The children's ages in the census lists of the 3rd revision: the problem of data reliability and birth rate calculation

Pankrat Ivan Alekseevich Postgraduate student; Faculty of History; Lomonosov Moscow State University specialist of the 1st category; Russian State Archive of Ancient Acts 119234, Russia, Moscow, Ramenki district, Lomonosovsky prospekt, 27 K. 4 ipancrat99@gmail.com Other publications by this author

Introduction

As a result of Peter the Great's transformations in the first quarter of the 18th century, new types of accounting documents appeared in the record-keeping system of Russian institutions, which contain mass information about the population and allow historians to analyze various demographic processes. Such mass sources include church records such as metric books and confessional murals, as well as audit tales, which are primary materials of censuses conducted periodically by the government for fiscal purposes and which recorded the available male tax population, including children of all ages.

Of particular value to the researcher are the tales of the third revision, conducted from 1761 to 1767 (the bulk of the tales were submitted in 1762-1764) ^{[1, pp. 64-65]}. Since this revision, a single printed form has been developed and officially approved to fill out the revision tales (which facilitates the process of formalizing the information contained in them to form an electronic database), which, most importantly, for the first time began to include information about women and girls ^{[1, p. 65]}.

The most difficult population accounting was in the southern counties of the state, where in the middle of the XVIII century. The processes of internal colonization and the formation of a network of settlements continued ^{[2, pp. 27-37, 50]}. Landlords played an active role in these processes, transferring their serfs to undeveloped lands. One of these intensively developed regions was the southern outskirts of the Middle Volga region ^{[2, pp. 36, 50]}, a significant part of which, before the Catherine administrative reform, was occupied by the vast Mining camp of Simbirsk Uyezd.

After the previous per capita census was conducted in 1745, 23 new settlements were founded in Zavalny Stan by landowners in addition to 274 already existing villages where noble soul ownership was present ^{[3, 4]}. In addition, about 10% of the landed peasants of the Zavalny camp, counted by the third revision (5.7 thousand audit souls out of 56 thousand), were immigrants who changed their place of residence after the second revision, as well as persons born into families of immigrants after being transferred to a new place ^{[3, 4]}.

In this regard, the question of the impact of migration on the family and age structure of peasant farms, as well as on the demographic behavior of serfs, manifested in fertility, mortality, marriage and natural growth, is of particular interest. The dynamics of the fertility rate, in particular, can demonstrate whether peasants consciously sought to regulate childbearing during their settlement in a new place, and exactly how fluctuations in the birth rate depended on the type of family and the conditions of resettlement.

Metric books are most suitable for calculating this coefficient, for the reason that regular registration of births and deaths (more precisely, baptisms and burials) was the main task of church accounting, whereas revision fairy tales did not record children who were born and died between revisions. At the same time, the fiscal purpose of the audit tales necessitated the accounting of all peasant migrations that took place after the previous census, since the movement of the taxable population directly affected the volume of poll taxes.

Thus, when analyzing the impact of resettlement on fertility, it is the combination of metric books and revision fairy tales that would allow us to obtain the most accurate results: if revision fairy tales make it possible to establish the very fact of transfer to a new place, then the dates of birth and death of the children of the settlers can be identified from the metric books.

Meanwhile, for most of the XVIII century, such a comparison of sources is difficult due to the extremely poor preservation of metric books ^{[5, p. 38, 197 – 198, 279]}; ^{[6, pp. 64, 78]}, which is also characteristic of the region under consideration. During the Catherine provincial reform, the territory of Simbirsk Uyezd was divided between the Simbirsk and Saratov governorates, and then, in accordance with the new administrative boundaries, the dioceses were reorganized, in the spiritual consistories of which parish registers were kept. According to the inventories of the consistory funds published on the websites of the Ulyanovsk and Saratov regional archives, the earliest copies of the metrics date back to the 1770s and 1780s and are represented by single cases for individual years, and the more or less systematic preservation of church records begins only with the onset of the XIX century ^[7-9].

It follows from this that the revision tales remain the only source for analyzing the birth rates of serfs in the Simbirsk district in the middle of the XVIII century. When calculated on the basis of audit materials, these figures will be underestimated due to the underestimation of children whose birth and death occurred during the 17-year inter-revision interval. However, this disadvantage is mitigated by the fact that when considering the impact of resettlement on fertility, the focus will be not so much on absolute indicators as on their changes depending on the conditions of resettlement.

At the same time, the procedure for compiling revision fairy tales had several features that had a negative impact on the display of demographic indicators of the peasant population. So, usually the storytellers were elders and elected representatives from among the local residents, who in the vast majority of cases were not literate. An analysis of one of the files of the RGADA Foundation No. 350, containing 187 copies of the revision tales of the third revision in Simbirsk County, shows that only 9 "informants" independently filled out the original document and signed it, while all other tales were compiled by third–party literate people under dictation ^{[3, l. 1-672]}. The revision form He demanded to report not the exact date of birth, but the number of full years, and illiterate peasants who had fairly general ideas about their age and the age of their fellow villagers were inclined to give approximate figures, often multiples of 5. This leads to a distortion of the picture of the total number of different age groups – that is, in the language of demography, to the emergence of age accumulation.

Researchers who analyzed the revision tales as a demographic source (B. N. Mironov ^{[10, pp. 521-530]}, I. A. Troitskaya ^{[6, pp. 49, 77-78]}, N. L. Yurchenko ^{[11, pp. 64-71]}), paid attention to this phenomenon, among others. However, accumulation was considered only for adults – the question of the presence of errors and distortions in children's ages, called from memory by the compilers of fairy tales, has not yet become the subject of a special study. To measure the degree of accumulation in works on historical demography, the Whipple index was widely used, which by default is calculated only for people aged 23-62 (this limitation is usually explained by the fact that the age structure of children, young people and the elderly is less susceptible to distortions associated with preferences in favor of certain end figures) ^{[12, p. 138]}.

Since the birth rate analysis based on the revision materials is based on the reported ages, in this paper we will try to establish how accurately the revision tales of the third revision reflected the age structure of children of both sexes under 17, who were not taken into account by the previous census and for whom, accordingly, there was no data to correct the information received from the storytellers.

To achieve this goal, it is necessary to solve two interrelated tasks: firstly, to determine the limits of possible errors when specifying children's ages, and secondly, to identify the presence or absence of underestimation of certain age groups of children, similar to the age accumulation in the adult population. If these distortions are indeed present in the revision tales and strongly influence the dynamics of the birth rate, it is also necessary to separately consider statistical methods for correcting the initial data, which make it possible to more accurately reconstruct the sex and age structure.

The object of the study is the ages of peasant children under 17 years old, recorded in the revision tales of the third revision. In turn, the subject of the study is possible errors and distortions that were made by illiterate storytellers in the oral assessment of ages, as well as the nature of their occurrence and the impact on the accuracy of demographic analysis.

Methods

The research methodology is based on the formalization of information from revision tales in the form of an electronic database designed for statistical analysis of demographic indicators of serfs. The database was built in Excel and includes information on 1,2952 landowner peasants of both sexes who were transferred between the second and third revisions to various settlements in the Zavalny Camp of Simbirsk county. This group includes not only the migrants themselves, but also members of their families who did not participate in the migration, who left before this event or appeared after being transferred to a new place (usually the form of the revision tale did not separate these peasants from the direct participants in the resettlement).

In total, more than 2.1 thousand revision tales were submitted to the settlements of the Zavalny Camp during the III audit, 419 of them contain information about the transfer of serfs to a new location and were used to fill out the database ^{[3, 4]}. The upper limit of the age range under consideration was limited to 17 years due to the fact that the main part of this set of sources The total of 344 fairy tales (82%) was compiled 17 years after the previous revision, that is, in 1762, while 43 and 32 documents (10% and 8%) accounted for 1763 and 1764, respectively.

The following data is provided about each of the peasants in the database: the year the tale was submitted (1762, 1763 or 1764), gender, age according to revision II, age according to revision III, in case of loss between two censuses, its cause (death, flight, recruitment, transfer to a new location, sale, etc. d.), as well as the year of disposal.

The serfs included in the database belong to a total of 2,642 families. For 617 of them (23.4%) transferred within Simbirsk County (including the settlements of the Industrial Camp and the "Zavolzhsky side", whose revision tales are mainly contained in other files of the foundation No. 350 ^{[13, 14]}), a comparison was made with the information contained in the revision tale of the former place of residence; all The discrepancies between the "old" and "new" fairy tales were recorded in an additional special column.

The accuracy of determining children's ages

So, to what extent did the peasants make mistakes in estimating children's ages, and what were the limits of possible deviations between the actual and "revision" ages of children aged 0-17? To answer this question, it is possible to analyze cases of double counting of the child population that arose during the submission of audit fairy tales. We are talking about peasant families who moved during the census (1762-1764) and, as a result, were recorded both in their old and new places of residence in full, as well as individual cases when the storyteller at the old place included in the revision tale the evicted children who were born after the second revision (although from the point of view of the form it was not necessary at all).

To ensure comparability of the data, data from the revision fairy tales of 1763-1764 were excluded from the analysis, since there is a significant time gap between the presentation of the old and new fairy tales (up to one or two years) it could significantly affect the accuracy of the age indication. In total, 93 cases of double counting of children according to the revision fairy tales of 1762 were recorded in the database, and in 16 cases the age information differs between the two versions of the record. The calculation of the average absolute error in all 93 cases shows that it was 0.38 years for boys, 0.66 years for girls, and 0.48 years for children of both sexes, which may indicate a relatively high accuracy in determining children's ages.

For comparison, we calculate this indicator for women over the age of 18. Unlike their male peers, whose age according to the third revision was compared with the data of the previous census, they, like children under 17, were taken into account by the revision tales for the first time and solely on the basis of oral testimony from the storytellers. In this group, the average absolute error was 1.93 years (a total of 23 discrepancies in 96 cases of double counting). In other words, the errors in determining the age of adult women averaged about two years, both in the direction of overestimation and underestimation, whereas in children they usually did not exceed six months.

Meanwhile, it cannot be ruled out that in some cases, fairy tales for both the old and the new place of residence were compiled by the same person from among the servants of the patrimonial apparatus, and individual landlords could have written accounting systems for peasant households and their inhabitants, information from which was subsequently transferred to audit materials. To check how much such theoretical situations underestimated the real value of the average absolute error, let us turn to the data on the mass transfer of 110 peasants of both sexes from the village of Nikolaevsky (Shuvatovo) to the village of Uspenskoye (Repevka), carried out by court counselor I. N. Morkov. In Shuvatov, future settlers were accounted for by an audit tale submitted to the Simbirsk provincial Chancellery on March 12, 1762 ^{[14, l. 363-376]}, after which they were re–recorded in the tale of the new residence, dated November 6, 1763 ^{[3, l. 304-325]}.

The peculiarity of this case lies in the fact that the probability of the influence of the theoretical situations mentioned above is practically excluded. Both storytellers were illiterate peasants, instead of whom outsiders "had a hand" in the tale – Vasily Leontiev, a servant of Morkov, and Vasily Dementyev, a plough soldier who lived in the village of Gorenki near Shuvatov ^{[3, l. 325]}; ^{[14, l. 376]}. According to the revision tale of 1762, the village of Shuvatovo was in the possession of the Life Guards of Second Lieutenant D. A. Nikitin, but by the autumn of 1763, Morkov had withdrawn a significant part of the peasants from his estate, claiming the inheritance of his deceased grandmother ^{[3, 318-322]}. Consequently, there was a property conflict between the two landlords, which made both the exchange of intra–patrimonial documentation and the mechanical copying of information from one revision tale to another unlikely.

The results of the comparison confirm the conclusions obtained based on the results of the general calculation based on the totality of the revision tales of 1762. In the sources under consideration, 27 boys and 23 girls aged 0-18 years were taken into account at both places of residence (the upper limit of the child's age was shifted by one year, since the tale of the new place was submitted in 1763). For boys, discrepancies in the indication of age were revealed in 23 cases, while the average absolute error was 2.03 years; In turn, the ages of girls differ in 17 out of 23 cases, and the average absolute error reaches 2.17 years. However, these figures cannot be considered as a direct measure of the inaccuracy of age information, because 1 year, 7 months and 23 days have passed between the submission of two revision tales – therefore, even with an ideally accurate age indication, the average absolute difference between the two records should be 1.65 years. After subtracting this natural time interval, the actual error value turns out to be significantly lower: 0.38 years for boys and 0.52 years for girls.

For comparison, data on women over the age of 18 were analyzed again. They were accounted for 39 times at both places of residence, with age discrepancies detected in 17 cases; the observed average absolute error was 3.5 years, and after adjusting for the time interval between the compilation of fairy tales, it decreased to 1.85 years.

Thus, the results of the analysis almost completely coincide with the indicators obtained from the audit materials of 1762: the average absolute error in determining the age of children is about half a year (0.38 years for boys and 0.52–0.66 years for girls), while for adult women it is close to two years (1.85–1.93 years), which means that This corresponds to the observations of researchers about the increased degree of female age accumulation among the Russian peasantry ^{[10, pp. 525, 527]}.

From all of the above, it follows that when determining the age of the same child from memory, different peasants, on average, gave almost identical estimates. At the same time, according to the results of calculating weather fertility rates based on revision ages, significant interannual fluctuations are found: with an average inter-revision fertility rate of 30.8%, the average amplitude of fluctuations is 7.2%, and in some cases the difference between neighboring years reaches 18-19% (Graph 1).

Graph 1. The dynamics of fertility in the male population between the II and III revisions

In relation to the era under consideration, such large–scale fluctuations can only be explained by the impact of extreme factors - crop failures, famines, epidemics and other crisis phenomena that can simultaneously increase mortality and reduce the proportion of children of appropriate ages in the gender and age structure of the population. However, a comparison of the data obtained with the dynamics of population decline does not confirm this interpretation: a visual comparison of the birth and death charts shows that there is no overlap between periods of increased decline and sharp falls in the birth rate (see graph 2). This is also confirmed by the correlation coefficient between the weather series of the two indicators for 1745-1762, which is practically zero (r=-0.005).

Graph 2. The dynamics of mortality in the male population between the II and III revisions

Thus, the revealed fluctuations in the birth rate cannot be explained by real demographic processes – with a high degree of probability they follow from the peculiarities of fixing ages in the revision tales and require a full-fledged analysis of the gender and age structure.

Age accumulation

In graph 1, the main attention is drawn to the fact that the sharpest birth rate spikes occurred in 1750 and 1752. This picture fully corresponds to the distribution of children's ages in the age-sex pyramid constructed according to the data of the revision tales of 1762, where the age groups of ten- and twelve–year-old male children are sharply distinguished in terms of numbers (see graph 3.1, as well as its abbreviated version, covering only the population aged 0-17 years - Graph 3.2). This configuration indicates the presence of age accumulation and is inconsistent with the demographic patterns typical of a traditional agrarian society. In the absence of the consequences of major catastrophic events, the number of one–year-old age groups should decrease smoothly and gradually as they move from younger to older ages.

Graph 3.1. The age-sex pyramid of peasants who migrated between the second and third revisions, as of 1762.

Graph 3.2. Fragment of the age-sex pyramid of 1762 (children 0-17 years old)

Age accumulation is more clearly revealed by calculating the ratio between the number of two adjacent age groups (Table 1). If the value of the indicator exceeds one, this indicates the predominance of this age group over the next, while values below one reflect the opposite situation. The most significant positive deviations of the coefficient from unity, exceeding its average value for the entire age range under consideration (they are highlighted in bold in the table), make it possible to identify the ages that were most preferred by storytellers and accumulated part of the population of neighboring age groups.

Table 1. The ratio of the number of this age group to the subsequent one according to the data of the third revision

As can be seen from the table above, both boys and girls were most strongly affected by age accumulation in the 10-year-old age group, while peasant children aged 9 and 11 were systematically and very significantly under-educated in its favor. In addition, pronounced accumulation was observed at the ages of 8, 12 and 15 years, to a lesser extent – also at the ages of 3 and 5 years. At the same time, it is interesting to note that the age of 15 stood out much more strongly among girls than among boys (2.05 versus 1.24). All this allows us to conclude that in the imagination of the peasants who submitted the revision tales, there was a set of six preferred age groups, to which they were inclined to include the children of their fellow villagers. At the same time, the degree of preference for individual ages varied depending on gender: for boys, it can be presented in sequence 10, 8, 12, 15, 5, 3, Whereas for girls – 10, 15, 8, 12, 3, 5.

Apparently, the accumulation of children's ages was more or less typical for the majority of population censuses conducted in agrarian societies with low literacy levels. So, V. V. Paevsky and S. A. Novoselsky noted an increased concentration of the population around the age of 12, observed in the Russian censuses of 1897 and 1926 ^{[15, pp. 46, 55]}. The UN Manual on Demography, published in 1955, cited the age-sex pyramid of Turkey according to the 1945 census as an example of accumulation, which shows population surges in the same age groups as in the revision tales of the third revision ^{[16, p. 34]}. Despite the prevalence of this phenomenon, it has so far practically not been the subject of a separate study in the demographic literature – as a rule, researchers have limited themselves to stating the very fact of child accumulation, noting a characteristic bias in favor of even numbers like 8, 10 and 12 ^[17]. One of the few exceptions is a special manual on the problem of age accumulation, published in 1981 by the publishing house of the National Academies of the USA, in which the distortions of the age structure of the child population were considered based on the materials of censuses and surveys of the population of African countries ^{[18, p. 48-59]}.

For what reason exactly 3-, 5-, 8-, 10-, 12- and 15-year-olds enjoyed the greatest preference among storytellers during the third revision, and what explains the special position of these numbers in the age system of the peasant population? To try to answer this question, it is necessary to refer to the works on the ethnography of the old Russian village, which dealt with issues of age categories, age terminology and the process of growing up ^{[19, p. 24 – 29, 53 – 59, 122 – 124]}; ^{[20, C. 137 – 151]}; ^[21]; ^{[22, pp. 108-119]}; ^{[23, pp. 22-35]}. The materials collected in them clearly demonstrate that almost all the ages identified in the revision tales as centers of age accumulation coincide with the most important stages of labor socialization of peasant children.

Ethnographers have studied the stages of boys' growing up in the most detail. At about the age of five, they began to carry out the simplest household tasks, including looking after younger siblings, and the next important age milestone was at the age of 8, when the child began systematic development of peasant labor and began to participate in field work as a "waxer" or "harrower", driving a horse while harrowing. This period of study was divided into two stages – preparatory and basic, the boundary between which passed either at 10 or 12 years old. From that time on, teenagers began to take part in latrines, hire themselves as helpers, graze cattle on their own, fish, and master more difficult types of agricultural work – plowing, harrowing, mowing, and threshing. Finally, by the age of 15, young men turned into full-fledged workers who possessed all peasant labor skills.

Concluding the conversation about the problem of age accumulation, we will briefly focus on the methods of statistical correction of data that make it possible to smooth out fluctuations in the age structure and, as a result, the reconstructed birth rate dynamics for the period between 1745 and 1762. To solve such problems, demographers have developed both special mathematical methods ^{[15, 24]} and simpler ways of smoothing data that are available for practical use by historians of the humanities. These include combining ages into enlarged 5- and 10-year age cohorts ^{[10, p. 528]}, as well as using the moving average method with a three-year or five-year interval, which is based on replacing each value of a series with an arithmetic average of several neighboring values (for example, with three-year smoothing, the indicator value for each year is replaced by the average the value of this year, the previous and subsequent years) ^{[25, pp. 115-1120]}. A sample of the application of this method using a three-year step is shown in graph 4.

Graph 4. Smoothing the birth rate dynamics using the three-year moving average method

Conclusions

Summing up all the above, we note that the use of Revision III fairy tales to study the dynamics of fertility requires taking into account a number of specific distortions associated with the peculiarities of fixing children's ages. A comparison of the cases of double counting of migrants indicates that, on average, peasants estimated the age of the same children equally, and the discrepancies between their estimates were relatively small. Paradoxically, the downside of this uniformity was not a high degree of accuracy, but the effect of age accumulation – remembering the ages of the children of their fellow villagers, illiterate storytellers sought to assign them to one of the 5-6 age groups, which, as a rule, coincided with the main milestones of labor socialization known from ethnographic literature (5, 8, 10, 12, 15 years old).

At the same time, the use of statistical smoothing methods makes it possible to significantly compensate for the impact of such distortions and identify the main trends in population reproduction.

The article is published in the version approved by the reviewers (after receiving a positive review recommending the manuscript for publication) with corrections made by the author (after receiving the editor’s comments, if any).
Read all reviews on this article