Abstract
Despite the laborious process of gathering the data and some limitations due to selective incompleteness, catasti and estimi (fiscal cadastres) provide precious information on demographic differences within Italy during the Late Middle Ages and early Modern Age. The authors compare three areas: the countryside surrounding Lucca in 1411 (Tuscany, 5,792 individuals) and Varese in 1530 (Lombardy, 2,703), and the “quasi-city” of Legnago in 1430 (Veneto, 2,101), reconsidering also the Florentine catasto of 1427. Family-type distribution, age at marriage, and residence after marriage are noticeably different, showing that the territorial diversity in family of nineteenth-century Italy was already present four centuries earlier.
Introduction
Despite recent investigations which demonstrate the possibility of reconstructing demographic trends using baptisms alone, 1 a greater understanding of the Italian population dynamics during the fifteenth and sixteenth centuries is made difficult by the scarcity of parish registers, one of the most important primary sources employed in studies of the Italians demographic ancien régime. The scarcity of status animarum also negatively conditions demographic analyses, limiting the latter, for the most part, to the end of the 1500s.
There are, however, other quite useful sources of fiscal nature (property tax registers, usually called estimi or catasti) that can be employed in the study of population trends prior to the 1600s. The Florentine catasto of 1427 is among the most notable and cited of these sources—in large part because it covers much of the Tuscan region (with the exception of Lucca, Siena, and Massa Carrara). 2 Investigations of demographic phenomenon using fiscal registries have, however, also been carried out with regard to other areas. For example, a number of important works have been published using censuses from Lombardy, especially those concentrated in the period of 1530–45. While Besta pointed out the importance of these sources early on, 3 it was not until more than a decade later that these sources were examined using a demographic approach. 4 However, following these early studies, fiscal surveys remained relatively untouched by historical demographers, almost as if there was a shared reticence to use estimi for the study of economic history. 5 It was not until the 1980s that scholars returned to the demographic analysis of fiscal sources. 6 More recently, Leverotti examined family arrangement and the role of servants in Italy’s Po Valley, underlining the utility of estimi for the reconstruction of family structures. 7
Fiscal surveys always raise doubts concerning possible selection effects in terms of families and individuals. Such reservations—as we address in this article—are rightfully an issue. That said, however, the catasti remain a precious and rich source of information; an attentive examination allows for a better understanding of the Italian population and household arrangement during the fifteenth and sixteenth centuries. More specifically, we study three different areas in the center-north of Italy: the countryside surrounding Lucca in 1411–13, the “quasi-city” 8 of Legnago (Verona) in 1432, and the countryside surrounding Varese in 1530. 9 Our aim is to compare these different sources and areas through use of the same technique, such as to better demonstrate data quality issues and provide demographic results.
Sources
The area surrounding Varese (the so-called Varesotto; henceforth Varese: 2,703 individuals and 846 families). Fiscal survey data on bocche da sale (literally mouths eating salt) from 1530 are available in the Comuni (Comunes) and Censo (Census) funds of the State Archives of Milan, although they are rather scattered and of various levels of completeness. In this article, we favor the most comprehensive censuses, thereby analyzing four pievi (fiscal jurisdictions that included several communities) within the Duchy of Milan: Varese, Angera, Brebbia, and Valcuvia. 10 Together, these pievi occupied a relatively compact geographic area made up of the so-called Lombard-Piedmont Pre-Alpine belt, the land surrounding the city of Varese (not included in this analysis), and the area that spreads from the former to the banks of nearby Lake Maggiore. Documentation relative to the four pievi—compiled between September and November of 1530 by the Commissioner Cristoforo Crivelli—includes a list of the inhabitants of each community, grouped by household, as well as indicates the age of each individual, their relationship to the head of the household, the profession of the former, and his ownership of oxen and cows. The sex of each individual is discernable by name, age is almost always indicated, and parental ties are easily discernable by a rigid reporting of family members always listed in the following order: head of the household, spouse, male children, wives of the male children and their children, female children, and so on.
Legnago (2,101 individuals and 415 families). Chiappa et al. published a study which contains—in addition to a detailed description of the source—the complete registry of Legnago’s catasto from 1430 to 1432. 11 The original register, in good condition, is preserved in the ancient municipal archive of Legnago. This document was drawn up with the objective of securing Verona’s control over Legnago—the most populous of the Veronese contado—which, in 1405, wished to become part of Venice’s dominion and thus became a troublesome matter for the Scaliger family (Lords of Verona) and the city’s elites. More specifically, in 1425, the Council of Verona imposed the renewal of the catasto, under the supervision of Aleardo Gafforini (a prominent jurist and the Scaliger’s ambassador to Venice). Despite great opposition from Legnago’s municipal council, preliminary steps in the process of collecting a count of people and animals began in 1430. The procedure employed in producing the fiscal report generally consisted of either a declaration on the part of the taxpayer or an inspection committee delegated to the purpose (or multiple commissioners, generally three to five, with eventual elimination of the maximum and minimum estimates), followed by registry in an account book, and thus the assignment of a category for fiscal purposes. More broadly, Legnago’s catasto includes both information concerning the possession of real estate and land ownership and data relative to commercial activities and the transformation of raw material. Demographic data reported in the catasto are similar to those mentioned above for Varese: age, relationship to the head of the household, the profession of the former, and ownership of oxen and cows. As for Varese, sex, age, and kin relation to the head of the family are always reported or easy reconstructed.
The countryside surrounding Lucca (henceforth Lucca: 5,792 individuals and 1,312 families). Lucca’s catasto from 1411 to 1413 can be considered a true population survey. 12 In comparison to cadastres from the previous century—which include only the first and last name of the head of the household, names of fraternal siblings, and adult male children, and a generic reporting of the presence of wives and young children—the catasto of 1411–13 provides, for each family, a detailed description of each member, including the name, age (with some exceptions), and relationship to the head of the household. The survey also reports the presence of non-kin members and servants (famuli). Both native families and foreigners (whose origins are also noted) are included, even those recently settled and therefore exempt from fiscal duties. In addition to the survey of the population, the catasto contains information concerning personal property and real estate belonging to each family, such as land and houses (whether owned or rented) as well as ownership of animals and farming equipment. Collection of the data was carried out by officials nominated by Paolo Guinigi, the ruling Lord of Lucca, in an attempt to assure the veracity of the survey and hence reliably measure the effects of war, pestilence, and famines on the contado. More broadly, the city of Lucca hoped to measure demographic weight and population movement in order to begin a restructuring of the administration. The catasto is made up of twenty volumes, the majority of which are quite legible. That said, however, for a few of pivieri (districts), name and age are reported only for the head of the household and the first born son, along with a general indicator of the total number of family members (i.e., bocche, or mouths). Data used in this article thus concern only those families for which it was possible to reconstruct the typology (nuclear, extended, etc.). We further reflect on possible selection mechanisms caused by this decision in the conclusion. In addition, for 12.7 percent of the people—mainly those belonging to the same family or living alone—age is not reported (Table 1) and thus these individuals are included only in the analyses where age is not an issue. Families of individuals with unreported ages are mainly of recent immigration; despite this difference, the mean number of household members is similar to that of families for whom the data are complete.
Proportion of Missing Data for Sex and Age (Percent)
Population
Age Heaping
As is common in population surveys from the ancien régime, there is a strong attraction effect of ages ending in 0, and to a lesser degree, those ending in 5 (with the exception of Lucca): from 30 to 40 percent with regard to the former, and about 15 percent with regard to the latter (Figures 1 and 2). The age heaping begins around age 20 and continues up to the oldest ages and affects both men and women. For Legnago, we also observe a strong concentration around the age of 18, perhaps due to fiscal rules.
Population by sex and age (absolute values).
Percentage of people aged 3–72 by last digit.
Sex and Age Structures: Some Plausible and Selective Underreporting
There may be some selective underreporting by sex and age (Figure 3). Varese has very few children, as shown by an age pyramid which looks more characteristic of contemporary wealthy population than of communities prior to the demographic transition. The relative lack of children could be due to the fact that youth under the age of 4 were exempt from the tax on salt. This rather surprising result may also reflect the effects of the Italian Wars (1494–1559). In 1530, Lombardy had just come out of a thirty-year period of war conducted in loco (1499–1529), and had experienced the height of an economic, social, and demographic crisis precisely in the years immediately preceding this date (the period 1522–29 and especially the biennium of 1528–29 were marked by plague and famine). 13 The limited presence of children, the demographic imbalance between men and women, and the concentration of the population in the 20–50 year age range could thus in part be due to selective mortality (disfavoring, in particular, women and children) caused by such calamitous events. The high proportion of individuals in the 20–39 year age range is likely due to numerous famuli (servants): eighty individuals are classified as such (or 3.0 percent of the total population, thirty-nine males and forty-one females, 75 percent age 15 or older).
Age and sex structures.
Lucca, in contrast, has quite a large number of children in the 0–9 age group, perhaps the result of a postepidemic “recovery,” given the relatively young average age at first marriage (see also Table 4), and the plague outbreaks of 1389–90, 1399–1400, 1406, and 1411. 14 The rather small number of men age 10–19 can be linked to servant activities or servitude (famulato), for example, the practice of working as a shepherd for peasant families in often far-off locations. Famulato was typically carried out during one’s youth, ending before the age of 20, and thus having little or no impact on age at marriage. 15 Within the families of Lucca, famuli are, however, practically absent: only thirty individuals are classified as such (0.5 percent of the total population, twenty-one males and nine females, 50 percent age 15 or older). Interestingly, the age pyramid of the population inhabiting the countryside surrounding Lucca in 1411–13 is very similar to that of the population living in the countryside around Florence in 1427. 16
The age structure of Legnago is instead more regular and similar to that commonly observed before the demographic transition. The proportion of famuli lies in between the levels observed in the other two communities: the catasto includes fifty-two individuals (or 2.5 percent of total population, twenty-seven males and twenty-five females, 50 percent age 15 or older).
Thanks to the availability of data by family, which includes each member’s age and parental relationship to the head of the household, it is possible to measure the potential underreporting of children using the own-children method. 17 While a detailed description of this technique can be found in the aforementioned studies, below we provide a graph (Figure 4) which reflects the estimated average number of children in the time period between a year before each survey and fifteen years before. We assume a mortality rate (of mothers and of infants) equal to a life expectancy at birth to twenty-five years (West Model of the standard life tables by Coale and Demeny). 18 In order to provide a better picture of the effect of this hypothesis on mortality, we calculate that the total fertility rate (TFR)—for each year—would have been greater or less by one child, if life expectancy at birth was five years lower or higher. Moreover, in the unlikely hypothesis of the absence of mortality for both children and mothers, then the TFR in the fifteen years preceding the survey would have been on average 4.85 for Lucca, 5.01 for Legnago, and 1.87 for Varese.
Fertility estimates by the own-children method for the years preceding the survey. 3-terms moving averages. Note: Mortality estimate corresponds to e0 = 25.
The results depicted in Figure 4 suggest that not all children born in Varese were reported in the fiscal survey. It is, in fact, highly unlikely that the fertility rate was less than four children per woman given that—as we further discuss below—average age at first marriage for women was around the age of 18. That said, however, if infant mortality was high enough—as could occur during very difficult times—then a TFR calculated using the own-children method could be underestimated.
Results for Lucca similarly raise suspicion as to the presence of underreporting, given both very high fertility in the years immediately preceding the catasto (10–12 children per woman) and very low fertility during the 13–16 years leading up to the survey (around 2 children per woman). However, high birthrates in the years immediately prior to the catasto may quite accurately reflect true fertility levels in the case of a postepidemic recovery, accelerated by very high female marital rates (as calculated below, the average age at first marriage for women was sixteen). Results for Lucca again resemble those for Florence in 1427, this time in terms of high fertility. 19 After all, in Florence as well, individuals under the age of 15 made of 45 percent of the population, age at first marriage for women was around 16–17, and the population was rapidly “recovering” lives lost during a severe plague in 1423–24. On the other hand, very low fertility rates observed in the years farthest from the catasto could be misleading due to the precocious age at which youth left the parental home. In fact, estimates of the TFR using the own-children method consider only mothers with a cohabiting child, and at the beginning of the 1400s half of female sixteen-year-olds in Lucca were already married and thus were not counted as children.
Results from Legnago seem more realistic, given both relatively little fluctuation over the course of the fifteen-year period under consideration, and a TFR (6–9 children per woman, according to level of mortality) potentially quite close to actual fertility, given that age at first marriage for women was both fairly contained (20–21 years of age) and marriage was practically universal.
An analysis of the sex ratio (ratio of males to females, considered at different ages) for the population under the age of 20 provides further evidence of underreporting, especially for young women in Varese (Table 2). In the other two areas, the sex ratio at infancy and young age is also higher than the standard biological sex ratio at birth (105 males to every 100 females) but at relatively contained levels. The obvious underreporting of female infants is another aspect the catasto data examined here have in common with the Florentine catasto, where the sex ratio at age 0–4 is equal to 100 × M/F = 120, 20 not unlike the sex ratios observed in Lucca and Legnago. Whether due to an underreporting of children or the supermortality of women, this suggests that in late medieval Italy, young men were more highly regarded and cared for with greater attention.
Population by Five-Age Group and Sex and Sex Ratio M/F over the First Twenty Years of Life
The sex ratio for Varese and Legnago is more balanced for adults and the elderly (Figure 5), while in Lucca adult men and the elderly are clearly more numerous. In what follows, we further discuss this finding in terms of population composition by age and civil status.
Sex ratio of age classes (M/F Percent).
Marital Status, Age at First Marriage, and Difference in Age between Spouses
Marital status, especially for individuals within complex families, is discernable through an analysis of kinship ties. This was always possible for women, in that they are generally “dependent” upon a man. Men who live alone are instead classified here as single, although they could also be widowers. That said, the distribution of the data by civil status seems plausible, perhaps with the exception of women older than age fifteen in Lucca, where unmarried women make up only 4.5 percent of the population (Tables 3 and 4). Underreporting may have had a greater effect on unmarried women in this area, or perhaps many were employed as servants in the city.
Total Population and Population Aged 15+ by Sex and Marital Status
Note: aOnly for those localities where the age of each individual is indicated.
bThese numbers include seventeen male adults who have an indeterminable marital status.
Substantial underreporting of married or unmarried women or widows could lead to an underestimation of the average age at first marriage in Lucca, obtained here using Hajnal’s method. 21 However, the youthfulness characterizing women’s age at first marriage might also reflect actual demographic dynamics at the time, given Lucca’s close resemblance to the Florentine contado in 1427–29, 22 and several areas in the Parma contado at the beginning of the fifteenth century. 23 We may in fact be observing a Malthusian response aimed at recovering the population gap left by the plague of 1406 and 1411. Results show that age at first marriage for men—less affected by problems of data incompleteness—was 22–23 in Lucca and Varese and slightly higher in Legnago (24–25). This level for Legnago is not too far from the one published by Herlihy for Verona (Catasto of 1425). 24 The higher age of Legnago could be due to the semi-urban nature of the latter and the necessity of learning a profession before wedding, thus delaying marriage. In the countryside surrounding Lucca and Varese, agricultural labor did not require a long period of apprenticeship.
In the three areas under consideration, marriage is virtually universal for women while some men remained unmarried (especially in Lucca) perhaps due to a scarcity of women. It should be pointed out, however, that this source does not include the religious population.
For Legnago alone, we calculated age differences between spouses, although it was not possible to distinguish between first and successive marriages (Table 5 and Figure 6). The average age difference between husbands and wives is 4.9 years. This is about 1–2 years greater than the average difference in age at first marriage calculated using Hajnal’s method and may be linked to greater age differences in second marriages. That said there is great variability among the couples. In more than one third of the cases, the age difference between spouses is greater than 10 years, while marriages in which the woman is older than her husband are les common, but not unusual (Table 5). These percentages are not all that different from the contado of Parma at the beginning of the fifteenth century, where 42 percent of the 1,730 couples were characterized by an age difference of 1 to 5 years, and 26 percent by an age difference of 6 to 10 years. 25 Further research might attempt to verify whether such differences were typical or prevalent among certain age groups, such as to better understand the influence of external events such as plagues and famine on marital models.
Combined ages of spouses, Legnago. Note: The diagonal line represents spouses of the same age.
Percentage of Never Married at Age 50 and Average Age at First Marriage for Men and Women a
aThese items are based on polished data by moving average.
Age Differences in Years (D) between the Spouses (Husband – Wife). Percentage Distribution in Legnago a
aTwo cases lacking the age of one spouses are excluded.
bIdentical ages are likely overestimated due to the tendency to round off ages.
Family
Dimension
As the number of women and children certainly seems to be underreported in Varese, the number of family members and the proportion of extended and multiple families may be underestimated. On the other hand, an average of 3.2 family members may not be unrealistic (Table 6). Larismont Pergameni’s above-cited study of fiscal survey data from 1545 concerning bocche and biade (mouths and fodder) reveals that in the hills of Lombardy there were an average of 3.3 individuals per family (the area covered includes the pievi of Angera, Appiano, Casate, Castelseprio, Fino, Garlate, Leggiuno, Mariano, Somma, and Varese). 26 On the contrary, the pievi located on the plains—studied using the same source—had a larger average family size of 4.9. Differences between the plains and the hills are also observed by Vitaletti in his work on fiscal surveys from 1537 in the same area. 27 Such studies support the hypothesis that as one descends from the mountains to the plains, family size increases. Greater size seems particularly tied to profession and socioeconomic conditions. Discriminating factors may also include the nature of farming contracts, the size of the terrain cultivated, and the distribution of cereal stocks. Di Tullio finds a very close relationship between family structure, profession, and contractual typology (interpreted as the possession of farming apparatus and the size of cultivated land plots). 28 In many communities studied by Di Tullio, average family size is not much different than that observed for Varese. Farming families tend to be the largest and more commonly consist of extended families, although this is always relative to the size of the area farmed. In Varese, land plots were not large and farming families tended to be small; tenants and day laborers were more diffuse and characterized—as also seen in other studies of Lombardy—by smaller and simpler structures. It should also be mentioned that the famulato (both life-cycle servants and long-life service) was an important means for demographic reallocation. 29 Such circumstances may have contributed to limiting the number of family members and the proportion of complex families, even compared to Legnago and Lucca. More generally, families in Legnago and Lucca tended to be larger than in Varese, due to a greater proportion of multiple families (see the following section as well). In Lucca, the mean number of members is the same for families with complete information (i.e. sex, age, and relation to the head of the household for each family member, for 1,312 families and 5,792 people, see Table 6) as for families for whom it was not possible to entirely reconstruct the aforementioned indicators (i.e., 471 families and 2,040 people). 30 This result reflects the generally good quality of the data for Lucca. Moreover, the average number of family members in Lucca in 1411–13 is 4.4, exactly the same as that observed in the Florentine catasto of 1427. 31
Family and Individuals Classified by Typologies
aPeople living in the different family typology.
Family Structure and the Cycle of Family Life
Family typologies are quite similar among the three areas under consideration. Multiple and extended families and nuclear families are all present in considerable number: roughly half of the people lived in complex families and the other half in simple families (Table 6). Once again the distribution of family types in Lucca is quite similar to that of the Florentine catasto of 1427. 32 In Varese, the multiple families are clearly less common.
In what follows, we endeavor to broadly reconstruct the cycle of family life in Lucca, classifying individuals by sex, age, civil status, and family typology (Figures 7 and 8). A significant proportion of newlyweds remained in the parental home upon marriage, most commonly that belonging to the husband. Shortly thereafter, however, couples were obliged to form their own home, due to the death of older members. Finally, individuals who managed to grow old often lived with their children or with other younger relatives.
Percentage of people living in nuclear families, by age classes, Lucca.
People by sex, age, marital status, and family type, Lucca. Absolute values.
The custom of patrilocal residence upon marriage was more common in Lucca than in the other two areas. The use of a technique similar to that of Hanjal’s SMAM (mean age at first marriage), allows for an estimation of the average age at which men became, for the first time, head of the household. Because this approach includes only men older than the age of 17, any difficulties related to underreporting—more problematic for children and females—should not significantly affect the outcomes (Table 7). Results reveal an average age of 34.5 for Lucca, more than 10 years older that the average age at first marriage. The same value for Varese is 24.5, only slightly greater than the average age at first marriage, suggesting a widespread practice of neolocal residence. The result for Legnago falls somewhere between the first two areas with an average age of 29.5, five years older than the average age at first marriage.
Percentage of Men Head of the Household by Age (Male People Aged Seventeen or More)
Note: aEstimate of the mean age at which a man became head of the household for the first time.
bEstimate of the mean age at first marriage (see Table 4).
A brief consideration of the socioeconomic context further helps to explain differences between these three areas. In the farming community of Lucca, land was the primary source of labor, and married children did not leave the parental home unless the latter was bereft of basic goods. 33 Both the presence of a lake in Varese and the proximity with the city of Milan favored instead the possibility of other kinds of employment—for example, fishing, transportation, construction, manufacturing, and servants activities—allowing for the formation of autonomous families immediately following marriage. 34 In the semi-urban context of Legnago, the necessity of achieving professional independence may have delayed the formation of autonomous nuclear families. 35
Conclusions
Our research confirms that this kind of data is unable to provide a complete picture of the complexity and significant differences in family types and demographic dynamics characteristic of the ancien régime. It does, however, provide a snapshot of the family at a precise moment in its life cycle. In addition to considering the effects of earlier social and demographic events (famine, plagues, war, mobility, famulato, and other issues only briefly touched upon here), one must account for the geography of the area, the distribution and types of settlements, their economic organization, the presence of local material assets, and so on. Of no less importance are issues related to the quality of the data, raising questions somewhat difficult to disentangle. That said, in what follows we summarize the results obtained thus far, and in doing so discuss the novel aspects this source has to offer to investigations of the population living in central and northern Italy between the fifteenth and sixteenth centuries.
Data on Individuals
The probable underreporting of never-married women and children underlines the care that must be taken in using medieval fiscal sources (existent far earlier than emergence of a culture of gathering administrative and statistical data) for demographic purposes. It is consequently quite difficult to understand the extent to which profound differences in the sex and age structures of Lucca, Varese, and Legnago are truly due to diverse demographic dynamics, or whether they can instead be explained by an underreporting of children (especially female) and unmarried women. While the underreporting of children (especially female) for Varese in 1530 is quite evident, data on age classes, sex, and civil status for Legnago in 1430–32 seem to be of quite good quality. Results for Lucca—while somewhat surprising in suggesting that 45 percent of the population was under the age of sixteen—are almost identical to those from the Florentine catasto of 1427, a source considered of good quality. Generally speaking, the quality of data of Lucca does not seem lower than the quality of Florentine catasto. The estimo of Lucca of 1411–13 can thus be considered a model for the Florence data collection of 1427, sixteen years after.
The regular distribution by sex and age class in Legnago reveals a population with relatively high fertility (6–8 children per woman) throughout the 15-year period preceding the cadastre. In Lucca in 1411–13, fertility (as in Florence) during the 2–3 years immediately prior to the catasto must have been much higher than in Legnago (even as much as 11–12 children on average per woman), enough to guarantee a rapid recovery of the considerable loss in population due to plagues that occurred several years prior to the census.
Data on Families
The quality of the data on age at marriage, family structures, and the cycle of family life are satisfactory for the three areas, especially for men. In the countryside surrounding Varese and Lucca, men married for the first time at more or less the same age (around twenty-three). However, men from Varese tended to form an autonomous nuclear family, while those from Lucca stayed in the parental home after marriage, often waiting as much as a decade to become head of the household, perhaps upon the death of their father. Legnago is characterized by yet another pattern: men married later and couples spent—on average—an initial period of five years cohabiting with their parents. These findings are in line with results concerning different types of family structures across the three areas. The nuclear family and to some degree the extended family were more widespread in Varese, whereas in Lucca and Legnago, multiple families were more diffuse, although the nuclear family was—all in all—the most common typology, and the extended family was also well represented. Family size and typology in Lucca was very similar to that observed in the adjacent territory subject to Florence.
We have to wait until the middle of nineteenth century to have reliable, territorially detailed, and comprehensive census on the Italian population. Using these data, scholars demonstrated that the diversity of demographic and familial regimes within central–northern Italy was much greater than that observed in other European countries. 36 Significant differences in household structures in the areas surrounding Varese, Lucca, and Legnago between the fifteenth and sixteenth centuries suggest that—in all likelihood—a large variability within Italy was already quite present at the end of the Middle Ages. These demographic studies could be extended to other fifteenth- and sixteenth-century communities. Differential analyses according to economic condition could be carried out thanks to data contained in the cadastres (but not employed here) and would allow for a better understanding of the great capacity of Italian families during the ancien regime to modify their rule of formation, structure, and size in order to adapt to diverse economic and social contexts.
Footnotes
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.