. This was precisely not the case and was indeed the problem (find a spouse of appropriate age among irregular age classes). Marriageable people had to cope with shortage in the opposite sex, which led to irregular fertility and to future distortions in the marriage market, which in turn led to further irregular fluctuations in nuptiality. Marriage was taking place under disrupted rather than smooth conditions, and this, we believe, was a marker of the times.

Figure 2.

Women's singulate mean age at first marriage, in years, by residence-denomination in the Don Army Territory.

Figure 3.

Proportion of never married women between 20 and 29 by residence-denomination in the Don Army Territory.

Figure 4.

Standardized proportion married $I_{m}$ by residence-denomination in the Don Army Territory.

Figure 5.

Mean age difference in years between bride and groom by residence-denomination in the Don Army Territory.

Orthodox and Old Believers The SMAM for Orthodox in the Don Army Territory over the period 1897–1903 is closer (18.6 years for rural Orthodox, 21.5 for urban Orthodox) to the levels of White Russia (22.7), Ukraine (20.8),³⁶ and in the Balkans (20.3 years in Romania, 20.8 in Bulgaria, 20.1 in Serbia)³⁷ than to those for England (26.2) or Sweden (27.5)³⁸ around 1900. Likewise, the proportions never married at 20–29 years of age are at the level of Bulgaria (0.14), Romania (0.14), and Serbia (0.09) (not computed for White Russia and Ukraine), in contrast to 0.67 in England and 0.58 in Sweden³⁹. Orthodox then tended to marry early,⁴⁰ as it has been estimated elsewhere in Russia at that period, for example in Ekaterinburg over the period 1907–1919, where the majority of a sample of 285 men and women married at the age of 19.⁴¹ Until 1894, Old Believer men did marry slightly younger (4 months on average) than Orthodox, but the age difference between bride and groom was smaller for Old Believers than for Orthodox, so that Old Believer women married at the same age as Orthodox women. After 1894, Old Believer men married slightly older (7 months on average) than Orthodox men and women only 3 months older than Orthodox women. Table 1 and Figure 5 also show that Old Believers and Orthodox in rural areas had approximately the same small age difference between bride and groom before 1894 (when Nicolas II came to the throne). Orthodox city-dwellers married later and had larger age differences between partners (Table 1). The dispersion around the SMAM (measured by the standard deviation shown in parentheses in Table 1) was also larger in cities. The proportions of men and women marrying under 20 are also consistent with the age difference of 4–6 years for urban Orthodox men, who married later than rural Orthodox, rural Old Believers, and rural Buddhists, for whom the age difference was closer to 2 years. The variance was comparatively high for urban Orthodox, which is consistent with continuous immigration of Orthodox from rural districts.

Buddhists Early marriage concerned Buddhists, who were authorized to practice polygamy, although we have no data on how much it was practiced. Most Buddhists lived as nomads in the steppes of the district of Sal'skii. They were 14,392 men and 13,651 women at the 1897 census. After the year 1853, some began growing corn and vegetables. Table 1 and Figures 2 to 5 show that Buddhist men were like rural Orthodox in having early marriage and the age difference between bride and groom of around 1 year. An exception is the high value of the SMAM from 1898 to 1912. The excessive rainfall followed by summer drought in 1898 ruined most of the Don Army Territory and Sal'skii in particular,⁴² where this catastrophe, by reducing the resources available for marrying, may be behind the abrupt upturn in the SMAM after that date.

Lutherans Lutherans were most numerous in the district of Taganrogskii, but also in Dontezkii and rural Rostovskii (14,436 men and 13,870 women in the entire Territory at the 1897 census). They lived in their specific milieu marked by central European influences.⁴³ They married young around 1900 but age at marriage increased rapidly after that date (Figure 2), as did age difference between bride and groom from 2 to 4 years (Figure 5).

Armenian-Gregorians (13,896 men and 13,222 women in the whole Don Army Territory at the 1897 census) resided either in rural Rostovskii, occupied primarily with grain culture, sheep breeding, and limited amounts of sericulture and viticulture, or in the city of Nakhichevan’, where they specialized in commerce (mainly of agricultural products including timber, hides, and skins, as well as bricks and sheet glass). Rural women married as early as rural Orthodox (Figure 2), though the age difference between bride and groom was larger for both rural and urban (Figure 5).

Catholics Contrasts along denominational lines persisted among city-dwellers. For Catholics, the age difference between bride and groom fluctuated between 1 and 8 years, perhaps because of the relatively small numbers involved (1,068 men, 768 women in Rostov-on-Don at the 1897 census). Roman Catholicism was relatively tolerant toward spousal age differences.⁴⁴ The Catholics for the whole Territory included Germans, but for Rostov-on-Don, which is the only geographical entity for which the data are of sufficient quality to allow reconstruction, the Catholics were mainly Poles.⁴⁵

Jews were 7,852 men and 8,126 women at the 1897 census, of whom 5,771 and 6,012 women were in Rostov-on-Don. They formed a community closed in upon itself,⁴⁶ whose culture and daily life were deeply imbued with religion. They still married early and maintained a 3–4 year age difference between bride and groom. Schellekens⁴⁷ found that early marriage was common among Ashkenazi Jews in Eastern Europe in the seventeenth and early eighteenth centuries on religious grounds (“early marriage was seen as a means to prevent the sin of premarital sex”). Livi-Bacci⁴⁸ saw Jews as the “forerunners” during the decline in marital fertility, though Schellekens,⁴⁹ studying Amsterdam in the period 1625–1724, contested such a role for Jews in the transition from early and universal marriage to the “West European” pattern of late marriage and high levels of permanent celibacy. For Jews, too, Figures 2 to 5 show that the years 1890–1910 capture the moment of nuptiality transition. The strikes and riots in Rostov-on-Don from 1902 up to the 1905 Revolution were contemporaneous with the onset of the transition, a parallelism that suggests political events had a role in triggering the emergence of a new pattern, one perhaps already latent but hitherto held in check by tradition.

Muslims Muslims were 928 men and 372 women in Rostov-on-Don at the 1897 census. The 1897 census also provides the distribution of languages crossed with the districts indicates that Tatar was the most spoken language in Rostov-on-Don after Russian, Yiddish, German, and Polish, and much more spoken than many other languages. These Muslims were thus probably Muslim Tatars. As was the case for Catholics, the large fluctuations in the time series in Figures 2 to 5 are due partly to the small number of available women. The scarcity of women may have been exacerbated by the practice of polygamy.⁵⁰ For Muslims in seventeenth-century Amsterdam, Schellekens and Eisenbach⁵¹ referred to the “nature of familial relationships and the segregated roles of women” and to the “traditional Muslim family considered to be strongly patrilineal and patrilocal, with male dominance and responsibility prescribed by the Quran.” Such attitudes are scarcely conducive to marital transition. Early marriage of women was maintained by the fact that parents were “especially anxious for their daughters to marry young” so as to preserve pre-marital chastity because “honor and esteem depend largely on the sexual conduct of women in the household.”⁵²

Overall then, Figures 2 to 5 show that not all groups were involved in the transition toward late marriage (Catholics in Rostov-on-Don and rural Armenians-Gregorians kept marrying early). The transition was slow for Orthodox, and the three Orthodox groups were at different levels. The other groups, some urban like Jews or Armenian-Gregorians, some rural like Buddhists, Lutherans, and Old Believers, embraced the transition quickly after 1894.

Determinants

Resources: Marriage was directly related to available resources. For Cossacks (who were Orthodox, Old Believers, or Buddhists), weddings cost “between 50 and 100 roubles more than the yearly surplus produced by the average Cossack farm.”⁵³ The classical argument is that improved survival of parents and siblings created a shortage of resources that pushed back the date for new family formation.⁵⁴

In the Model section below, we show why the first differences of the SMAM, $I_{m}$ , the proportion of women never married between 20 and 29, and the age difference may depend on the variations of grain price, redemption payments, and the total number of marriageable people (more than directly on those of life expectancy—famines, wars, revolutions were filtered by the pyramids of marriageable people, which in addition absorbed the ups and downs of past fertility, mortality, and migration). Age at marriage surely in turn influences the total numbers of marriageable people,⁵⁵ but only in the medium term, the time it takes for the children of these marriages to become marriageable. Thus in the short term, which is of interest to us here, the total numbers of marriageable children are approximately exogenous to current age at marriage. This allows us to posit the econometric model in Eq. (4) below in order to test the importance of total numbers of marriageable people, grain price, and redemption payments in the annual variations of nuptiality indicators. We tried lagged grain price, but it provides no significant coefficient.

We estimate the total number of first-time marriageable persons as the sum of first-time married of the year and the total number of singles in a range of years around age at first marriage for each sex. The estimation of Eq. (4) below gives the largest total number of significant coefficients for $a_{m}$ equal to 1 year for every group and each sex.

Grain price: Grain price can play a direct role by modifying people's capacity to pay for a wedding and start a new household. The fluctuations in the price of barley and oats on the market of Rostov-on-Don⁵⁶ are close to those plotted by Mironov⁵⁷ for Russia as a whole. The short-term fluctuations in the time series of the prices of barley, rye, wheat, buckwheat, and oats are strongly correlated with each other (correlations range between 0.69 and 0.86).

Steel price: Metal industry constituted another source of employment and salaries, but it did not develop until 1900, following the establishment of metallurgy at Taganrog in 1896 and the Sulinovsky metallurgy factory in Cherkasskii in 1900. In 1894, the Don produced 24.3% of Russian coal but less than 1% of Russian charcoal and 1.1% of steel.⁵⁸ The fluctuations in the price of charcoal and steel would thus be of secondary importance as a determinant of marriage behavior for the Don populations, the great majority of whom relied on agriculture, crafts, and commerce for their livelihood.

Land price: Peasants (mostly Orthodox but non-Cossacks) had to pay for the land allotments they had received after the abolition of serfdom in 1861. The level of redemption payments varied greatly between districts, due to the different proportions of peasants and the different conditions under which land allotments were made. As stipulated in the edict of 19 February 1861, for noble-owned serfs, and “for the remainder of the peasant population in 1863 and 1866, the peasants were to ‘redeem’ the land by repaying the government in instalments spread over a term of 49 years.”⁵⁹ Our data show that indeed redemption payments ended in 1910 for all districts. Redemption payments could cover many unobservable differences in land quality or about the prevalence of serfdom.

Literacy could be a key variable for explaining changes in marriage behavior, notably because people who could read and write were more likely to move to cities where couples married older. Unlike rural agricultural jobs, most city jobs required the ability to read and write.⁶⁰ No time series evidence is available, but the proportions able to read and write increased rapidly for all districts between the 1873 census and that of 1897: typically for the rural Don, in the First Don for example, the proportion for the 7–15 old increased from 3% to 18% for girls and from 13.5% to 41.8% for boys. For the 16 years old and over, these increases were from 2.1% to 9.4% and from 23.3% to 53.3%. In Rostov-on-Don at the 1897 census, for the 7–15 old, 52.4% of girls and 68.9% of boys could read and write. These proportions were 34.3% and 66.0% for the 16 years old and over. Literacy was higher among men than among women, one reason being that men benefited from the great army reform of 1874, which cut the length of military service from 25 to 6 years and offered considerable further reductions for those who had received education before beginning their service.⁶¹ Although simple correlations between variables which are distributed spatially and temporally may be misleading, because of temporal and spatial correlations, and although we do not have literacy data by denomination, the correlations between the SMAM, $I_{m}$ , and the proportions of boys and girls who could read and write over the districts in 1897 (12 observations) give a hint at a relationship between literacy and marriage for Orthodox (the only denomination represented everywhere). The correlation with the SMAM is 0.79 (sd = 0.002) for the proportion of girls aged 7–15 able to read and write and 0.84 (sd = 0.001) for the proportion aged 16 and over. The correlations of the standardized proportion $I_{m}$ of married women with the same variables are − 0.73 (0.01) and − 0.63 (0.03). So the rise in literacy, itself correlated with economic resources and urban residence, is likely to have favored the rise of age at first marriage over the period.

Spatial heterogeneity: The follow-up of the SMAM, $I_{m}$ , the proportion never married 20–29, and age difference between bride and groom across the Territory for the Orthodox by district differentiates the rural upper Don (with low SMAM and high $I_{m}$ ) from the rural lower Don (with higher and increasing SMAM and decreasing $I_{m}$ ). The southern districts of Cherkasskii, rural Rostovskii, and rural Taganrogskii occupied intermediate positions throughout the period, with values closer to those in cities than was the case in northern districts. This reflects the exchanges that took place between cities and their rural hinterlands, as well as the remoteness of the northern districts from urban influences.

Vector Autoregressive Regression

First-differences of nuptiality indicators are approximately linear to those of grain price and numbers of marriageable We treat denominations-ensembles rather than districts because the correlation matrices are smaller and more readable. The SMAM avoids the age structure effect, but the age-specific rates that comprise it are still dependent on the marriage market of the current year, through age preferences between men and women. In two-sided matching theory,⁶² the total number of marriages $X_{a b}$ between bride of age a and groom of age b is $X_{a b} = c_{a b} X_{a 0}^{f} X_{b 0}^{m}$ , where $X_{a 0}^{f}$ and $X_{b 0}^{m}$ are the total numbers of marriageable women of age a and men of age b, and $c_{a b}$ measures the preference between ages a and b. As parents in the case of the Don Army Territory were constrained in financing weddings, this equation becomes $X_{a b} = π_{a b} c_{a b} X_{a 0}^{f} X_{b 0}^{m}$ , where $π_{a b}$ is the probability to pay the expenses of the wedding. Replacing the proportion $S_{a} - S_{50}$ of people who will marry after age a by its expression in terms of numbers of marriages $X_{a .} := \sum_{b = 15}^{50} X_{a b}$ yields $SMAM = 15 + \sum_{a = 15}^{49} \sum_{i = a}^{50} X_{i .} / \sum_{15}^{50} X_{a .}$ . Population growth beginning at $t_{0}$ brings larger younger cohorts to marriage. After $t_{0} + 15$ , there exists an age $15 \leq \bar{a} \leq 50$ years for which $d X_{a 0}^{f} / d t > 0$ for $a \leq \bar{a}$ and $d X_{a 0}^{f} / d t = 0$ for $a > \bar{a}$ . If the variations are approximately independent of age, $\frac{Δ π_{a b}}{π_{a b}} \approx \frac{Δ π}{π}$ , $\frac{Δ c_{a b}}{c_{a b}} \approx \frac{Δ c}{c}$ , $\frac{Δ X_{a 0}^{f}}{X_{a 0}^{f}} \approx \frac{Δ X_{0}^{f}}{X_{0}^{f}}$ , $\frac{Δ X_{a 0}^{m}}{X_{a 0}^{m}} \approx \frac{Δ X_{0}^{m}}{X_{0}^{m}}$ , where $Δ$ is the first differentiation operator. By differentiation,

Δ SMAM (t) = \frac{\sum_{a = 15}^{\bar{a}} \sum_{i = a}^{\bar{a}} Δ X_{i .}}{\sum_{a = 15}^{50} X_{a .}} - SMAM (t) \frac{\sum_{a = 15}^{\bar{a}} Δ X_{a .}}{\sum_{a = 15}^{50} X_{a .}} \approx (\frac{Δ π}{π} + \frac{Δ c}{c} + \frac{Δ X_{0}^{f}}{X_{0}^{f}} + \frac{Δ X_{0}^{m}}{X_{0}^{m}}) (\frac{\sum_{a = 15}^{\bar{a}} \sum_{i = a}^{\bar{a}} X_{i .}}{\sum_{a = 15}^{50} X_{a .}} - SMAM (t) \frac{\sum_{a = 15}^{\bar{a}} X_{a .}}{\sum_{a = 15}^{50} X_{a .}}) .

(1)

The probability of paying the expenses of the wedding is likely related to wealth, whose proxies are grain price p and redemption payments r:

\frac{Δ π_{a b}}{π_{a b}} \propto d_{1} Δ p + d_{2} Δ r

, where

d_{1}

and

d_{2}

are coefficients. Grain sellers are likely to be favored by an increase of grain price, and conversely for pure buyers. The preference is related to price, as resources modify the choice of spouse and the timing of wedding, to total numbers of marriageable people, which condition choices, and to the probability of being able to pay for the wedding (people interiorize resource constraints):

\frac{Δ c}{c} = η_{1} Δ p + η_{2} Δ r + η_{3} \frac{Δ X_{0}^{f}}{X_{0}^{f}} + η_{4} \frac{Δ X_{0}^{m}}{X_{0}^{m}}

, so that

Δ (SMAM)

is approximatively a linear function of first-differenced grain price and of relative variations of total numbers of marriageable people, for a given denomination-ensemble.

Standardized proportion of married women Likewise, from the count of the dead and the newly married, the total number of married women changes with the probability of dying $_{1} q_{a} (t)$ at age a and the newly married $X_{a .} (t)$ as $p_{f}^{m} (t + 1, a + 1) = p_{f}^{m} (t, a) (1 -_{1} q_{a} (t)) + X_{a .} (t)$ , and the total number of women as $p_{f} (t + 1, a + 1) = p_{f} (t, a) (1 -_{1} q_{a} (t))$ . The variation of $I_{m}$ from 1 year to the next is:

Δ I_{m} (t) = \frac{X_{15, .} (t + 1) h (15) + \sum_{a = 15}^{49} (- p_{f}^{m} (t, a)_{1} q_{a} + X_{a + 1, .} (t + 1)) h (a)}{\sum_{a = 16}^{50} p_{f} (t, a) h (a)} - I_{m} (t) \frac{- \sum_{a = 15}^{49} p_{f} (t, a)_{1} q_{a} h (a)}{\sum_{a = 16}^{50} p_{f} (t, a) h (a)} .

(2)

At equilibrium,

Δ I_{m} = 0

, the distribution of marriages is

X_{a .}^{equil}

. Using

X_{a .} (t + 1) = X_{a .}^{equil} (t) + Δ X_{a .} (t)

and developing (2) around the equilibrium yields

Δ I_{m} = \sum_{a = 15}^{\bar{a}} Δ X_{a .} h (a) / \sum_{a = 16}^{50} p_{f} (t, a) h (a)

. Developing

Δ X_{a .}

by differentiating

X_{a .} := \sum_{b = 15}^{50} X_{a b}

and using

X_{a b} = π_{a b} c_{a b} X_{a 0}^{f} X_{b 0}^{m}

yields

Δ I_{m}

as a linear function of

Δ p

Δ r

\frac{Δ X_{a 0}^{f}}{X_{a 0}^{f}}

, and

\frac{Δ X_{b 0}^{m}}{X_{b 0}^{m}}

, then, by averaging, as a linear function of

Δ p

Δ r

\frac{Δ X_{0}^{f}}{X_{0}^{f}}

, and

\frac{Δ X_{0}^{m}}{X_{0}^{m}}

Proportion of never-married women 20 to 29 Formally, the proportion of never married women between 20 and 29 years of age is obtained from the formula of $I_{m}$ by putting h(a) = 1 and replacing the range of variation from 15 to 50 by 20 to 29. This proportion therefore varies as a linear function of $Δ p$ , $Δ r$ , $\frac{Δ X_{0}^{f}}{X_{0}^{f}}$ , and $\frac{Δ X_{0}^{m}}{X_{0}^{m}}$ .

Age difference at marriage With $X ._{b} := \sum_{a = 15}^{80} X_{a b}$ the total number of men marrying at age b, the age difference between bride and groom is $A_{m} - A_{f} = \sum_{b = 15}^{80} b X_{. b} / \sum_{b = 15}^{80} X_{. b} - \sum_{a = 15}^{80} a X_{a .} / \sum_{a = 15}^{80} X_{a .}$ . By differentiation:

Δ (A_{m} - A_{f}) (t) = (1 - A_{m} (t)) \frac{\sum_{b = 15}^{\bar{b}} b Δ X_{. b}}{\sum_{b = 15}^{80} X_{. b}} - (1 - A_{f} (t)) \frac{\sum_{a = 15}^{\bar{a}} a Δ X_{a .}}{\sum_{a = 15}^{80} X_{a .}},

(3)

where

15 \leq \bar{b} \leq 80

is an age for which

d X_{b 0}^{m} / d t > 0

for

b \leq \bar{b}

and

d X_{b 0}^{m} / d t = 0

for

b > \bar{b}

, expressing the arrival of larger younger birth cohorts associated with population growth. As previously, by differentiating

X_{a .} := \sum_{b = 15}^{50} X_{a b}

and

X_{. b} := \sum_{a = 15}^{80} X_{a b}

, and using

X_{a b} = π_{a b} c_{a b} X_{a 0}^{f} X_{b 0}^{m}

allows for developing

Δ X_{. b}

and

Δ X_{a .}

. Averaging yields

Δ (A_{m} - A_{f})

as linear in

Δ p

Δ r

\frac{Δ X_{0}^{f}}{X_{0}^{f}}

, and

\frac{Δ X_{0}^{m}}{X_{0}^{m}}

Geographical-Denominational Covariations, Controlling for the Marriage Market and Grain Price

This incursion into the mechanisms underlying the indicators of nuptiality allows us to formulate, test, and validate the multivariate vector auto-regressive (VAR) model:

Δ Y_{j} (t) = a_{j}^{'} Δ Z_{j} (t) + b_{j}^{'} Δ Z_{j} (t - 1) + ε_{j} (t),

(4)

where j represents a denomination-ensemble, t is time in year, where

Δ

is again the first differentiation operator (

Δ Y_{j} (t) = Y_{j} (t + 1) - Y_{j} (t - 1)

) to make the time-series stationary (all the time-series have a temporal trend, including prices in constant roubles 1913, and all are stationary after differentiation (partial correlogramm, Dickey–Fuller, and Philipps–Perron tests (not shown)), Y represents successively the SMAM,

I_{m}

, the proportion never married 20–29, and age difference between bride and groom,

Y_{j}

this variable Y for denomination-ensemble j,

Z_{j}

explanatory variables for denomination-ensemble j,

a_{j}

and

b_{j}

coefficients to be estimated and tested, and

ε_{j}

the innovation of zero mean. There is no reason why Y should have the same coefficients of explanatory variables for all denominations-ensembles. A regression of Y where denomination-ensemble appears as an additive explanatory variable is thus misleading. On the contrary, our objective is to identify the specificity of each denomination-ensemble. That is why a regression of Y for each denomination-ensemble j, hence

Y_{j}

, on its covariates

Z_{j}

, is necessary. However, these denominations-ensembles can influence each other, or simply vary simultaneously, through social learning or institutional influence such as the Church's. Correlations between the residuals of each regression capture the interactions between denominations-ensembles, net the effects of the covariates. They are quantified in the correlation matrix

Ω = (Ω_{i j})_{i, j}

of the perturbations

ε_{j}

Ω_{i j} = corr (ε_{i}, ε_{j})

. The VAR of Eq. (4) is then a (classical) seemingly unrelated regression.⁶³

The time-series are not of the same length (Rostov-on-Don containing a large mix of denominations was appended only in 1888, and statistics before that date are not available). We distinguish two periods to estimate Eq. (4): 1867–1889 for the (rural) Upper Don Orthodox, urban Orthodox, rural Old Believers, and Buddhists (results in Table 2), then again these groups for the period 1889–1916 to which we add (rural) Lower Don Orthodox (results in Table 3). The two periods are approximately of equal length and long enough to allow econometric estimation. We then re-estimate Eq. (4) with rural Armenians-Gregorians over 1889–1912 (Table 3), making sure that the coefficients for the previous groups have not changed substantially. And so on until all groups are considered. To each of these tables and for each dependent variable corresponds a correlation matrix (Tables 4 to 7).

Table 2.

Seemingly unrelated regressions (by group), 1867–1889: N = 23 years × 4 groups. Standard deviations in parentheses. All variables are first-differenced.

Dependent variable →	Singulate mean age at marriage, women(R2 = 0.4)				Standardized proportion of married women $I_{m}$ (R2 = 0.5)
First-differenced co-variate→	Log of the total number of marriable		0.01 × rye price	Redemption payments	Log of the total number of marriable		0.01 × rye price	Redemption payments
Group↓	Men	Women	0.01 × rye price	Redemption payments	Men	Women	0.01 × rye price	Redemption payments
Orthodox rural, Upper Don	−0.1 (0.4)	0.3 (0.2)	−0.3 (0.3)	−0.6** (0.2)	−1.0 (1.5)	0.0 (0.9)	0.1 (0.1)	−2.9** (0.1)
Orthodox urban	−0.2 (0.2)	0.4 (0.2)	0.3 (0.5)	-	−0.6 (1.4)	−0.7 (1.4)	0.3 (0.3)	-
Old Believer rural	0.4 (0.4)	−0.1 (0.3)	−0.2 (0.7)	−1.4** (0.4)	2.0* (1.0)	−2.2** (0.9)	1.3 (1.9)	1.9 (9.8)
Buddhist (rural)	0.0 (0.1)	0.0 (0.1)	−0.5 (0.6)	-	0.1 (0.4)	−0.2** (0.1)	0.1 (0.2)	-
Dependent variable →	Proportion never married 20–29 years (R2 = 0.2)				Age difference at marriage × 0.01 (R2 = 0.4)
First-differenced co-variate →	0.01 × Log of the total number of marriable		0.001 × rye price	0.01 × redemption payments	Log of the total number of marriable		0.01 × rye price	Redemption payments
Group ↓	Men	Women	0.001 × rye price	0.01 × redemption payments	Men	Women	0.01 × rye price	Redemption payments
Orthodox rural, Upper Don	2.2 (1.8)	2.0 (1.3)	0.1 (0.1)	0.1 (1.0)	−0.2 (0.3)	−0.2 (0.3)	0.2 (0.3)	−0.1 (0.2)
Orthodox urban	−0.6 (1.5)	1.2 (1.5)	0.1 (0.3)	-	0.6** (0.2)	−0.4** (0.2)	0.8 (0.5)	-
Old Believer rural	1.2 (1.9)	1.2 (1.5)	−0.1 (0.3)	−0.7 (1.8)	−1.9 (1.6)	3.2 (1.4)	0.2* (0.1)	2.9 (1.8)
Buddhist (rural)	−0.4 (0.4)	−0.1 (0.1)	0.1 (0.2)	-	0.0 (0.3)	0.0 (0.4)	1.8 (1.6)	-

*: significant at 10%; **: significant at 5%.

Table 3.

Seemingly unrelated regressions (by group), 1889–1916: N = number of group-years. Standard deviations in parentheses. All variables are first-differenced.

Dependent variable→	Singulate mean age at marriage, Women (R2 = 0.3)			Standardized proportion of married women $I_{m}$ (R2 = 0.5)			Proportion never married 20–29 years(R2 = 0.3)			Age difference at marriage × 0.01(R2 = 0.4)
Co-Variate→	Relative variation of the total number of marriable		0.01 × price of rye	Relative variation of the total number of marriable		0.01 × price of rye	0.01 × Relative variation of the total number of marriable		0.001 × price of rye	Relative variation of the total number of marriable		0.01 × price of rye
Group↓	Men	Women	0.01 × price of rye	Men	Women	0.01 × price of rye	Men	Women	0.001 × price of rye	Men	Women	0.01 × price of rye
					1889–1916
Orthodox, rural Upper Don	2.0** (0.7)	−0.8 (0.6)	0.9 (0.5)	1.1* (0.5)	−0.6 (0.5)	0.3 (0.5)	0.7 (2.4)	−0.6 (2.1)	0 + (0.2)	−0.3 (0.4)	−0.2 (0.3)	−1.6 (3.5)
Orthodox, rural Lower Don	0.4 (0.8)	0.1 (0.5)	−0.2 (1.1)	0.1 (0.7)	−0.6 (0.5)	−0.3 (0.9)	0.3 (0.8)	0.2 (0.5)	0.5** (0.1)	0.4 (1.2)	−0.7 (0.8)	−1.0 (1.3)
Orthodox urban	−0.4 (0.8)	0.4 (0.6)	−0.3 (0.8)	3.1 (4.6)	9.3** (3.1)	1.3 (5.8)	0.04* (0.02)	0.07** (0.02)	0.3 (0.2)	−3.1 (3.5)	10.1** (2.5)	1.6 (6.0)
Old Believer rural	−0.2 (0.5)	1.5 (1.0)	0.6 (1.1)	−0.3 (1.4)	−1.7 (2.3)	1.8 (3.4)	0.13** (0.03)	−0.04 (0.06)	0.2 (0.7)	−0.6 (1.4)	−4.2 (2.5)	2.7 (3.4)
Buddhist (rural)	−0.1 (0.1)	−0.3 (0.2)	−0.5 (1.2)	0.8 (0.9)	2.0 (2.1)	3.6 (8.2)	0.04 (0.04)	0 + (0.1)	0.3 (0.3)	0.8 (0.8)	−0.5 (1.9)	−6.0 (7.9)
					1889–1912
Armen-Gregor.rural	2.0** (0.9)	0.8 (0.8)	0.5 (2.0)	−1.1 (0.7)	−1.0 (0.7)	−2.1 (2.0)	0.09* (0.04)	−0.05 (0.04)	1.6 (0.9)	−3.1** (1.0)	−1.1 (0.8)	−2.0 (2.0)
					1889–1907
Armen-Gregor.urban	−0.4 (0.6)	0.2 (0.3)	−2.0 (1.6)	3.5 (4.1)	4.6* (2.4)	−6.4 (18.4)	8.2** (3.0)	0.2 (1.7)	−2.1** (0.9)	4.0 (3.7)	3.8 (2.3)	−4.4 (18.9)
					1890–1910
Jew (urban)	0.1 (0.7)	0.2 (0.3)	0.1 (1.7)	0.5 (5.6)	−3.0 (2.0)	−3.0 (1.6)	10.2 (7.9)	9.1 (26.8)	−13.7 (19/0)	−3.0 (4.7)	−1.1 (1.7)	−2.9* (1.4)
					1897–1909
Lutheran rural	0.9** (0.3)	0.8** (0.3)	1.3 (5.2)	−3.8 (2.4)	3.0 (2.2)	4.4 (2.9)	1.2 (1.3)	5.1** (1.5)	0.9 (1.9)	−0.3 (4.2)	4.1 (2.6)	0.8 (4.5)
					1889–1907
Catholic (urban)	0.2 (0.2)	0 + (0.1)	1.3 (1.6)	1.0 (1.3)	−0.5 (0.5)	−1.2 (9.0)	9.9* (5.3)	−4.1 (2.7)	−2.4** (0;5)	1.2 (1.5)	0.3 (0.7)	−2.5 (9.6)
Muslim (urban)	3.6** (1.2)	−3.0** (1.1)	5.9 (3.7)	4.4* (2.5)	−3.2 (1.8)	0.8 (7.3)	−3.6 (2.2)	3.1 (2.0)	7.7 (7.0)	1.0 (2.3)	0.2 (2.0)	−4.8 (6.3)

*: significant at 10%; **: significant at 5%; In addition, redemption payments for Eq. (4) estimated from 1889 to 1910 on women's SMAM: Orthodox Upper Don: 0.2 (0.4), Orthodox Lower Don − 0.2 (0.2), rural Old Believers 1.0 (0.8); Standardized proportion of married women: Orthodox Upper Don: 0 + (0.02), Orthodox Lower Don 0.01 (0.02), rural Old Believers 0.02 (0.02) ; proportion never married 20–29 Orthodox Upper Don: 0 + (0.01), Orthodox Lower Don − 0.003 (0.002), rural Old Believers − 0.11** (0.05); on age difference at marriage Orthodox Upper Don: 0.1 (0.3), Orthodox Lower Don 0.5 (0.3), rural Old Believers 3.9 (2.3).

Therefore elasticity differences can be compared between groups, with tests performed between the coefficients calculated for each system (e.g., the comparison between Buddhists and Catholics is performed for the system estimated for the period 1889–1907).

Results

The estimation of (4) yielded no significant coefficient associated with 1-year lagged variables; so we re-estimated (4) without lagged variables and present the results in Tables 2 and 3.

Debt constraints For the period 1867 to 1889 (Table 2), more redemption payments are associated with younger age at marriage among rural Upper Don Orthodox and Old Believers (coefficients − 0.6, − 1.4 for the SMAM) and less contribution to marital fertility (− 2.9 for Upper Don Orthodox for $I_{m}$ ): years of lower income would lead to postponing payments and weddings. From 1889 to 1910 (footnote of Table 3), coefficients of redemption payments are not significant, which supports the possibility that grain price no longer affected nuptiality.

Grain price For the period 1867–1889, we find no significant coefficient of grain price. For the period 1889–1916, as we include Lower Don Orthodox and urban groups, an increase in grain price increases the proportion never married between 20 and 29 (coefficient 0.5) among Lower Don Orthodox, which is consistent with the cost of marriage, and decreases this proportion among urban Catholics and urban Armenian-Gregorians (− 2.1 and − 2.4), which is consistent with the fact that these groups lived from trade. However, for the SMAM, $I_{m}$ , and age difference, grain price has no significant effect.

The marriage market For the period 1867–1889 (Table 2), consistently, the decrease in the total number of marriageable women for rural Old Believers and Buddhists increases $I_{m}$ (− 2.2 and − 0.2), but this effect is not significant for the other groups. For urban Orthodox, age difference between bride and groom increases with the total number of marriageable men (0.6) and decreases with the total number of marriageable women (− 0.4), which is consistent with the fact that for Orthodox, migrant men in the city were 3–4 years younger on average and more numerous than migrant women, maintaining an imbalanced matrimonial market.

For the period 1889–1916 (Table 3), the SMAM increases with the total number of marriageable men for rural Upper Don Orthodox (coefficient 2.0), urban Armenian-Gregorians (2.0), rural Lutherans (0.9), and (urban) Muslims (3.6), but is not influenced by the total number of marriageable women, except for Lutherans (0.8) and Muslims (− 3.0). The former consecrated marriage after a period of living together (interview in 2013 with Lutheran pastor Jacques-Noël Pérès, from the Faculté théologique of Paris) and the latter practiced polygamy. More marriageable men are associated with a higher proportion of never-married people between the ages of 20 and 29 among Catholics, rural and urban Armenian-Gregorians, rural Old Believers, and urban Orthodox: the causality that makes sense is that more unmarried women aged 20–29 left more unmarried men and thus more marriageable men. The groups concerned by significant coefficients in Table 3 are also groups where women were relatively rare. This is not the case with rural Orthodox or Jews, for which the coefficients, consistently, are not significant. This mechanism applies to rural Lutheran and urban Orthodox women. The former case does not concern a group with scarcity of women and the latter case may reflect the delay in the integration of migrant women into the urban milieu. The age difference between bride and groom increases with the total number of marriageable women for Urban Orthodox (coefficient 10.1 in Table 3), as women migrated at least as old as men. The other groups did not suffer as much from a chronic deficit of women and no effect is significant on the age difference.

Co-variation between denominations The correlation matrices $Ω$ between the residuals of short-run fluctuations in Eq. (4), for each dependent variable, are presented in Tables 4 to 7. For the SMAM (Table 4), this matrix shows firstly, the relative isolation of (rural) Upper Don Orthodox, rural Old Believers, and rural Lutherans; secondly, substantial correlations of Lower Don Orthodox with rural southerners and urban Orthodox: 0.45 with urban Orthodox, 0.53 with Buddhists, 0.40 with rural Armenian-Gregorians; thirdly, very low correlations between rural and urban Armenian-Gregorians; fourthly, substantial and positive correlations between urban Armenian-Gregorians and urban Orthodox; fifthly, very low correlations of both these groups with Jews, Catholics, or Muslims. The correlations presented in Table 5 for the proportion never married among women aged 20–29 confirm the weak covariation of most groups with respect to celibacy. Noteworthy although still low correlations of 0.45 (weakly) link urban Orthodox to Lower Don Orthodox and to urban Armenian-Gregorians. For the standardized proportion $I_{m}$ of married women, the correlation of 0.76 in Table 6 between urban Armenian-Gregorians and urban Orthodox confirms this urban link. For the age difference between bride and groom, rural Orthodox and rural Old Believers are again correlated (0.43 in Table 7). The low or negative correlations for the other pairs of groups show that age differences varied almost independently from one group to another.

Discussion

Orthodox: The Alliance of Church and Tradition Tested by Economy, Demography, and Ecology

While Upper Don Orthodox perpetuated early marriage (brides aged 17–18), conceding over the years a slight upward trend, urban Orthodox adopted a slightly later marriage (20 years until 1889, 22 years after) and Lower Don Orthodox (19 years until 1889, 20 years after) fell between these two models. Figures 2 to 5 showed that all Orthodox groups are distinct from other denominational groups, even (rural) Buddhists and rural Old Believers, who might have been thought to be close by virtue of their rurality and their participation in the irregular army as Cossacks.

This situation is consistent with five keys to reading: economic, demographic, and ecological, which go in the direction of postponing age of marriage; politico-religious and traditional, which push toward early age.

Economic: In the countryside, the rule of periodic division of peasant land encouraged newlyweds to form a new production unit, eligible for its own share of land: pressure for early marriage and thus for more labor units came equally from the landlord, the peasant family, and the commune.⁶⁴ This steppe-specific possibility distinguished the Don Army Territory from Western European societies, where the selection of a single heir often put younger brothers at higher risk of celibacy. ⁶⁵ In proto-industrialization theory, ⁶⁶ resource constraints forced young men to delay marriage until they had accumulated sufficient capital, a theory that was to be modulated by the place of residence after marriage.⁶⁷ In the Don Army territory, according to this theory, if resources remained abundant, the availability of accessible and profitable land made it possible to persist in early marriage. In addition, our econometric analysis showed that short-term fluctuations did not respond much to that of grain price, which rules out homeostatic regulation of marriage by grain price. Until 1889 however, fluctuations in redemption payments were related to those in age at marriage, in the expected sense that a delay in payment accompanied a delay in marriage.

Demographic: The downward trend in mortality from the mid-1880s onward,⁶⁸ without a simultaneous associated decline in fertility, tended to result in a relative excess of marriageable women, because they had comparatively lived through better mortality conditions, as they married younger than men. If families made an effort to marry off their daughters, then the oldest marriageable women (but still very young) could be preferred, which delayed age at marriage by that much. The process continued as long as mortality declined and fertility did not compensate for it to decrease the total number of births. Also, a chronic increase in the total number of marriageable women led to a decrease in $I_{m}$ among rural Orthodox. Figure 2 showed that the SMAM increased among Orthodox, and Table 3 that fluctuations in the SMAM responded to fluctuations in the total number of marriageable men but not to fluctuations in the total number of marriageable women: The competition for women was made more intense by the presence of more marriageable men. This shows that the marriage market was driven by women.

In addition, a sustained increase in the total number of young people led to a greater financial demand on the part of parents to support more surviving children. Then there was the need to equip young men for military service, since the 1875 law required Cossacks to pay for their own equipment. It was necessary to finance weddings, which were very expensive social events for families, amounting to 1 year's income.⁶⁹

Ecological: The increasing total number of households resulting from the maintenance of early marriage among Orthodox favored the ecological crisis, the reasons for which lie in the absence of a mechanism for preserving the communal system that allowed free and equal access to natural resources. This crisis worsened in the second half of the nineteenth century.⁷⁰

Political-religious and traditional: We have mentioned the role of marriage as a factor of social recognition in villages. This role was encouraged by the Russification campaign launched by Alexander III and continued by Nicholas II, through which Orthodox priests strengthened their control over the population and encouraged religious observance, tradition, and allegiance to the tsar (we are not saying that they “regulated” age at marriage, but note that early age of marriage was not likely to increase if tradition and conservatism were advocated). The persistence of early marriage in Figures 2 to 5 is consistent with the effectiveness of this campaign in impeding the adoption of late marriage among the Orthodox. This campaign intensified under Nicholas II beginning in 1894, while other groups were beginning their transition to late marriage. This favored the maintenance of $I_{m}$ among rural Upper Don Orthodox women above 0.8, and still high—the highest among those which decreased—levels for rural Lower Don and urban Orthodox. The strengthening of social control exercised by the Orthodox Church at the instigation of the monarchy and the resulting coercion can explain the decrease of the SMAM for urban Orthodox after 1909 to around 19 years old, close to the age observed for rural Upper Don Orthodox (Figure 2). The Orthodox religion thus acted as a conservative control toward early marriage.

Urbanization as a Factor of Delaying Marriage for Orthodox

The growth of cities, industry, and commerce provided sources of income other than field work, especially with the resource crisis in the countryside, where good-quality land was becoming rarer. Migration to cities developed rapidly (for example, between 1889 and 1907, the average annual rate of population growth was 4.2% for Orthodox in Rostov-on-Don and 3.3% for Armenian-Gregorians in Nakhchivan’). Migrants to cities were more likely to be men than women (e.g., net migration of 3642 women to cities for 4921 men in 1885–1889). Immigrant women to cities were also at least as old as men: 23.3 years old for women, 23.5 for men from 1872 to 1895; these figures are 24.2 and 22.4 from 1896 and 1899; 25.4 and 24.2 from 1900 to 1916. As men usually married younger women (Figure 5), relatively older migrant women participated in delaying age of marriage. Cities also created an environment in which young men could take the time to find and court a partner, rather than being married off early by parents or matchmakers, as in the countryside.⁷¹ In the Don countryside, the means of achieving independence and acquiring wealth or land in rural society was marriage; in cities, opportunities for wage labor brought the possibility of independence without immediate marriage. Life in city provided access to other forms of income for which early marriage was not a precondition. It thus challenged the traditional marriage system and the social norm that encouraged young people to establish independent households.

Toward Late Marriage among Urban Jews and Urban Armenian-Gregorians

For the other denominations, Figures 2 to 5 and Eq. (4) showed similarities in trajectories between (urban) Jews (residing in Rostov-on-Don) and urban Armenian-Gregorians (residing mostly in Nakhichevan’, which is a city adjacent to Rostov-on-Don), but the correlation between the residuals of the regression (4) for these two groups is not significant. Meanwhile, the correlation of 0.45 for the SMAM in Table 4 between residuals of (4) for urban Armenian-Gregorians and urban Orthodox may reflect the fact that these two groups had recently moved out of the countryside⁷² and were likely to seek the same urban professional opportunities. In contrast, Jews for example had been confined to the cities for a long time and were inserted there, with more established relationships—we may speculate—conditioning the choice of spouses. Jews show similar trajectories to urban Armenian-Gregorians, too, but the correlation of short-term fluctuations is not significant. Thus, the transformation of the marriage pattern was not something mechanical or uniform: if the result was late marriage, the groups on the way could vary independently of one another or not.

In Rural Areas, Buddhists, Old Believers, and Lutherans Toward Later Marriage

However, late marriage from 1894 onwards was not confined to city dwellers, but was adapted by (rural) Old Believers and Buddhists, who might have been expected to be close to rural Orthodox, who still married early. Despite their relative geographical isolation in the southern steppes of Sal'skii and their Mongolian origin which distinguished them from other inhabitants of the Don Army Territory, Buddhists also undertook this movement towards later marriage with a SMAM increasing from 17 years in 1894 to 20 years in 1914 with a peak of 26 years in 1909. Even nomads had to face the increasing total number of women due to the decrease in mortality and its consequences on sex balance in the marriage market.

The social and geographical isolation of Old Believers and rural Lutherans was not enough to stop the drift towards late marriage. Old Believers could have been expected to be conservative because of their isolated way of living (since their rejection of the reform introduced by Patriarch Nikon in 1666–1667). They however adopted late marriage after 1894. The schism with the Orthodox preserved them from the authority of the Holy Synod, which worked for early marriage for Orthodox as part of the Russification campaign.

Lutherans, of German origin and language and residing in the hinterland of the city of Taganrog and allowed to reside in rural areas only since the 1870’s, may have been influenced by the westerner model of later marriage, if not by the city life in Taganrog. However, Lutherans were well isolated from other groups, making a point of self-sufficiency for their colonies in Russian territory, economically as well as culturally, educationally, or linguistically.⁷³

The persistence of early marriage among (urban) Catholics On the other hand, Catholics who lived mainly in Rostov-on-Don remained faithful to the early marriage model: city life alone thus does not explain the transition; religion interfered for this group as it did for urban Orthodox. The reason we proposed was that the Catholics of Rostov-on-Don were mostly Poles. In Taganrog, the Catholics also included Germans, but the data are not of sufficient quality to allow reconstruction. In Rostov-on-Don, in any case, the origin allied with religion succeeded in preventing the transition to late marriage.

Muslims Among Muslims, the large fluctuations between late and early marriage are based on small numbers, so it is difficult to say.

Divergence among Armenian-Gregorians

While urban Armenian-Gregorians married later, rural Armenian-Gregorians persisted in marrying early. Batiev and Sushchij⁷⁴ explained the divergence: while rural Armenian villages had become isolated, endogamous enclaves, Armenian-Gregorians residing in the city of Nakhichevan’ lost their language to Russian and chose Western education, mores, and clothing. Thus, a same denomination could be divided on the behavior of marriage.

The nuptiality transition is multiple The links between religion, residence, and economy are thus not simple. There is no mechanistic determination between marriage pattern, denomination, and residence. Each group was caught in a web of determinisms: The city should have favored late marriage because of the relative freedom of the formation of couples and the relative independence from parents; the countryside should have favored the tradition of early age; a single obedience should have standardized behavior. However, the case of the Don Army Territory, with its multiple denominations and groups, demonstrates that the pattern of marriage retained the imprint of history left by the choices of individuals and the struggles for influence. The case of urban Armenian-Gregorians is typical, where the choice of acculturation led them to follow the Westernizing trend, while their rural counterparts caught in communal isolation ignored the transition of nuptiality. In contrast, but also typical, urban and Lower Don Orthodox tended to delay age of marriage, against the directives of the Holy Synod, which tightened discipline. This was not so much a matter of religion as of the deliberate action of the archconservative K. P. Pobiedonostsev, Over Procurator of the Most Holy Synod of the Russian Orthodox Church. In office from 1880 to 1905, but keeping influence until 1917, he professed hatred of the industrial revolution and urban growth and supported a reactionary policy to counter the influence of Western-oriented liberal and secular ideas.⁷⁵ Assuming that the transition to late marriage was underway, agents could therefore accelerate the transition, as among urban Armenian-Gregorians, or slow it down against economic or ecological necessity, as among Orthodox.

Conclusion

It is fortunate to have in the same territory in the period of demographic transition eight denominational groups combining with the urban-rural contrast and different economic modes (agrarian, commercial, and industrial). The transition offers a rare moment when behaviors can be distinguished from each other, and helps point out situations of isolation or, on the contrary, covariations. The picture that we reveal shows less a mechanism, although late marriage settles durably before the first world war for many groups, urban as rural, than a place of struggles of influences between economic mode, urbanity, religion, westernization, and influence of neighbors. Urban lifestyle favored late marriage, as with Jews or urban Armenian-Gregorians, but not exclusively. Catholics were urban, but kept marrying early; Buddhists, Lutherans, Old Believers were rural, and even socially conservative and isolated, due to geography for the first, that they were German settlers for the second, to the schism having led to social ostracism for the third, but still moved toward late marriage. The Orthodox Church strengthened its control over the faithful—we do not say that it went as far as regulating age at marriage. In rural areas, the Orthodox Church found as an ally the rules of Cossack communities where weddings were an element of social cohesion. There, a drama was being played out, that of demographic growth sustained by the downward trend in mortality, coupled with the ecological crisis and the resulting shortage of quality land. The maintenance at all costs of early marriage, in the Cossack tradition, despite some emigration to the cities, precipitated the ecological crisis at the end of the nineteenth century.⁷⁶

The transition of nuptiality was therefore not linked to a single determinant; the determinants did not add up to each other, but reflected choices or political coercion for each group. The transition was played out within each group, and these groups were more or less linked to each other, mainly according to geographical criteria, but never strongly, and denominational groups behaved in relative isolation from the Orthodox, although this latter denomination constituted the numerical majority.

In the theory of minority group status, fear of acculturation and economic or political disadvantage encourages group preservation and a high birth rate. Minorities in the Don Army Territory were defined by their religious allegiances but also by their “ethnicity” (Jews, Lutherans (mainly Germans), Muslims, Armenian-Gregorians), language (Yiddish, Armenian, German), occupation (trade and industry versus agriculture), and by their experience of political persecution. The latter included the loss of churches for the Armenian-Gregorians from 1903; a settlement zone for Jews created in 1791 and a series of repressive policies, including 1882 (reduction of the settlement zone), 1887 (numerus clausus for Jewish children in secondary schools), 1891 (10,000 Jews expelled from Moscow), 1892 (exclusion of Jews from civil service positions), 1903 (pogrom in Kishinev) and 1905 (pogroms in Russia, including Rostov-on-Don), 1914–1917 (ban on Jews settling even temporarily in Cossack territories).

This segregation of minorities may have shaped attitudes toward marriage, contributing to the formation of distinct identities. One might think that the struggle for minority survival would encourage early marriage from a pro-natalist perspective, a corollary of nationalism. This is not what we observe for Jews, Lutherans, urban Armenian Gregorians, even rural Old Believers, and to a lesser extent Buddhists: on the contrary, for these minorities, the decline of mortality was associated with later marriage and subsequently lower overall fertility. For non-Orthodox groups, late marriage as a strategy to avoid the burden of large families was comparable to that observed in the Western demographic transition underway since at least the mid-nineteenth century.

The disparity in marriage behavior after 1894 split society between and within denominations or drew connections (between urban Armenian-Gregorians and urban Orthodox for example). The interest of this case study is to clarify the contrasts within the same denominations, the coexistence of different denominations within the same cities, and to relativize situations of social or geographical isolation. This could not have been done without the reconstruction by district-denomination over a fairly long period, when the data for the other regions of Russia are often based solely on the census of 1897, incapable by construction of revealing temporal follow-up. This could not have been done either without the seemingly unrelated regressions that preserved the specificity of each group over time while taking into account and quantifying covariation.

The nuptiality transition finally was not an unequivocal matter where the same causes produce the same effects: the proximity of residence or denomination was not mechanical but only created a structure indicating the direction of the marriage model adapted to the economy of the moment. Within this temporal structure, the coercion imposed by a politico-religious authority or the force of conservatism could still thwart the movement. On the contrary, the flight forward to the model of late marriage, sometimes at the cost of acculturation, could accelerate it. The transition of nuptiality put in play religion and economy. Thus a social question concerning the intimate revealed the opposition between Westernization and conservatism, which, in the Orthodox case at least, took nationalistic tints (through Russification).

Table 4.

Correlation matrix for the singulate mean age at first marriage in Eq. (4) and Table 3. Each correlation refers to the longest common period within 1889–1916 of the reconstructed SMAM.

Rural Ortho. Upper Don	1.00
Rural Ortho. Lower Don	0.22	1.00
Urban Orthodox	−0.24	0 . 45	1.00
Rural Old Believer	−0.04	0.27	0.33	1.00
Buddhist	−0.07	0.53	0.26	0.18	1.00
Rural Lutheran	0.42	0.04	0.33	0.13	0.30	1.00
Rural Arm.-Gregorian	0.25	0.40	0.35	0.26	0.27	0.11	1.00
Urban Arm.-Gregorian	−0.31	0.14	0.45	−0.06	−0.11	−0.18	0.21	1.00
Jew	0.10	0.02	−0.19	0.19	0.24	−0.23	−0.22	−0.19	1.00
Catholic	0.13	0.03	0.20	0.35	0.16	0.18	0.30	−0.05	0.20	1.00
Muslim	−0.39	−0.25	0.10	0.11	0.11	0.21	−0.27	0.33	0.33	0.14	1.00

Values in bold face are significant at the 5 % level.

Table 5.

Correlation matrix for the proportion of never married women 20–29 (Eq. (4) and Table 2). Each correlation refers to the longest common period of the reconstructed proportion.

Rural Ortho. Upper Don	1.00
Rural Ortho. Lower Don	0.22	1.00
Urban Orthodox	0.19	0 . 45	1.00
Rural Old Believer	0.15	0.27	0.40	1.00
Buddhist	0.16	0.33	0.26	0.18	1.00
Rural Lutheran	0.32	0.03	0.33	0.13	0.30	1.00
Rural Arm.-Gregorian	0.25	0.40	0.35	0.26	0.27	0.11	1.00
Urban Arm.-Gregorian	−0.31	0.14	0.45	−0.06	−0.11	0.18	0.22	1.00
Jew	0.10	0.02	−0.19	0.19	0.44	−0.23	−0.21	−0.19	1.00
Catholic	0.13	0.03	0.20	0.25	0.16	0.32	0.36	−0.05	0.20	1.00
Muslim	−0.39	−0.26	−0.35	0.10	0.11	0.37	0.27	−0.05	0.36	0.14	1.00

Values in bold face are significant at the 5 % level.

Table 6.

Correlation matrix for the standardized proportion of married women $I_{m}$ (Eq. (4) and Table 3). Each correlation refers to the longest common period of the reconstructed $I_{m}$ .

Rural Ortho. Upper Don	1.00
Rural Ortho. Lower Don	−0.47	1.00
Urban Orthodox	−0.27	−0.17	1.00
Rural Old Believer	0.37	−0.25	−0.11	1.00
Buddhist	0.38	−0.50	−0.23	0.76	1.00
Rural Lutheran	0.00	0.14	0.26	−0.18	0.03	1.00
Rural Arm.-Gregorian	0.36	0.08	−0.42	0.20	0.18	0.11	1.00
Urban Arm.-Gregorian	−0.07	−0.15	0.76	−0.25	−0.41	0.16	−0.31	1.00
Jew	−0.43	0.19	−0.11	−0.09	−0.20	0.03	−0.02	−0.27	1.00
Catholic	0.02	−0.30	0.18	0.22	0.03	−0.36	−0.16	−0.11	−0.01	1.00
Muslim	0.26	−0.37	−0.02	0.23	0.36	0.32	−0.20	−0.15	0.28	−0.35	1.00

Values in bold face are significant at the 5 % level.

Table 7.

Correlation matrix for the age difference at marriage (Eq. (4) and Table 2). Each correlation refers to the longest common period of the reconstructed age difference.

Rural Ortho. Upper Don	1.00
Rural Ortho. Lower Don	0.24	1.00
Urban Orthodox	−0.04	0.33	1.00
Rural Old Believer	0 . 43	−0.22	−0.27	1.00
Buddhist	−0.19	−0.31	0.11	−0.24	1.00
Rural Lutheran	0.27	−0.25	−0.24	0.12	0.31	1.00
Rural Arm.-Gregorian	−0.14	−0.05	−0.12	−0.44	0.28	0.06	1.00
Urban Arm.-Gregorian	−0.21	0.06	0.63	0.04	0.25	0.29	0.07	1.00
Jew	−0.40	0.26	0.64	−0.62	0.09	−0.21	0.34	0.27	1.00
Catholic	0.13	0.03	0.20	0.23	0.16	0.14	0.26	−0.05	0.20	1.00
Muslim	−0.38	−0.26	0.10	0.11	0.11	0.29	−0.27	0.33	0.33	0.14	1.00

Values in bold face are significant at the 5 % level.

Footnotes

Appendix A

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article

ORCID iD

Noël Bonneuil

Notes

Author Biographies

Noël Bonneuil is senior researcher and full professor. His major interest is in the formalization of change in human populations, as a mathematician, statistician-economist, and historian. He is the author of Transformation of the French Demographic Landscape, 1806-1906. Oxford: Clarendon Press (1997) and Introduction à la modélisation démographique. Paris: Armand Colin (1997).

Elena Fursa is associate professor in economic demography. Her major area of research is about children in difficult situations, precarious families, and quality of life. She was responsible for the Russian part of a French-German-Russian project on child protection and a member of a French- Russian project on forms of poverty and social vulnerability.

Between Economy and Social Coercion: Nuptiality in Transition. The Case of the Don Army Territory (Southern Russia),1867–1916

Abstract

Keywords

Context and Data

Inter-Denominational Marriages

Descriptive Analysis

Determinants

Vector Autoregressive Regression

Geographical-Denominational Covariations, Controlling for the Marriage Market and Grain Price

Results

Discussion

Orthodox: The Alliance of Church and Tradition Tested by Economy, Demography, and Ecology

Urbanization as a Factor of Delaying Marriage for Orthodox

Toward Late Marriage among Urban Jews and Urban Armenian-Gregorians

In Rural Areas, Buddhists, Old Believers, and Lutherans Toward Later Marriage

Divergence among Armenian-Gregorians

Conclusion

Footnotes

Appendix A

Declaration of Conflicting Interests

Funding

ORCID iD

Notes

Author Biographies