[1] For information in English on the kinship, familial, and tribal systems of the southern Slavs see inter alios Halpern (1958), Halpern and Anderson (1970), Hammel (1957, 1968 , 1972 , 1976 , 1977, 1980a, 1980b, Hammel and Soc 1973). For comparative evidence see Shah (1974). See also the critique by Todorova (1993). The material on Croatian and specifically Slavonian ethnography in Croatian begins with Reljkovic (1973 (c. 1760)), Engel (Engel 1971 (1786), first published 1786), Ilic Oriovcanin (1846), and Lovretic and Juric (1897), this last being the second volume in the massive series on the national life and customs of the southern Slavs (i.e. the Croatians) published by the Yugoslav Academy of Sciences and Arts (i.e. the Croatian Academy) and the parallel series on Serbian ethnography (Srpski Etnografski Zbornik) published by the Serbian Academy of Sciences and Arts since 1894. Information on native legal systems is given in Bogisic (1874). Overviews of the historical record are given in Moacanin and Valentic (1981), Pavicic (1953), Pavlicevic (1984), Rothenberg (1960 , 1966 ), Smiciklas (1891b), Tomasevich (1955 ), Valentic (1981 ).

[2] In this census inquilini appear as members of households headed by non-inquilini, thus as lodgers. In later records inquilini appear as heads of independent households having little or no land in feudal tenure. In 1698 inquilini appear in villages with military functions and in those without such functions. In later data inquilini are found only in Civil Croatia, not in the Military Border.

[3] Gelo and Krivosic (1990:16) note that some categories of persons without potential or existing feudal obligation were not listed; these were for example socalled subinquilini (beggars, servants, certain widows without working coresident males, etc.). In the city of Pozega in 1702 these omissions resulted in a 22% undercount, but such persons might have been more common in towns than in villages, which constitute the bulk of the settlements.

[4] The meaning of the Latin is not clear. Fossa is a ditch or furrow but may refer here to a row of grape vines planted along a furrow. Fossor is literally "digger" but might mean a row, furrow, or pit.

[5] The jugerum (from iugum, "yoke") was sometimes defined as the area that could be sown with a set amount of grain, sometimes as a day's plowing, sometimes by size in fathoms or ulnas, varied in definition between arable and meadow, but generally seems to have fallen between about 2400 and 3600 sq. meters. See Capo (1990:28ff.). In later times in Austria it was about 1.5 acres.

[6] Our translation could be misleading. What might be meant is the amount of land that would yield a cartload of hay. Similarly in some households falcator is listed. Falcator can be a reaper, or a scythe, or it could be the amount of meadow that can be scythed by one man in a day.

[7] A session of land constituted a full allotment for a colonus (serf). In later times coloni often had less, and inquilini (by then no longer "lodgers" but analogous to "cottagers") had less than a quarter of a session. The area of a session depended on the region, soil quality, and other factors.

8 Elsewhere (Mazuran 1988:41) he gives the total of occupied settlements as 491 and that of unoccupied settlements as 240.

[9] Data from the detailed listings were entered in computer disk files. The data for each village include not only the listing by individual household but also a village summary of totals in each category, e.g. total sons, brothers, pigs, etc. We summed the household level data independently and checked these sums against the Austrian sums in the census. Where there were no differences we accepted the data; where there were discrepancies between the sums we re-examined our data entry. We took the published household level data literally, attributing any rare remaining differences from the published village sums to Austrian arithmetic error.

[10] The districts with no kinship information are nos. 3 - 8 in the census as published by Mazuran (1988), namely Valpovo, Miholjac, Podravska Moslavina, Slatina, Virovitica, and Nasice. Mazuran (1993:Table 1) lists four more in summary form, namely Osijek, Ivankovo, Djakovo, and Vocin. All of these are in the Drava basin. Some villages, or sections of villages, do not have any kinship information, and sometimes such sections are noted as containing persons from Bosnia. Where census counts are available for villages or districts, but no kinship information within households is given, the census takes the form of only a list of names of inhabitants. These are surely the heads of the households, so that counts of households are possible. Widows are indicated, and sometimes an additional man is listed as a brother.

[11] A kind of kinship terminology typical of strongly patrilineal societies is called "Omaha" by anthropologists, after an American Indian tribe in which it was identified. In Omaha terminologies the males of the mother's patrilineage are terminologically merged to greater or lesser degree across generational levels (Hammel 1965, Lounsbury 1964, Morgan 1870). For example, mother's father, mother's brother, and mother's brother's son may all be called by the same kinship term. A convenient example from Latin is the quasi-merging in the terms avus (grandfather) and avunculus (mother's brother as differentiated from patruus, father's brother). Avunculus is the diminutive of avus, thus "little grandfather" and reflects early Latin patrilineal kinship organization. Less dramatic examples of the influence of social organization on kinship terminology in Serbo-Croatian are the distinction between father's and mother's brothers (stric and ujak) but the merging of father's sister and mother's sister (tetka). Stric can easily coreside in one's natal house, while ujak would not, and after their marriage no tetka coresides in one's natal house. Aunts by marriage are not called tetka but strina and ujna, respectively.

[12] For example, domazet (from dom, house, and zet, sister's or daughter's husband), uljez (with the meaning of intruder), mirascija ("one who enjoys dowry", a derivative of miraz, dowry, from Turkish miras from Arabic mirat).

[13] By "word" here we mean a monolexemic, simple expression, not a compound or circumlocution. For example, in South Slavic the kinship term for "husband's brother's wife", which in English would be "sister-in-law", is simply jetrva. Some caution is necessary. In many dialects of Serbo-Croatian the word brat means either brother or cousin although the meaning can be made specific through an adjective or prepositional phrase (rodjen brat, "born brother", brat od strica, "brother through my father's brother", etc.). In other dialects the word rodjak is used for cousin. Nephew is necak in some dialects, bratic or sinovac or bratanac or sestric in others. Latin has patruus for father's brother, fratruelis for cousin, patruelis for father's brother's son, and nepos for nephew or grandson (note the generational merging, although it is bilateral through either sons or daughters, brothers or sisters). None of these locutions are found.

[14] Karl Kaser (personal communication) notes that while ordinary ethnic differences may have no import for household structure, the ecology of the area of origin of migrants did. In a census of 1712 for the Lika region he could show that residents noted in the census as Orthodox or Catholic "Vlachs" (Vlasi or Bunjevci) or as converted Muslims (neochristiani) generally showed around 50 percent or more of households as complex. Converted Muslims may well have been previously Muslim Vlachs. "Vlachs" are predominantly pastoral. By contrast, residents who were not classied as Vlachs but as Croats (Croate) showed a much lower proportion of complex households. These households were from core Croatian areas, where the economy was more agricultural (1994a, 1994b). In similar work on a census from the Erdut region in 1736 he found that those from pastoral regions had joint family households while those from nonpastoral regions did not (personal communication). We are entirely sympathetic to this finding and would anticipate confirming it if the origins of households were more precisely given in the 1698 census. (For general confirmatory information on this ecological-economic emphasis see Hammel 1968 , Hammel 1995, Hammel and Kohler 1995). Kaser also notes that inmigrants apparently came in nuclear households but rapidly developed joint households if they had come from pastoral areas, while inmigrants from nonpastoral areas did not. He basis this on evidence of 342 households that had arrived in Lika only a few months before the census. The scenario suggested is one in which pastoral inmigrants had lived in joint households in their areas of origin, broke into nuclear households to migrate, and then rebuilt joint families after arrival (1994a: 262). It is not entirely clear what it meant to "rebuild" joint families. One scenario is rebuilding by natural growth. In that case we would expect to see households composed of fathers and married sons but no married brothers, if such households had been in the area about 20 years -- that is, long enough for an original inmigrant nuclear household to produce one or more married sons but not long enough for the father to die and leave his married sons behind as married brothers. If, on the other hand, households resident about 20 years showed fraternal joint organization, we could imagine that the joint families that had existed on the Bosnian side of the border and reported themselves as nuclear on the Croatian side were only engaging in the usual peasant duplicity in order to get more land on a per-household basis, and that they then proceeded to conduct their household economy in the ordinary joint way. Kaser proposes the first of these scenarios, because he finds few households sharing a patronym. In an economic analysis subsequent to this paper, Hammel and Kohler (1995) found that in the 1698 census the ratio of last names to households was 1063:1657 for demonstrably Catholic households and 835:1341 for demonstrably Orthodox households, thus .64 and .62 for Catholic and Orthodox, respectively. A household selected at random would share a last name about 2 times out of 1000 with another randomly selected household. Sharing of last names is more suggestive of actual kinship relationship at close distances. In the economic analysis, putative kinship was assumed on the basis of name identity but with an exponential decrease according to geographical distance. Kinship thus imputed was found to have important consequences for the extent of land cultivated in grain and the status of oxen as a factor of production in grain cultivation, leading to the conclusion that oxen were probably shared between putatively related households but not between unrelated households.

[15] It is important to remember how the census was taken. Village elders, no doubt monolingual and illiterate, reported to Habsburg scribes who knew Latin. The scribes may have known some dialect of Croatian or Serbian, since the Empire had administered the Military Border and had been the suzerain of civil Croatia since the early 16th century, or they may have used interpreters who knew at least Slavic and German if not Slavic and Latin. In any case, the Slavic kin terms must have been glossed in Latin.