Evaluating the Slavonian Census of 1698

Part I: Structure and Meaning

E. A. Hammel

Kenneth W. Wachter

Department of Demography

University of California

Berkeley, CA 94720


(510) 642 9800 [voice]

(510) 643 8558 [fax]

gene@demog.berkeley.edu [email for Hammel]

wachter@demog.berkeley.edu [email for Wachter]


Hammel, E. A. and Kenneth W. Wachter, 1995, The Slavonian Census of 1698. Part I: Structure and Meaning, European Journal of Population.

Microsimulation, other demographic tools, and evidence of history and ethnography are used to evaluate an important 17th century household census. Linguistic, ethnographic, and internal evidence allow adjustment of anomalies in census categories. Microsimulation based on historically and ethnographically plausible rates and household formation scenarios produces simulated households in accord with those of the adjusted census. Results permit estimation of the true population of the region, of the kinship and age composition of households under frontier conditions, and the probable future composition of households as the frontier stabilized and land shortage began to exert pressure for greater density and household complexity. Part I concentrates on historical, ethnographic, and linguistic evidence.

Keywords: historical demography, family and household, kinship composition, microsimulation. 1. Introduction Historical census returns are a precious source of clues to the social and economic structure of the past. To be sure, older censuses and population counts usually include only certain categories of people, and multipliers and adjustments based on demographic theory have been employed from Hume (1752:419) onwards to infer total population size and structure from information on included subsets. Brunt's (1971) estimates of the total free population in ancient Roman Italy from the Roman Republic's census counts of adult males and Laiou-Thomodakis' (1977:72-107, 223-66) analysis of household formation and refined use of household coefficients are classic examples. Until now, however, such indirect estimation has generally used nothing more than information on age and sex and sometimes legal or marital status, even though many surviving nominative household listings are also rich in kinship information, household by household. The restriction to age and sex reflects the long-standing reliance on stable population theory, which takes no account of marital linkages or lateral kinship ties among members of a population.

Indirect estimation is always a two-way process, back and forth between calculation and textual interpretation. Brunt's (1971:117-20) calculations with proportions by age and sex led him to conclude that the Roman Imperial Census under Augustus must have included women and children and excluded infants; this conclusion induced him to reinterpret the words introducing the counts in the ancient sources. Likewise, our calculations suggest new interpretations for some of the language in the 1698 census documents.

The process of inferring totals which the census does not give is closely related to the task of assessing the plausibility of the information which it does give. With the 1698 census there is pervasive uncertainty as to whether kin of certain types are absent in the census because they were not counted or because they were not present, or because the kinship terms mean more than they seem. Such questions can never be settled with certainty. However, in this case, we believe that we have found reasonable solutions to the major puzzles raised by the census, solutions which are consistent with the data and compatible with the constraints of demographic plausibility. We would not of course deny the uncertainty of this and similar historical enterprises that are obliged to navigate, often by a kind of dead reckoning, between conflicting dangers in a fog of obscure information, enjoying only an occasional glimpse of a guiding star or a headland in the data.

This analysis is in two separately published parts (at the suggestion of the Editor and anonymous reviewers). The first part describes and evaluates one of the most important and controversial censuses of early modern Europe, the 1698 Austrian Census of Slavonia in what has recently become the independent country of Croatia (1991). Historical, ethnographic, and linguistic information are the core of this evaluation and reveal serious problems in any literal interpretation of the listings. The second part develops a methodology of re-estimation of the structure and membership of censused households, using microsimulation techniques, and goes on to estimate the partly counted and uncounted population of the region in 1698. While the approach here employed is of course specific to the empirical data at hand, we expect that the general approaches in these papers can be applied quite widely to nominative census listings from the past, as long as comparable independent information is available. 2. The Census The census which we examine here is particularly important, because it is the first in the newly reconquered territory of Slavonia after the ejection of the Ottomans. Fragments were published by Smiciklas (1891a), a fuller transcription by Mazuran (1988), and a yet more extensive tabular summary by Mazuran (1993). Narrowly defined, Slavonia is a triangle lying between the Drava River on the north and the Sava River on the south, its eastern apex now ending at the border with the region of Srem, roughly defined by a line between the western end of the Fruska Gora hills and the Drina-Sava confluence, its western base along the Ilova River at the modern border of Moslavina. The location of the districts on which this analysis is based are approximately given on the map (Fig. 1).

We address the issue of household composition apparent in the census and use the tools of historiography, ethnography, and demography to inquire into the reasonableness of the census and what it may tell us about life on the edge of Catholic Europe in a period still feudal in its economic and political organization. (Civil serfdom was abolished in Croatia-Slavonia in 1848; military serfdom in the Croatian border facing the Ottomans ended only in 1871, and full civil status was not achieved until 1881.) Because the 1698 census is regarded as a foundation for the understanding of the social and demographic history of this historically pivotal frontier region (Gelo and Krivosic 1990, Mazuran 1988, Stipetic 1988), it is important to maximize the amount of quantitative information that can be gleaned from it and to assess which aspects of the text can be taken at face value and which require deeper interpretation.

There are two facets to the importance of the document. First, it was taken only 22 years before the date of the first parish records that are still preserved from central Slavonia (Cernik parish) and provides a baseline for analysis of a corpus of over 200,000 baptisms, marriages, and burials c. 1714-1900 that show early evidence of fertility control (Hammel 1984, Hammel 1985, Hammel 1990b, Hammel 1993, Hammel 1995, Hammel and Kohler 1995). Second, it provides evidence for the construction of the historical legitimacy of the modern Croatian state in the region or conversely for counter-claims by other ethnic groups that would deny that legitimacy. The ethnic heterogeneity that led to the collapse of the Yugoslav state in 1991 is already manifest in the census of 1698.

Some "reading" of this census is unavoidable. No scholar imagines that it can be taken as a complete listing of all members of the population. It must be that some categories of persons have been excluded. No one is listed as a wife, and the number of listed women per listed man is only 0.26. The average number of listed "sons" and "daughters" per household is only 1.47, which is much too low for a high-fertility population like this one, as evidence from parish records c. 1720-80 suggests. Historians generally assume that children under age 15 were not listed and that married women were not listed unless they were widowed heads of households, but they caution that the social evaluation of age may have played a role. Our analysis, spelled out in later sections, lends support to this view. But historians have not come to grips with what other categories may or must have been excluded, and with what the total membership of households should have looked like, counting unlisted and listed members together. These are the questions we address.

Throughout, we depend heavily on an extensive corpus of historical and ethnographic investigation, some of it going back to as early as the 1760s in Habsburg Croatia or to the 14th Century in mediaeval or Ottoman Serbia, but especially on materials published by Croatian ethnographers since 1896.[1] These materials are especially consistent in their depiction of traditional household formation, and they accord well with the anthropological literature on patrilineally organized societies.

Initial scrutiny of the recorded census reveals two great puzzles in a territory characterized then and later by the strong patrilineal extended and fraternal joint household organization at least later typical of this part of the Balkans.

* Why are there no brothers listed as coresident kin in the first 14 (eastern) districts visited by the census commission, whereas brothers are listed in the last three (western) districts (the Brothers Puzzle)?

* Why is the number listed as "daughters" so sizable a fraction of the number listed as "sons", especially if daughters only included the unmarried and sons included married and unmarried alike (the Sons and the Daughters Puzzle)?

Other, related puzzles also exist, for example the complete absence of listed sisters or sons-in-law, and these will also be addressed. 3. Historical Background The Croatian tribes occupied the basins of the Sava and Drava rivers and, according to some claims, also the territory between the Drina and the Adriatic (i.e. Bosnia-Hercegovina) around the 8th or 9th century. They were closely related linguistically to other Slavic groups in the Alpine regions even westward into the Tyrol (now the Slovenes), to others still further north and northeast (now the Czechs and Slovaks) and to still others to the south and southeast (now the Serbs). The Germanic advance to the east after the 9th century cleared most of the Alpine zone of Slavs (except in modern Slovenia and Carinthia) and separated the Croats from the Slavs of Bohemia, Moravia, Slovakia and the Ciscarpathian Ukraine. Beginning in the 10th century the Croats came under increasing Magyar influence and by 1102 had recognized Hungarian suzerainty. Habsburg knowledge and administration of Slavic groups was confined to the Alpine (Slovenian) and more northerly Slavs. It is important to note that there is no historical evidence for joint family structure among the Alpine Slavs (Winner 1971); these Slavs seem to have had nuclear or stem family households, although the tradition of joint family organization has very deep Indo-European roots. The Alpine Slavs may have adopted nuclear and stem organization quite early, perhaps for ecological reasons. We have no direct historical evidence on family organization in Croatia between the 12th and 18th centuries, but we do have good evidence from documents of the 14th and 15th centuries that joint family organization was well established among Orthodox groups. After the Ottoman victory at Kosovo in 1389 and more northerly conquests up to the fall of Bosnia in 1493 and of Belgrade in 1521, Orthodox (or what would now be called Serbian) refugees began to flow into what is now Croatian territory. The degree to which the Habsburgs became familiar with such organization is unknown; they played no administrative role with respect to such groups until after the collapse of the Hungarian state in 1526, and clear evidence of their cognizance of joint family organization is not apparent until 1737 with the institution of the Hildburghausen reforms (see below).

The Ottoman advance deep into Pannonia after their overwhelming victory at Mohacs in 1526 is thought to have emptied Slavonia of much of its Croatian Catholic peasantry. From Bosnia and Serbia into this population vacuum between 1526 and 1683 flowed large numbers of islamicized Slavs, Turks, and other ethnic groups. Some unknown and disputed proportion of the native population remained. There also came large numbers of Orthodox (and some Catholics) of pastoral economy, known as Vlachs. The word "Vlach" has complex meanings. Here it is best understood as "pastoralist" or "inmigrant." Most Vlachs were Orthodox, but some were Catholic, a distinction that in modern times would be read as Serbian versus Croatian. (See Rothenberg 1966 :16). The remaining Croatian nobility on the fringes of the Alpine zone (e.g. near Graz) and west of the Ilova to the Adriatic, thence around the shoulder of Bosnia about to Zadar began to organize a military defense zone about 1525 in the vicinity of Senj, into which they invited refugees from Bosnia, many of whom were Orthodox, to serve as military serfs who received land in return for perpetual military service (Fig. 2). At the same time (1526), the Croat nobles recognized the suzerainty of the Habsburgs. By the last part of the 16th century administration of the Border fell to the Inner Austrian Estates (Styria, Carniola and Carinthia), although the Croatian feudatories continually struggled to maintain their own control of the region. By the 1680s the military zone included the highlands from a point north of Knin and west of the modern border of Bosnia north to the Kupa, thence east to the Ilova, then in a northward extension west of the Ilova to the Drava, protecting Zagreb. In 1683 when the removal of the Ottomans from Slavonia began, the Croatian feudatories assumed direct control of Regiments V and VI (the "Banska" regiments, i.e. under control of the Croatian Ban, or king), although general military policy was still set by the Austrians. The Border was subsequently extended to the Danube by 1718 and afterwards into the Carpathians. The Transylvanian regiments were decommissioned after 1848 and the Croatian Border was reunited with Civil Croatia only in 1881. Vestiges of its social organization, e.g. communal ownership of forest, remained in force until 1945.

Codification of the direct subordination of the Border population to the Emperor appeared in the Statuta Valachorum ("the law of the Vlachs", herinafter SV) in 1630. Rothenberg (1966 :11) claims that the SV recognized the traditional joint family household (zadruga, Hauskommunion) as the "recipient of the land grant". This is far from certain; Rothenberg's source (Sucevic 1953) notes only that the Vlachs lived in joint family households. Indeed, Article 8 of the SV explicitly gives shares in the estate to children of both sexes, an idea quite contrary to the exclusively agnatic inheritance of South Slavic customary law under which women do not inherit real property or stock. It also reveals an expectation of primogenitural inheritance typical of Germanic stem family organization.

"Starb ein Hausvater kinderlos, so hatte der nächste Verwandte mit der hinterlassenen Witwe der Wirtschaft vorzustehen. Hinterliess dagegen der Erblasser auch Kinder, dann leitete die Wittwe mit dem Formund oder Curator die Wirtschaft und auch das jüngste Kind hatte mit den übrigen Geschwistern ohne Unterschied des Geschlechtes das ungeschmälerte Recht auf den ihm zufallenden Erbteil" (Vanicek 1875:94).

Prof. Karl Kaser (personal communication and 1986, 1994a, 1994b) has kindly shared his insights on the legal status of the joint family and notes that the uskoci (guerilla fighters, Bordermen) of the Zumberacka Gora (Sichelberger Uskoken) received special feudal privileges in 1535 and lived in joint family households, but the privileges accorded them make no mention of zadruga ownership or inheritance. Only in 1737 with the establishment of the Hildburghausen reforms are zadruga ownership and the inheritance practices of customary law recognized. Habsburg military officers at the local level were certainly familiar with the joint household organization of their soldiers, especially since the regions west of Karlovac were than, as now, preponderantly Orthodox, among whom joint family organization is clearly traditional from the mediaeval Serbian evidence. That portion of the Border running north from the Sava to the Drava just east of Zagreb had a lower proportion of Orthodox. Whether the distinct civil administrators of the Hofkammer in Vienna or even of the Inner Austrian Estates understood this organization, being familiar only with administration of the Alpine Slavs, is moot. As we will see, a possible misunderstanding about the status of persons within joint households may have contributed to curious discrepancies in the listings.

Before the historic Ottoman defeat at Vienna in 1683, the population of Slavonia is thought to have been about 200,000. After seven years of destructive fighting following 1683, the subsequent Ottoman defeat within Slavonia in 1691, and the flight of the Muslim population and some of its allies, the number is said to have been only about 40,000 (Mazuran 1988:40). A formal peace was achieved in 1699 with the Treaty of Carlowitz (Karlovci), but relations between the interested superpowers (Austria, Venice, and the Porte) were not settled until the Treaty of Pozarevac in 1718. The exact ethnic/religious composition of the remnant population is unknown, although most Croatian historians seem to assume that it was mostly Catholic and Croat. The region was then flooded presumably by mostly Catholic migrants from other regions of Croatia and Hungary, and by Christian refugees from Bosnia, many of whom were Orthodox. Especially important was a strong flow of 30,000 Orthodox refugee families who crossed into Habsburg territory from Serbia in 1690 after the Austrian defeat and the failed Serbian uprising incited by the Habsburgs at Pec (in Kosovo-Metohija) in 1689. These Orthodox refugees were resettled mostly in southern Hungary and the eastern apex of Slavonia and in the more western regions near the Ilova River. By 1698 the population of Slavonia itself is thought to have reached between 65,000 and 80,000 (Gelo and Krivosic 1990:17, Mazuran 1988:42), implying an annual increase between 5 and 10 percent from 1691 to 1698. Gelo and Krivosic (1990:21) give the population in 1780 as 380,700, so that the growth rate from 1698 to 1780 would have slowed to between about 1.5 and 2 percent. Rothenberg's work on the Military Border (1960 , 1966 ) and Capo's (1990, 1991) and Gelo and Krivosic's (1990) analyses of family names suggest that growth before about 1700 was driven mostly by migration but after 1780 mostly by natural increase, migration having slowed except into a few towns. The growth rate after cessation of the migration flow was a few percent per annum. We conjecture from these data that we might expect about 2 percent per annum in crude natural increase around the time of the census in 1698.

One consequence of the Ottoman defeat was an enlargement of the Border (Fig. 2). The Croatian Sabor (Assembly) raised its own troops after 1683 and by 1690 had liberated the region south of the Kupa and from about the Glina to the Sava-Una confluence; thereafter the Croatian feudatories were in control of this region, although it was considered part of the overall Border organization. The Habsburgs extended the border in a narrow strip on the left bank of the Sava from the Una eastward, eventually (by the Treaty of Pozarevac in 1718) to the Danube (and ultimately into Transylvania). Part of the task of the census of 1698 was to decide where the northern boundary of the military zone should be. The military authorities wanted that zone to be as wide as possible, while the civilian authorities (who were in actual charge of the census) wanted that zone to be narrow, leaving as much reconquered territory as possible as spoils to be divided among favorites of the Court or sold to the highest bidder among the magnates.

The shifting tides of humanity 1683-1698 must stand as one of the most substantial population turnovers in early modern Europe (Stipetic 1988). Strong traces of the resulting ethnic differences remain even today in the dialect distributions and political conformations of the area. 4. The Census of 1698 In 1698 a commission of the Austrian Crown conducted the first thorough and non-ecclesiastical census of the 12,500 square kilometers between the Ilova, Drava and Sava rivers that had just been liberated from the Ottoman Empire. The census was undertaken to solve pressing questions of governmental organization and lay the basis for taxation and conscription, especially the extension of the Military Border along the Sava line. Planning for it began in October, 1691 after a failed Turkish counter-offensive, when the defeat of the Ottomans seemed certain, and the spoils of war within the Habsburg grasp. (Nevertheless, skirmishes with the Ottomans continued and indeed prevented the taking of the census or precensal reconnaissance in some areas, e.g. Srem.) It was executed only after protracted conflict between the civil and military arms of Habsburg administration and between them and the native Croatian nobility, all three of whom were competing for control of the new territories. The commission held an orientation for its district enumerators on March 3, 1698 and then traveled widely, taking testimony, redressing wrongs, and adjudicating disputes. The enumerators finished their work with few incidents by the middle of August. The actual implementation of the census was done by four work groups in four geographic areas. The first group worked in the arc from the Drava to the Sava, north of the central Slavonian massif and and down the Ilova, including the extreme western districts, the second from the Drava along the Sava drainage up to the limits of the western districts, the third in the Pozega basin and its surrounding mountains, and the fourth in the city of Osijek in the Drava basin. See Fig. 1. The census was received with apprehension by the populace, and 500 families are reported to have fled Slavonia to avoid it. In one village of District 20 (Mala Vlaska), the enumerator, one Gabriel Hapcz, was made to cool his heels for a week until the inhabitants received firm answers from higher authorities on certain questions of feudal status and tax and corvée liability . In each place, a pair of local, responsible persons was sworn to their task by a formidable oath in Latin and in Croatian and collected the census information. The information was almost surely collected in the local Slavic vernacular and then transcribed into Latin, whether first through a translation into German or directly, we do not know. We also do not know whether the sworn census takers actually visited the households in their catchment area, or interviewed the household heads, or whether they practiced what has come to be known as "curbside estimation." According to Mazuran (1993:24) all of the census takers and their assistants went through the same orientation in Osijek on March 3, 1698, but these did not include the "pair of village notables or elder persons" who reported to the census takers. It is possible that the official enumerators interviewed the pair of village elders, who reported from memory and their own knowledge without actually visiting the households. The average number of households per village was about 13, with district means ranging from 5 to 35.

The census listed (with supposedly standardized omissions) the members of households by category of kinship with the household head, the number of inquilini ("lodgers")[2] in each household and that of the sons and daughters of such lodgers, by district by village, together with their productive assets, down to the individual beehive.[3] Some villages or blocks of households within them are shown as having migrated from Bosnia, and such origin is also often evident in familial names, e.g. Bosnyak. As noted earlier, historians generally assume that children under age 15 were not listed, and married women were not listed unless they were widowed heads of households. There are 82 widows, or two percent of the households. Fully a third (28) of the widows are in the city of Pozega.

In a long account for each settlement, apparently based on a standard protocol since the paragraphs are consistently numbered, the census reported the history of settlement during the Turkish period and the recent war, the military role of the population, their religious affiliation, their needs, their feudal obligations, and limits on productivity such as the quality of land, insufficiency of cattle for plowing, and so on.

An example of a household listing is the following from the village of Jazavica in the district of Kraljeva Velika near the Ilova-Sava confluence in the west:

Marko Philipovich, fratres 2, filii 1, filiae 2, equi 3, boves 4, vaccae 3, vituli 3, oves et caprae 15, porci 10, alvearea 7, vineae fossorum 2, tritici jugera 8, hordei jugera 1/2, avenae jugera 1/2, milii jugera 4, currus foeni 6, terrae incultae jugera 5. (Mazuran 1988:541).

Marko Filipovic, brothers 2, sons 1, daughters 2, horses 3, oxen 4, cows 3, calves 3, sheep and goats 15, swine 10, beehives 7, vine rows[4] 2, yokes[5] of wheat 8, yokes of barley 1/2, yokes of oats 1/2, yokes of millet 4, haycarts[6] 6, yokes of fallow land 6.

By contrast is another kind of household from the village of Ternovac in the district of Erdut along the shores of the Drava in the east:

Vukovan Ternovazan, sessio 1, filii 1, equi 3, boves 3, vaccae 3, vituli 3, porci 15, frumentum jugera 14, hordei jugera 2, avenae jugera 2, kukuruz jugera 1, inquilini 2, filii inquilini 4.

Vukovan Ternovacan, sessions[7] of land 1, sons 1, horses 3, oxen 3, cows 3, calves 3, pigs 15, grain 14, barley 2, oats 2, maize 1, lodgers 1, lodger's sons 4

Mazuran's tabulation (1993:29-30) shows 464 occupied and 165 unoccupied villages (pagus desertus) in 28 named regions, and 6,613 households containing 385 brothers, 3,949 sons, 2,725 daughters, and 1,368 inquilini.8 Our own analysis is based not on his summary tabulations (which are themselves probably based on the village level summaries in the census) but on the detailed household level data for 4,453 households in those 330 villages in 17 districts for which the recording gave information on the kinship relation of residents to the household head.[9], [10] The missing districts are concentrated at the north end of the Drava valley. Because of this concentration it is possible that there is some selection bias in the corpus. The Drava valley was the main route for armies moving between Belgrade and Vienna and probably suffered equally from the Ottomans and the Habsburgs. However, there is no reason to think that the northern end of the Drava valley was affected more severely than the southern, which is represented in the census. 4. Expectations of Household Structure Ethnographies of the region have always stressed the extended and fraternal joint households found there, called zadruga (lit. "cooperative") by intellectuals but not called anything in particular by peasants. Households are simply called kuca (Serbia) or hiza (Croatia), meaning "house." Sometimes the words porodica or vamilija (Serbia) or obitelj (Croatia), meaning "family", are used. Peasants sometimes identify nuclear households by a special phrase, e.g. inokosna kuca, "lonesome house." The literature on "the zadruga" is enormous. (For recent discussion see Halpern and Wagner 1984, Hammel 1968 , Hammel 1972 , Hammel 1980a, Hammel 1990a, Hammel and Soc 1973, Todorova 1993). The formation of such households is rooted in ancient Indo-European principles of agnatic filiation and reflected in the kinship terminologies of Slavic and other languages, including Early Latin (Hammel 1957, Hammel 1968 , Lounsbury 1964).[11] Their existence in the area is suggested by mediaeval Serbian sources combined with 17th century evidence on Orthodox populations, Croatian historical sources of the 18th-19th centuries, and Austrian statutes of the 18th century. They exhibit, on the ethnographic evidence, what in anthropological jargon is called patrivirilocal residence, that is, a bride lives with her husband in the household of his father. If such principles of coresidence are followed without frequent fission, then under conditions of positive natural increase households can become quite large and complex, containing married cousins related in the male line, and may reach a membership in excess of a hundred. Households of these sizes, although rare, have been reported from the 19th and early 20th centuries. The size of larger households may be a function of increased expectation of life which continues the authority of the paterfamilias for longer (Halpern and Anderson 1970), and of emerging land shortage that forced denser coresidence and more intensive agriculture (Hammel 1990b, Hammel 1995). However, households have principles of fission as well as of fusion, and in fact even with strong rules of initial coresidence, divisive factors generally resulted in about 50 percent of households in a census being nuclear. So powerful was the tradition of complex households that Croatian and Serbian ethnographers almost uniformly regarded their division as a social pathology, not recognizing the cyclical nature of household formation. Peasants understand division very well and have institutionalized procedures for the division of property. They usually attribute division to arguments between female affines, that is between the wives of coresident sons or brothers, showing thus that agnatic filiation is a cultural principle as well as a social one. Indeed, the mother's brother (ujak), as a cognatically close but agnatically unrelated elder, often presided over such divisions as a distinterested outsider. (See Hammel 1968 , Hammel 1972 , Hammel 1976 , Hammel 1977, Hammel 1980a, Hammel 1980b, Todorova 1993). Fission of households in this region may be expected to take place typically as sons of coresident brothers approach puberty or adulthood, and domestic conflicts emerge between those who are not sons of the same father and thus not bound by immediate agnatic solidarity, which in patrilineal societies is typically strongest among brothers and progressively weaker for cousins of increasing collaterality. In the presence of the preferences that ethnography and history lead us to expect, and despite the potency of processes of fission, we would imagine that the census would expose particular kinds of evidence of such complex structures, especially in the fertile plains and hills of Slavonia, where large and complex households were common in later times, particularly in the Military Border where military labor demands encouraged large households. The censuses of 1702 and 1736 (Gelo and Krivosic 1990, Mazuran 1993), overlapping in area with that of 1698, and the work of Capo on villages of central Slavonia in the 18th and 19th centuries show abundant evidence of such complexity. Population, density, and complexity appear to increase as land resources remained relatively constant. Since households without sons but with at least one daughter would on the marriage of the latter "import" her husband; we would expect at least some trace of households with in-marrying sons-in-law, whose status has distinct names in various dialects of Serbo-Croatian.[12] Even if patrilocal extended households dissolved rapidly, and even if fraternal joint households lacking the binding authority of the father dissolved even more rapidly, we would expect some incidence of fraternal joint households of orphans, since unmarried sons would become unmarried brothers on the death of their parents and would remain together for a time even if they did not coreside with a married brother. Such households would often contain sisters. We would expect at least some nephews in households cohabited by brothers unless the cycle of fission was unusually rapid, and even some cousins in a large sample of households. 5. Census versus Expectations These expectations come up short against the census document itself, even in intuitive examination. Table 1 shows the mean number of kin of expectably included types as reported in the census. There are no coresident brothers reported in vast stretches of Slavonia, specifically in all but the three western districts of Stupcenica, Subocka, and Kraljeva Velika (nos. 21, 22, and 23). There is only one brother, in one household in one village in District 20 (Mala Vlaska). If the absence of reported brothers in the eastern zone is taken to mean an absence of actual brothers, this is an unbelievable outcome for a population with the kind of kinship and household system so consistently reported in the ethnographies and mediaeval history and recognized by 1737 in Austrian statute. There are no sisters listed in the census, even though we would expect some orphaned unmarried sisters to remain until marriage in households headed by a brother. There are no nephews listed, even in households in which there are brothers. If it were true that the 1698 census would have listed nephews if they were present, these data can only mean that married brothers left the joint household before their sons reached countability. There are no nieces. There are no grandchildren of countable age. There are no cousins listed, even though as noted in the traditional joint families agnatic cousins do sometimes coreside. There are no in-marrying sons-in-law listed. More broadly, there are no wives listed at all. Words for all these relationships exist both in the local Slavic vernaculars and often as regular words in Latin.[13]

There is not a single instance of the naming of a person of a generation older than the head of household. Thus there are no fathers, no mothers, no uncles, even though surviving mothers and debilitated elder males were surely coresident with sons or nephews who had assumed headship. That means either that the census takers must always have taken a male of the senior generation, or failing that a widow, as the head of household, or that they did not count males beyond some age or in some condition of disability unless they were solitaries or in isolated spousal pairs.

How can we account for such discrepancies? We wondered whether this important census should be consigned to the dustbin, or whether there was some way to rescue it. 6. Some Unsatisfactory Explanations for Curiosities in the Census In the customary ethnic and political logic of the Balkans, explanations for some of the anomalies and regional differences fairly leap to mind. The Orthodox are reputed to have larger families, a more archaic family organization, and to have been involved more in military feudalism. Under Austrian practice, complex households were encouraged in order to to insure the economic viability of households if men were sent on military duty, and the per capita burden of military service was lessened if households contained more adult males. The concentration of "brothers" in the western zone would be easy to attribute to the prevalence of Orthodox Serbs and the general importance of military rule there. But a detailed comparison at the village level shows that almost all districts were heterogeneous in religion, ethnicity, military versus civil status, and household organization. Indeed, even villages were sometimes heterogeneous in these respects. Apart from small differences in household size and relative numbers of kinds of kin, contemporary military involvement or religion/ethnicity seemed to make little difference in household organization. Some districts in which brothers are listed were finally included in the formal Military Border, but some were not. Most importantly, there were very substantial numbers of Orthodox Serbs in regions and villages in which no brothers were listed at all. We were unable to construct reasonable explanations for household structural differences based on ethnicity or political status.[14] 7. Census Redux Though not impossible, it would certainly be surprising if persons such as those omitted were not in Slavonian households as they existed in actual operation. It would be less surprising if the kinship terms used in the census meant more than they seem. A fundamental point of this analysis rests on the anthropological commonplace that it is a gross error to assume the same genealogical or social meanings for kinship terms across languages, especially in using the most restricted meaning in one language to gloss the apparently matching term in another.

Were the kinship terms employed to designate persons defined by the immediate genealogical denotation of the Latin words? Were the Latin words glosses on Slavic kinship terms but still interpretable in a narrow genealogical context? Were the Latin words, on the other hand, convenient glosses for socially defined categories?[15] In other words, does filius mean precisely a child of the head, or does it mean a lineally related male member of the household one or more generations below that of the head of household, or even just a junior male? Does fratres mean exactly a male child with the same parents as the head, or any collateral male relative of the same generation as the head? Does filia mean a female relative analogous to filius in any of the senses given above for filius, or can it also mean any unmarried female in the house who is a jural minor, thus including persons that would have been listed as soror (sister) if that word had been used anywhere in the census?

Everything that anthropologists know about kinship usage would suggest that the kinship terms used in the census would have broad, rather than narrow meanings. Exactly how broad the meanings were, we cannot say, but some aspects of the local culture and language, and some features of the internal structure of the census are informative.

Where "son" is given in the listings, it appears not only after the head of the household but also after any brothers that may be listed. Since "nephews" are not listed by the literal Latin term (nepos), we conclude that "son" means the son of either the head of household or of any listed brothers, so that "son" in the census means sons plus nephews. We also note that in South Slavic joint family organization the sons of coresident brothers are socially equivalent, and that the term for "brother's son" in some dialects (sinovac) is a linguistic derivative of the term for son (sin). If people were reported in Slavic but noted in Latin, these effects might easily appear. For all of these reasons we consider filius in the census to mean at least sons plus nephews of the head. The clinching argument is the positional one from the census listings: head, brothers, "sons." One could go further by including grandsons under the social rubric of sons, but we do not explicitly make that extension. Parallel arguments (except for the linguistic derivative of brother's son from son) apply to filia (daughter), and we consider filia in the census to mean at least daughters plus nieces of the head. Further consideration of the meaning of filia will be pursued below.

The term fratres (brother) occurs in the census only in the three westernmost districts. (See the example of the Philipovich household given earlier.) It is not found in the other fourteen. In these fourteen there are many "lodgers" listed (inquilini), but no "lodgers" are listed in the three westernmost districts (see the example of the Ternovazan household listed earlier). There is only one exception to this distribution: in one village in one eastern district verging on the western three, where "lodgers" are listed; there is one listed brother in one household, but there are no lodgers in that household. The geographical distribution of lodgers and brothers is thus completely disjoint. We propose that the Austrian census takers, familiar with a stem family system of the Alpine regions in which non-inheriting sons could stay on with their inheriting brother as Knechte (farmhands), usually classified coresident brothers as lodgers. They could have done so even if they were familiar with the zadruga form of organization, provided that they did not fully appreciate that all males were equal coparceners in the zadruga estate, regardless of which one was the head. The exceptions to the general absence of brothers all occur in the area enumerated by census work group I, headed by the unlucky Gabriel Hapcz (see above). Indeed, the change from lodgers to brothers occurs exactly in the village of Podborje, where the inhabitants obliged Mr. Hapcz to wait. We suppose that for almost all of their tour this work group used the same definitions as other work groups, changing the terminology only at the end, perhaps because they decided to use the kinship term rather than the misapplied status term to describe the relationship. Whether the near-revolt of the populace altered their linguistic or cultural sensitivity can only be a matter of amused speculation. For our purposes, we therefore classify "lodgers" as "brothers", "lodger's sons" as "sons", and "lodger's daughters" as "daughters." If there were truly unrelated lodgers in households in the eastern zone, our equivalencing will overestimate brothers in that zone, but we think probably not by much. In later times Austrian military administrators managing land allotments and supervising the organization of zadrugas sometimes did oblige unrelated persons to coreside in order to maintain sufficient military strength on the homestead, but such arrangements were rare.

Beyond this we note that in the local Slavic dialects of the time (and indeed in many of them today such as in standard Serbian), siblings and cousins are denoted by the same word. Both a brother and a male cousin are brat (cognate with English "brother), and both a sister and a female cousin are sestra (cognate with English "sister"). (See notes 11-13.) Socially, male cousins of the head in a joint household are equivalent to his brothers. If counts of people were reported in Slavic and glossed in Latin, these effects could easily occur. For all of these reasons, we consider the Latin gloss, fratres, to be equivalent to the Slavic brat, and to mean not only brothers (and lodgers, as indicated earlier) but also male cousins.

Finally, and with some trepidation, we reconsider the Latin filia, which we have held to mean both "daughter" and "niece", analogously to filius. It will be remembered that there are no sisters as such listed in the census, although some must have existed at countable ages. We propose that it is reasonable to consider the Latin filia as a gloss for all countable unmarried women. There is no direct linguistic or structural evidence internal to the census for this terminological merger, contrary to the situation for the other interpretations above. Nevertheless, the merger is ethnographically reasonable, since all unmarried daughters, nieces, and sisters were jural minors in their household and expected to depart it on marriage. All nubile women, in this sense, are "daughters of the lineage" (Hammel 1957, Hammel 1968 , Lounsbury 1964). There is some indirect linguistic evidence even though the Slavic terms for sister (sestra) and daughter (kci, kcer, cerka) are distinct and never confused. The husbands of daughters and of sisters are known by a single Slavic kinship term (zet); that is, son-in-law and brother-in-law are identically named; sisters and daughters are the wives or potential wives of identically named men. We would not suggest this terminological merger if any "sister" (soror) had been listed as such in the census at all, but none were, and the interpretation seems not unreasonable. To accommodate this possibility, we later present both interpretations of the census counts, in one of which filia means daughters and nieces, and in the other of which it means these plus sisters. In either of these instances, we could include granddaughters under the social rubric of daughters, but we do not do so explicitly here.

The results of these re-readings of the text are shown in Table 2. We propose that these linguistically and ethnographically informed "readings" of the census text do much to resolve the Brother's Puzzle, the Son's Puzzle, and the Daughter's Puzzle. The outcome is an "adjusted census", in which the counts of persons are affected only insofar as "lodgers" are merged with "brothers", and the children of lodgers are merged with "sons" and "daughters". It is the interpretation of these counts that is most changed, a shift that has meaning when we go on in Part II to compare the simulation results with the census and use these interpretations to aggregate the exactly classifiable computer-simulated coresidents into the categories we think were intended by the census takers. 10. Conclusion In this paper we examined an important foundation document for the history of Croatia and of the Balkans in general, asking whether the counts of kin within households were plausible in the light of general knowledge of the ethnography and history of the region. The first answer to that question is clearly negative. The census listings, if interpreted literally, make no sense. Expectable and important categories of kin are entirely missing even under the apparent census protocol, while the counts of those included are inconsistent with plausible qualitative expectations.

As a first step in the rehabilitation of the census for scholarly study we reconsidered the meanings of the kinship terms as reported in Latin. Some were given broader meanings on the grounds of structural features of the listings (filius as son plus nephew, filia as daughter plus niece). Some were given broader scope on the grounds of the meanings of the Slavic glosses in Serbo-Croatian (fratres = brat as brother plus cousin). An extension of filia = daughter plus niece plus sister was proposed on the basis of the legal status of unmarried females in patrilineal systems and tentatively implemented. A plausible extension of filius = son plus nephew plus grandson (and similarly for filia) was considered but not implemented.

These qualitative steps transform the census listings into a more reasonable and believable corpus, better in accord with the expectations generated by ethnography and history. In Part II of our enterprise we will turn to microsimulation techniques to obtain more precisely drawn quantitative expectations of coresident kin, using a range of plausible demographic scenarios and the most likely household formation scenario and census protocol. From those results we will go on to estimate the population that was counted only by enumeration of household heads rather than by coresidents in households and to estimate the population that was indicated only by counts of households or villages. The results accord closely with population counts estimated by other means by Croatian historians and lead us to consider other processes such as migration, the instability of large households, and the need to combine very small households in the turbulent environment at the close of the 16th century when a temporary peace with the Ottomans was within grasp.

Table 1

Mean Numbers of Expectable Kin in Households in the Empirical Census

              Sons  Daughter  Brother  Siste  Grandchildre  Nephews  Niece  Cousins  
                    s         s        rs     n                      s               
East          0.85  0.633     0        0      0             0        0      0        
West          0.90  0.486     0.524    0      0             0        0      0        
Total         0.87  0.604     0.412    0      0             0        0      0        

              Lodgers  Lodger's      Lodger's           Head's Wives  Son's Wives   
                       Sons          Daughters                                      
East          0.368    0.011         0.006              0             0             
West          0.040    0.001         0.000              0             0             
Total         0.295    0.009         0.005              0             0             

               Daughter's     Brother's     Nephew's      Niece's      Brother's    
                Husbands        Wives        Wives        Husbands       Wives      
East          0              0            0             0             0             
West          0              0            0             0             0             
Total         0              0            0             0             0             

Census data are for the 3 western and 14 eastern districts. (See Fig. 1.)

Table 2

Mean Numbers of Expectable Kin in Households in the Adjusted Census

        Sons plus Lodger's Sons    Daughters plus Lodger's        Lodgers plus      
          (Sons and Nephews)       Daughters (Daughters and   Brothers  (Brothers   
                                           Nieces)            and Male Cousinse )   
East             0.870                      0.639                    0.368          
West             0.907                      0.486                    0.564          
Total            0.888                      0.609                    0.707          

The italicized kinship terms suggest what the census categories really might have included.Acknowlegements

This research was supported by grant DBS-9120159 from the National Science Foundation and grant RO1 HD 29512 from the National Institute of Child Health and Human Development of NIH, and by the facilities of the Department of Demography, the Institute of International Studies at the University of California, Berkeley, the Institute for the Study of Ethnology and Folkloristics and the Archive of Croatia, both of Zagreb. None of these institutions is responsible for the data or interpretations found in this paper.

References Cited

Bogisic, V., 1874, Gragja u odgovorima iz razlicnih krajeva slovenskog juga. (Jugoslavenska Akademija Znanosti i Umjetnosti, Zagreb).

Brunt, P. A., 1971, Italian manpower 225 BC-AD14 (Clarendon Press, Oxford).

Capo, J., 1990, Economic and demographic history of peasant households on a Croatian estate, 1756-1848. Doctoral dissertation (Demography, University of California, Berkeley).

Capo, J., 1991, Vlastelinstvo Cernik: gospodarstvene i demografske promjene na hrvatskome selu u kasnome feudalizmu (Institut za Etnologiju i Folkloristiku, Zagreb).

Engel, F. S., 1971 (1786), Opis kraljevine Slavonije i vojvodstva Srema, Zbornik Matice Srpske za Knjizevnost i Jezik XIX, 289-356.

Gelo, J., and S. Krivosic, 1990, Razvitak stanovnistva na tlu hrvatske (Ekonomski Fakultet, Institut za Ekonomska Istrazivanja, Zagreb).

Halpern, J. M., 1958, A Serbian village (Columbia University Press, New York).

Halpern, J. M., and D. Anderson, 1970, The zadruga, a century of change, Anthropologica 12, 83-97.

Halpern, J. M., and R. A. Wagner, 1984, Time and social structure: a Yugoslav case study, Journal of Family History 9, 229-244.

Hammel, E. A., 1957, Serbo-Croatian kinship terminology, Kroeber Anthropological Society Papers 16, 45-75.

Hammel, E. A., 1965, An algorithm for Crow-Omaha solutions, American Anthropologist 67, 118-26.

Hammel, E. A., 1968 , Alternative social structures and ritual relations in the Balkans (Prentice-Hall, Englewood Cliffs).

Hammel, E. A., 1972 , The zadruga as process, in: P. Laslett (ed.) Household and family in past time (Cambridge University Press, Cambridge) 335-73..

Hammel, E. A., 1976 , Some mediaeval evidence on the Serbian zadruga: a preliminary analysis of the chrysobulls of Decani, in: R. F. Byrnes (ed.) Communal families in the Balkans ( University of Notre Dame Press, Notre Dame) 100-115..

Hammel, E. A., 1977, Reflections on the zadruga, Ethnologia Slavica VII, 141-151.

Hammel, E. A., 1980a, Household structure in 14th century Macedonia, Journal of Family History 5, 242-73.

Hammel, E. A., 1980b, Sensitivity analysis of household structure in medieval Serbian censuses, Historical Methods 13, 105-118.

Hammel, E. A., 1984, Opadanje fertiliteta u vojnoj krajini Hrvatske (Anthropological Society of Croatia and Ethnological Society of Croatia, Zagreb, 1984).

Hammel, E. A., 1985, Short term demographic fluctuations in the Croatian military border of Austria, 1830-47, European Journal of Population , 265-90.

Hammel, E. A., 1990a, Demographic constraints on the formation of traditional Balkan households, in: A. Laiou (ed.) Dumbarton Oaks Papers: Marriage and the Family in Byzantium, vol. 44, Washington, D.C.) 173-180..

Hammel, E. A., 1990b, The early decline of fertility in Slavonia, Studia Ethnologica 2, 143-162.

Hammel, E. A., 1993, Censored intervals in family reconstitution: a sensitivity test of alternative strategies with historical Croatian data, in: D. Reher and R. Schofield (ed.) Old and new methods in historical demography (Clarendon University Press, Oxford) 125-44..

Hammel, E. A., 1995, Economics 1: Culture 0. Fertility change and differences in the northwest Balkans 1700-1900, in: S. Greenhalgh (ed.) Rethinking reproduction: culture and political economy in demographic behavior (Cambridge University Press, Cambridge) 115-258..

Hammel, E. A., and D. Soc, 1973, The lineage cycle in southern and eastern Yugoslavia, American Anthropologist 75, 802-814.

Hammel, E. A., and H.-P. Kohler, 1995, Kinship-based resource sharing in the agrarian economy of frontier Slavonia, 1698: evidence from an early census (Social Science History Association, Chicago, 1995).

Hume, D., 1752, On the Populousness of ancient nations: Essays, moral, political, and literary , ).

Ilic Oriovcanin, L., 1846, Narodni slavonski obicaji , Zagreb).

Kaser, K., 1986, Freier Bauer und Soldat. Die Militarisierung der agrarischen Gesellschaft in der kroatisch-slawonischen Militargrenze (1535-1881) , Graz).

Kaser, K., 1994a, The Balkan joint family household: seeking its origins, Continuity and Change 9, 45-68.

Kaser, K., 1994b, The Balkan joint family: redefining a problem, Social Science History 18, 243-269.

Laiou-Thomadakis, A. E., 1977, Peasant society in the late Byzantine Empire: a social and demographic study (Princeton University Press, Princeton).

Lounsbury, F., 1964, The formal analysis of Crow- and Omaha-type kinship terminologies, in: W. H. Goodenough (ed.) Explorations in cultural anthropology (McGraw-Hill, New York) 351-394..

Lovretic, J., and B. Juric, 1897, Otok: narodni zivot i obicaji, in: (ed.) Zbornik za narodni zivot i obicaje u juznih slavena, vol. II (Jugoslavenska Akademija Znanosti i Umjetnosti, Zagreb) 91-459..

Mazuran, I., 1988, Popis naselja i stanovnistva u Slavoniji 1698. godine, Vol. 2, Radovi Zavoda za Znanstveni Rad u Osijeku (Jugoslavenska Akademija Znanosti i Umjetnosti, Osijek).

Mazuran, I., 1993, Stanovnistvo i vlastelinstva u Slavoniji 1736. godine i njihova ekonomska podloga, Radovi Zavoda za Znanstveni Rad u Osijeku (Hrvatska Akademija Znanosti i Umjetnosti, Osijek).

Moacanin, F., and M. Valentic, 1981, Vojna krajina u Hrvatskoj , Zagreb).

Morgan, L. H., 1870, Systems of consanguinity and affinity of the human family , Washington).

Pavicic, S., 1953, Podrijetlo hrvatskih i srpskih naselja i govora u Slavoniji, Vol. 47, Djela (Jugoslavenska akademija znanosti i umjetnosti, Zagreb).

Pavlicevic, D., Editor, 1984, Vojna krajina (Sveucilisna Naklada Liber, Zagreb).

Reljkovic, M. A., 1973 (c. 1760), Satir, ili divji covik, Zbornik stihova i proze XVIII stoljeca (Matica hrvatska, Zagreb).

Rothenberg, G. E., 1960 , The Austrian military border in Croatia, 1522-1747, Vol. 48, Illinois Studies in the Social Sciences, (University of Illinois Press, Urbana).

Rothenberg, G. E., 1966 , The Austrian military border in Croatia, 1740-1881 (University of Chicago Press, Chicago).

Shah, A. M., 1974, The household dimension of the family in India; a field study in a Gujarat village and a review of other studies (University of California Press, Berkeley).

Smiciklas, T., 1891a, Dvijestagodisnjica oslobodjenja Slavonije. Drugi dio: Spomenici o Slavoniji u XVII. vijeku (1640-1702), Vol. XI, 2, Djela (Jugoslavenske Akademije Znanosti i Umjetnosti, Zagreb).

Smiciklas, T., 1891b, Dvijestagodisnjica oslobodjenja Slavonije. Prvi dio: Slavonija i druge hrvatske zemlje pod turskom i rat oslobodjenja, Vol. XI, 1, Djela (Jugoslavenske Akademije Znanosti i Umjetnosti, Zagreb).

Stipetic, V., 1988, Predgovor, in: I. Mazuran (ed.) Popis naselja i stanovnistva u Slavoniji 1698. godine, vol. 2, Radovi Zavoda za Znanstveni Rad u Osijeku (Jugoslavenska Akademija Znanosti i Umjetnosti, ) ..

Sucevic, B. P., 1953, Razvitak `Vlaskih Prava' u Varazdinskom generalatu, Historijski Zbornik 6, 33-70.

Todorova, M. N., 1993, Balkan family structure and the European pattern: demographic developments in Ottoman Bulgaria (American University Press, Washington, D.C.).

Tomasevich, J., 1955 , Peasants, politics and economic change in Yugoslavia (Stanford University Press, Stanford).

Valentic, M., 1981 , Vojna krajina i pitanje njezina sjedinjenja s Hrvatskom 1849-1881 , Zagreb).

Vanicek, F., 1875, Specialgeschichte der Militärgrenze, Vol. I (Kaiserlichen-Königlichen Hof u. Staatsdrückerei, Vienna).

Winner, I. P., 1971, A Slovenian village: Zerovnica (Brown University Press, Providence).