1When discussing of social network in historical communities, historical demographers have often focused their attention and efforts on the co-residential patterns of living arrangement, that is to say the household structure. Classification of communities according to their household composition as well as theoretical models linking this feature to demographic regimes has represented a useful interpretative tool for explaining demographic evolution of past human populations (Le Play, 1871; Laslett, 1983; Hajnal, 1983; Skinner, 1997). The basic assumption was that the presence of kin, in this case only co-residential kin, could somehow influence demographic outcomes by providing social and economic support to individuals and biological couples.

2Needless to say, scholars were aware of the limits of this approach as it took into account only the co-residential component of kinship, thereby considering neither the broader network of kin nor the network of social relationships like godfathers and godmothers, marriage alliances and, more in general, the entire system of friendship. This situation is as true as to induce Perrenoud (1998, 2) to say, “the study of family structures and strategies has focused on the domestic group as if it could be abstracted from its larger familial environment”.

3Though some attempts to take into account also the larger kin network beyond the household have been recently proposed (Das Gupta, 1997), Levi’s critics (1990) are still well founded and justified. He strongly criticized the household approach, inviting all the scholars in family history to abandon the typological analysis on household structure and to put much more efforts in reconstructing the entire network of social relationships of individuals. In his opinion, this would have helped not only to better specify the “real conditions in which choices and strategies arise” (Levi, 1990, 577), but also to understand much more in depth the connections between social network and demographic systems, and to re-think the real role of family in this fundamental process.

4However, it is clear that the stress on household-based analyses in historical demography derived mainly from the impossibility to reconstruct the entire network of social relationship on account of the inadequacy of archival material and sources of the past (Plakans, 1984; Kertzer, 1992; Post et al., 1997).

5As well known at this point, in the Italian society of xix^th century, and still nowadays, family and strong family ties have always played a central role in people everyday life. Yet, some important differences existed. The dichotomy sharecroppers-day laborers, involving different forms of household typology, respectively complex and simple structures, is perhaps one of the most studied examples in which family ties had such different roles and importance as to be roughly described, according to Reher’s definition, as ‘weak and strong family systems’ (Reher, 1998).

6In this paper, we will examine, thanks to a careful exploitation of historical sources, the local kinship network of a Tuscan community during the xix^th century, in which sharecroppers and day laborers represented the two most important agricultural groups there working. Our purpose is to describe, firstly, the characteristics of such a network, paying particular attention to the relationship between different typologies of household structure (coresident kin) and the kinship network beyond the household unit. In the last part of the paper we will finally provide some evidences of the importance of the kinship network in the circulation and emigration of individuals.

The community studied: Casalguidi, 1819-59

7The parish of Casalguidi is situated in the countryside of what is today the province of Pistoia, and throughout the period of analysis, 1819-59, it was included in the Grand Duchy of Tuscany.

8In the period studied, the population rose from 1,906 inhabitants in 1819 to 2,697 in 1859, for the most part living on agricultural labors. Yet, the social structure was diversified and not homogeneous, and even among farmers there were quite large differences in socioeconomic condition. This was mainly due to the fact that both the holding of property and the characteristics of the agricultural contracts differed greatly among the various agricultural groups (Giorgetti, 1973; Luzzati, 1979). Sharecroppers, on the one hand, and day laborers, on the other hand, were the two most represented categories of farmers. The former lived usually in complex households, whose members were bound to the landowner as a single working unit in which the head (usually a man) had great power and control over the other members. The contract provided that the farm produce had to be split evenly between the sharecropping family and the landowner. Day laborers lived, on the contrary, in simple-family households (Barbagli, 1998), residing in small houses in the village center and not directly on the farm.

9On the whole, complex households amounted to 35.3%, involving 48.2% of inhabitants, while nuclear ones accounted for 58.1% of total households and 49.6% of the population. The overall mean household size was about 5.2 members in the period considered. [1] Obviously, marriage played an important role in this situation of large diffusion of complex households. It was the result of the typical patrilocal system of Italian sharecropping societies (Barbagli, 1988 and 1990; Barbagli and Kertzer, 1992; Viazzo, 1990; Rettaroli, 1993), characterized by a high age at first marriage for both the sexes (around 28 and 25 years, respectively for men and women) and high level of celibacy among females, around 11%.

10From a demographic point of view, Casalguidi reflects many of the typical features of nineteenth century Italy. The life expectancy (e₀) was little over 35 years, while fertility (TFR) was about 5.5 children per woman on average. These demographic features contributed to determine a continuous upward trend of the total population (8.3‰ annual growth rate), interrupted only in 1855 because of a serious outbreak of cholera.

Sources used and methodology adopted

11Basically, two kinds of historical sources have been used in this analysis for the period 1819-59: Parish Registers and Family Tax Registers. Parish registers includes the classic ecclesiastical vital registers—baptism, burial and marriage acts—along with the Status Animarum (Registers of Souls). The first books are well known and they have been the basis of almost every work in Italian historical demography, while the second ones need further comments. [2] The Status Animarum was actually a sort of annual census the parish priest recorded on the occasion of pastoral visits to families during Easter. For each household residing in the parish territory, the parson recorded name, surname, age, marital status and relationship to the head of each of the household’s members. Information about the property of the house was present as well. Since the complete series of annual Status Animarum for the period studied is available, the linkage of information between the two sources (census and vital registers) allowed us to reconstruct the life-histories of individuals and families for the part of their life they spent in Casalguidi (Manfredini, 1996). By assigning a unique code number to each person and his/her parents and spouse(s) we were able to identify the individual kinship network by tracing relationships both horizontally and vertically in the genealogy line. Thanks to the information on co-residential patterns provided by Status Animarum, it is now possible to determine the kinship network of each person year after year, specifying the relatives living at Casalguidi both outside and within the household of the index person. According to this methodology, the reconstruction of non-coresidential kinship is inevitably limited to the local network of kin.

12At the end, while the Stati Animarum provide us, year by year, information on the structure of co-residential patterns, vital registers keep us informed about the events which interested the members of such households as well as some of their biological and demographic characteristics (sex, age, etc.). Cross-checking information permitted us to improve not only the quality and quantity of data, but also, thanks to the continuity of Status Animarum over time, to build up a new meta-source with longitudinal data on individuals and households.

13The second documentation we exploited is the Family Tax Register. It contains information about the tax amount each family had to pay annually according to its economic status. Through name and surname of the household head recorded in these registers, we linked such information to the life-histories of individuals, determining with precision the socio-economic status each person (and household) has experienced over his life-course. [3]

14In the final part of this paper, reconstructed longitudinal data have been analyzed using event history analysis—one of the best and most powerful statistical tools in dealing with this kind of data (Allison 1984; Trussell and Guinnane, 1983; Gutmann and Alter, 1983). A discrete-time approach is used, since the data set is characterized by annual repeated observations.

Some notes on the kinship network at Casalguidi

15Since in Casalguidi coexisted two different patterns of household structure—the large and complex household system of sharecroppers and the nuclear family system of day laborers—it is interesting to verify whether the social network of these two systems differed or not. Laslett (1988) hypothesized that social network beyond family bonds could be of much more importance, though smaller, for nuclear than it was for complex households. This is certainly true for rural Italy, where it is actually difficult to imagine households as separate self-supporting working groups. In particular, day laborers and their nuclear households could be part of wider contexts of related supporting family groups that could allow them to diversify activities and share risks. On the other hand, sharecropping households, though much more complex, were themselves often submitted to landowners’ decisions and they have to strictly respect the balance between landholding size, on the one hand, and household size, on the other hand (Poni, 1982). Hence, a network of non-coresident kin could represent even for this category a useful leeway from landowners’ constraints (Levi, 1990).

16In the light of that situation, we firstly examined the kinship size, here split into its two components of co-resident and non-coresident relatives. The calculations include parents, children, the spouse, sibling, grandparents, grandchildren and nephews, uncles and aunts, first cousins. The first six groups of relatives have been classified as close relatives, whilst the latter three groups as distant relations. The analysis of the social network has been then extended also to parents-in-law and children’s parents-in-law in order to take into account possible marriage alliances among families.

17The degree of kinship we used here to define the entire network of relatives depends on the limits imposed by data rather than on some substantial and conceptual hypothesis. There is no reason, in fact, to doubt that even distant relations could somehow play a role in affecting mobility of individuals, though it is likely arguable that closer relatives were the first ones to turn to in case of need. Besides, we cannot evaluate the role of close friends and other non-relatives as well as weak ties like acquaintances, and, moreover, we are not ready yet to consider the role of godparents, so important also in the Italian context to set up family alliances and strategies (Bigi et al., 1981; Alfani, 2004). The inclusion of “marriage alliances” in the analysis of the kinship network goes just in the direction to allow for other forms of relationships and strong ties beyond the kin network.

18Figure 1 shows the mean number of relatives by age of the index person and typology of residential pattern for the whole population. Starting from around 12 relatives present in Casalguidi at the birth of the index person, the kinship network displays then a steady decline up to the age of around 50-55 years, when about 8 relatives have been found.

19Although these figures may bring further evidences to the fact that in xix^th century Italy people were not enclosed in very large network of relatives, they could in any case impress for the very low levels found in this community. However, it is to remind that those values are certainly underestimates, as it is actually impossible to reconstruct all the possible links among relatives living in different households if we do not have previous and precise indication relating one person to another, that is to say a common ancestor. This situation regards mainly immigrants just settled down in Casalguidi, spouses arrived by marriage as well as those individuals who were already present in the first years of the period studied. For these people we cannot assess whether other relatives, especially ascending or collateral ones, were already present in the community.

20If we limit the calculation to the part of individuals whose father and/or mother were found within the population and linked to the index person (fig. 2), the mean width of the kinship network changes, obviously, especially for adults and elderly people, showing, respectively, 11.6 relatives at 30 years and 12.3 at 50 years. Moreover, the curve relative to non-coresident kin does not show any significant decline over the life-course, but from the age of 30 it shows a profile very similar to that of coresident kin.

Fig. 1-2

Mean Number of Relatives by Age and Typology of Residential Pattern. Whole Population (above), Only Individuals Linked to Father and/or Mother (below)

Mean Number of Relatives by Age and Typology of Residential Pattern. Whole Population (above), Only Individuals Linked to Father and/or Mother (below)

21However, it is very hard to say whether these increases reflect more reliable estimates of the real size of the kinship network existent at Casalguidi or they are rather the consequence of a selection of a more stable and less mobile sub-population. In fact, mobility is a key factor in the study of kinship, especially as far as its non-coresidential component is concerned. As Laslett said, “displacement is of crucial importance to kinship relationships and their effectiveness for support”, and that is the reason why he connected “kinship consciousness and kinship interchange with the complex rather than with the simple family household” (Laslett, 1988, 159).

22In previous studies always concerning Casalguidi, we found just evidence of a higher mobility of nuclear households in comparison to complex ones (Manfredini, 2003; Breschi and Manfredini, 2002, see table 1). The former group was quite mobile, showing a turnover rate of 70 per thousand, 35% higher than that of complex households. The necessity of nuclear family groups, mainly made up by day laborers, to move around in search of job was surely a key-factor.

Tab. 1

Household Migratory Rates by Household Structure. Rates per Thousand

Household structure In-migration Out-migration Net migration Turnover Nuclear 37.8 32.0 5.8 69.8 Complex 22.7 22.5 0.2 45.2 No structure 46.4 43.8 2.6 90.2 One-person household 82.3 131.9 -49.6 214.2 Total 34.6 33.3 1.3 67.9

Household Migratory Rates by Household Structure. Rates per Thousand

23On the other hand, the lower mobility of complex households, and especially of multiple ones which were for the most part formed by sharecroppers and tenants, depended on the opportunity of these agricultural categories to prolong their contracts and, hence, their permanence on the same farm. Moreover, the mobility of sharecropping households was also ruled and strictly controlled by landowners. They actually let a sharecropping family settling down on their farms in Casalguidi only once another one had left because its contract had not been renewed, and this conditioning explains the net migration figure close to zero.

24Accordingly to Laslett’s hypothesis, we would expect that the less mobile multiple households showed a broader kin network beyond the household than simple family-units. In our case, it has been supposed that the lower mobility of sharecroppers household allowed them to be deeply rooted in the territory, thereby able to raise their local human capital in terms, among others, of family networks (Oris, 2003). However, looking at table 2 we can see that the community of Casalguidi does not exactly follow that theoretical model. First of all, for each age considered, non-coresident relatives outnumber coresident kin in nuclear households, whilst the opposite happens in multiple ones. Furthermore, as far as only linked individuals are concerned multiple and nuclear households show similar figures of non-coresident kin for infants, but they differ both at 30 and 50 years. Consequently, unlikely what supposed by Laslett and Oris at these ages the external network of kin is actually larger among nuclear than it is among multiple households. Such differentials have much to do with close relatives than distant relations. Actually, while the difference in the mean number of non-coresident distant relations is very similar for whatever age over 15 years (mean difference between multiple and nuclear households = 0.064, t = 0.795, p-value = 0.430), the mean network sizes of close relatives living in other households of Casalguidi show a statistical significant difference (mean difference = -0.357, t = -2.950, p-value = 0.005).

Tab. 2

Mean Number of Relatives by Typology of Residential Pattern and Household Structure

Kinship network Nuclear Households Multiple households At birth 30 years 50 years At birth 30 years 50 years Whole population Coresident kin 5.1 4.0 4.9 9.1 7.8 8.4 Non-coresident kin 5.7 4.3 2.2 5.9 3.8 2.8 Close relatives 0.6 2.3 1.1 0.5 1.5 1.2 Distant relations 5.1 2.0 1.1 5.4 2.3 1.6 Total 10.8 8.3 7.1 15.0 11.6 11.2 Only linked individuals Coresident kin 5.2 4.2 5.0 9.1 7.7 8.1 Non-coresident kin 6.0 6.1 6.5 6.0 4.8 5.9 Close relatives 0.7 2.9 2.8 0.5 1.9 2.3 Distant relations 5.3 3.2 3.7 5.5 2.9 3.6 Total 11.2 10.3 11.5 15.1 12.5 14.0

Mean Number of Relatives by Typology of Residential Pattern and Household Structure

25A more detailed analysis reveals, however, that differences between the two household systems were also present in the structure of the network of non-coresidential distant relations, regarding namely marriage alliances. In fact, while during adulthood (between 15 and 45 years) the number of parents-in-law and sibling’s parents-in-law was significantly higher for members of nuclear households than multiple households (mean difference = -0.061, t = -4.035, p-value = 0.000), in old age the situation is reversed (mean difference = -0.186, t = 4.202, p-value = 0.000). An explanation of this finding must account for the form of living arrangement after marriage that differentiates day laborers and artisans, on the one hand, and sharecroppers, on the other hand. The former established usually a brand new family after marriage as an application of the neolocal rule of household formation, whilst the latter remained generally in their native multiple households. This entails that after marriage those couples forming a new family had their parents-in-law in other households, whilst those who remained in their native family coresided with at least the bridegroom’s parents-in-law. On the contrary, in old age this type of network was mainly formed by sibling’s parents-in-law, which could find hospitality, especially if widowed, in the households of the parents of their son/daughter-in-law. Obviously, multiple households were in the best position to provide hospitality to such special distant relations, and this explain why at older ages this particular network of relations was bigger for multiple than nuclear households (Laslett, 1988; Oris and Ochiai, 2002; Oris, Ritschard and Ryczkowska, 2004).

26At the end, all those evidences prove that the Laslett’s hypothesis is not perfectly fit to describe the kin network system(s) existent in Casalguidi: in this community, in fact, the lower mobility of complex households did not necessarily imply that they were embedded in significantly larger networks of related households.

27Is it however possible to hypothesize, for this Tuscan community, the existence of a reverse causal relationship between mobility and size of the non-coresidential kin network, in which the presence of many relatives outside the household may induce a reduction of emigration? It is obviously impossible to check for the existence of an inter-dependent network of mutual aid and support among related families that could somehow reduce the risk of emigration. Nevertheless, it is possible to argue that help and support could take the form of an exchange and displacement of household members in case of necessity, thereby avoiding the complete emigration of the household. First evidences of this mechanism were found by Levi for an area of the Piedmont region, where the solidarity networks were especially prevalent among non-coresident brothers or brothers-in-law, which were “ready to exchange farms, contracts, labor force, money” (Levi, 1990, 570). [4] However, it is also possible that the constraints imposed on the households of this Italian peasant society by agricultural contracts and, indirectly, also by landowners, could be so strong as to reduce the possibility of mutual help among related households, with consequences also on the possibility to prevent emigration.

28At the end, in the next section we will test whether a supporting network of relatives beyond the household could reduce the risk of emigration by harboring some members of related households. In the next section we will try to give answer to some of these points.

Mobility and the kinship network: a longitudinal analysis

29The opportunities provided by the complete reconstruction of individual and household life histories allow a deeper analysis of the role of the kinship network in the mobility of people and families. It is in fact evident that this relationship involves so many other aspects of everyday life that it is absolutely necessary to adopt a methodology able to analyze all those factors together as well as their possible interactions: Event History Analysis provides the right approach. The analysis presented here concerns both out-migration and internal mobility but not in-migration because, obviously, nothing is possible to know about the history of households and individuals before their arrival in Casalguidi.

30The technique of Event History Analysis is one of the best and most powerful statistical tools in dealing with longitudinal data (Allison, 1984; Gutmann and Alter, 1983). Briefly, this technique is based on a regression model, in our case the logistic model on account of the discrete partition of time here used accordingly to our sources, in which the dependent variable indicates whether or not the event studied has occurred in any given year. The group of explanatory variables included in the model (called covariates) describes the context (economic, epidemiological, etc.) existing at the beginning of each year; the aim is to verify the possible influence of such factors on mobility. In the case of categorical variables, the estimated coefficients indicate the risk of migration (or exit from the household) for each category in relation to a reference category, for each of the covariates included in the model. On the other hand, for quantitative and continuous covariates, the coefficients express the relative risk for each unitary change of the covariate.

31From an operative point of view, we run two sets of models. Each of them did concern only individual mobility, thereby excluding emigration and mobility of entire households. In the first model, we limited the analysis to individual out-migration, whilst in the second one only individual internal mobility has been considered, excluding all the movements outside the community. Mobility for marriage, characterized by a completely different causality pattern, and one-person households has not been taken into account in any of the above models.

32The aim is at evaluating possible differential impacts of coresident and non-coresident kinship networks on the likelihood of leaving the household to go either outside the village or in another family group of Casalguidi. As anticipated above, our hypothesis is that the presence of an extended kinship network in loco might somehow prevent the emigration of entire households by giving hospitality to some of their members. If this holds true, it follows that the presence of a kinship network should increase internal mobility of single individuals among related households of Casalguidi.

33Furthermore, a final model will try to assess, by means of interaction terms, whether a relationship between household structure and non-coresident kinship network on the likelihood of internal mobility is present. This model aims at giving answer to the hypothesis of a different household emigration rates between complex and nuclear households on account of a differential recourse to displacement of individuals among related households in the community.

34We used two different covariates to capture the possible effects linked to the presence of close relatives, on the one hand, and distant relatives, on the other hand. Households allied by marriage were also considered. Yet, we decided to consider the number of related households rather than the crude number of relatives since, in our opinion, a large number of related families diversifies and expands the network individuals might turn to in case of necessity.

35Other covariates have been included in the models in order to control for the very complex mechanism of mobility. Individual as well as household level variables have been considered along with other covariates controlling for communal factors, such as economic short-term indicator (wheat price), year(s) of cholera epidemic and so on.

36Socio-economic structures were approached as well. We used two covariates, Tax Level and the possession of the house, in order to capture possible confounding effects due to constraints linked both to the economic position and the possession of the house in which the family lives. Owning a house represents a clear sign of a strong tie with the territory and the community and it could actually represent a powerful disincentive to emigrate regardless of the size of the kinship (see, for example, the study on the French population of Ancien Régime by Le Roy Ladurie, 1998, 301-302). More partially, this indicator could also be used to control for the more stable component of the population.

37Table 3 shows the results for emigration and internal mobility of individuals. The discussion will be limited to the issues more strictly linked to the kinship network and the household composition. [5]

Tab. 3

Individual Out-Migration and Internal Mobility. Risk models. Casalguidi 1820-58

Covariates Out-migration Internal mobility Mean Rel. risk Mean Rel. risk Age (ref. 25-54 years) 57.6 1.000 57.6 1.000 < 25 years 29.9 1.046 29.9 0.841 * 55+ years 12.5 0.926 12.5 0.561 ** Sex (ref. Male) 50.3 1.000 50.3 1.000 Female 49.7 1.040 49.7 1.099 Household structure (ref. Nuclear) 50.3 1.000 50.2 1.000 Complex 48.4 1.460 ** 48.6 3.114 ** No-structure households 1.3 3.465 ** 1.2 4.981 ** # of households with close relatives 0.7 0.965 0.7 1.293 ** # of households with only distant relatives 0.8 0.873 ** 0.8 1.027 # of households allied by marriage 0.1 1.044 0.1 1.001 Tax level (ref. Low Tax) 27.9 1.000 28.0 1.000 High & medium tax 17.9 0.791 ** 17.9 0.615 ** Untaxed 54.1 0.700 ** 54.1 0.889 Property of the house (ref. Yes) 27.0 1.000 27.0 1.000 No 73.0 1.005 73.0 0.574 ** Cholera epidemic (ref. No) 97.2 1.000 97.2 1.000 Yes 2.8 1.027 2.8 1.265 Wheat price (logged) 2.530 ** 3.1 1.267 Log likelihood 7615.4 -4822.7 Total number of household-years 79.627 78.984 Events 1608 929 * significance level at 0.05 ** significance level at 0.01

Individual Out-Migration and Internal Mobility. Risk models. Casalguidi 1820-58

38The two models present both clear differences and similarities as for the coresidential and non-coresidential components of the kinship. First of all, members of complex households were more likely either to emigrate or to change family within Casalguidi than members of nuclear households were. It is evident, given the higher household out-migration rates of the latter, that complex households were more prone to modify their size and composition rather than to emigrate as a whole (Manfredini, 2003). As already anticipated, these complex households were set up, for the most part, by sharecroppers, whose family size had to be contractually tailored on the landholding size. This fact implied that members of sharecropping families were sometimes compelled to leave (Poni, 1982).

39Unlike the household structure, the non-coresidential kin network shows different responses in the two models. As for out-migration, the relative risks associated to the network of related families prove the protective effect on emigration played by the presence of households of close and distant relatives (the latter shows a significant 13 percent decrease of the risk).

40Opposite is the effect on internal mobility. Individuals with related households in loco, in particular those including close relatives, were around 30 percent more likely to move among the households of Casalguidi. At the end, the logical conclusion is that the presence of a network of related households represented a potential barrier for preventing from emigration, but, at the same time, it made easier exchanging and dislocating people among households when needed.

41In table 4 we show the results concerning the models with interactions. As it is immediately evident, the inclusion of the interaction terms does not bring in any significant increase in the explanatory power of any of the two models. Hence, there is no significant differential by household structure in the recourse to non-coresident kin, either close or distant.

Tab. 4

Individual Out-Migration and Internal Mobility. Risk Models with Interactions

Covariates Out-migration Internal mobility Mean Rel. risk Mean Rel. risk Age (ref. 25-54 years) 57.6 1.000 57.6 1.000 < 25 years 29.9 1.046 29.9 0.840 * 55+ years 12.5 0.926 12.5 0.561 ** Sex (ref. Male) 50.3 1.000 50.3 1.000 Female 49.7 1.040 49.7 1.098 Household structure (ref. Nuclear) 50.3 1.000 50.2 1.000 Complex 48.4 1.459 ** 48.6 3.033 ** No-structure households 1.3 3.445 ** 1.2 5.219 ** # of households with close relatives 0.7 0.964 0.7 1.261 ** # of households with only distant relatives 0.8 0.873 ** 0.8 1.027 # of households allied by marriage 0.1 1.045 0.1 1.001 Interactions terms Complex hh * # of hh with close relatives 1.001 1.032 No-str. hh * # of hh with distant relatives 1.044 0.616 Tax level (ref. Low Tax) 27.9 1.000 28.0 1.000 High & medium tax 17.9 0.791 ** 17.9 0.616 ** Untaxed 54.1 0.700 ** 54.1 0.890 Property of the house (ref. Yes) 27.0 1.000 27.0 1.000 No 73.0 1.005 73.0 0.575 ** Cholera epidemic (ref. No) 97.2 1.000 97.2 1.000 Yes 2.8 1.028 2.8 1.265 Wheat price (logged) 2.531 ** 3.1 1.266 Log likelihood -7615.4 -4822.1 Total number of household-years 79,627 78,984 Events 1608 929 * significance level at 0.05 ** significance level at 0.01

Individual Out-Migration and Internal Mobility. Risk Models with Interactions

42These results permit to describe quite clearly the two different strategies that complex and nuclear households set up with regard to mobility. The former were less likely to emigrate as whole family groups, but, on the other hand, their members were more likely to move individually than nuclear households’ members did. Conversely, nuclear households were more likely to emigrate as single family-unit, but they do show lower risks of individual mobility either within the community or outside.

43Furthermore, these movements did not happen on account of a differential recourse by household typology to external kin-network. In fact, the presence of relatives beyond the family bonds acted to prevent people from emigration regardless of the type of household they come from. The presence of related households in Casalguidi was therefore fundamental in increasing the circulation of individuals among different family groups at Casalguidi, and, at the same time, it dampened the departures from the parish territory.

44On the other hand, households tied by marriage alliances do not result involved in this pattern of circulation of individuals among different family groups.

45Identical models using only individuals linked to father and/or mother as population at risk do show results consistent with those displayed above for the entire population.

46We can therefore argue that complex households may be more flexible and able to adapt themselves whenever urged by necessity, modifying size and structure as occurred, for instance, on the occasion of the death of the household head (Manfredini, 2003).

47It is therefore probable that agricultural contracts and constraints linked to head’s occupation rather than kin-network might significantly affect the chances of mobility of farmers.

Conclusion

48This study represents a preliminary attempt that aims at following Levi’s suggestion to study circulation of individuals among related households and its influence on the mobility patterns in sharecropping communities.

49Our findings suggest that the existence of large local networks of non-coresident kin might act as a barrier to emigration of individuals, probably by providing them support and the opportunity to move within Casalguidi without leaving. Moreover, this behavior concerned equally members of nuclear and complex households. This fact gains even much more prominence in the light of the great differentiation characterizing the two household systems, not only from an economic point of view but also in terms of mobility pattern. In fact, sharecropping households were more rooted on the territory, and used marriage and mainly individual mobility of members as a sort of “preventive check” to control household size and preserve the actual income and household welfare (Cazzola, 1996). They seem fitting what Derouet (1994) called a “defensive family systems”, aimed at preserving and maintaining the contract that tied the household to the land “en se tournant vers la restriction de la population” (Lorenzetti and Neven, 2000, 87). On the other hand, the poorer and landless day laborers were less stable on the territory and move around in search of work as entire family group. The necessity to diversify both income and the opportunity of work among the household members, determined small-sized household, on the one hand, mobility and geographical dispersion of kin, on the other hand. [6] At the root of these contrasting differences in the family system, economic status, and mobility pattern is the tie to the land, which was definitely lacking on the day laborers side.

50However, the kin network played a strategic role in re-dislocating kin for both these social groups of agricultural laborers. More long-term rootedness on the sharecroppers’ side, short-distance movements and fragmentation of family units on the day laborers’ side, both produced kin networks, which, in this case-study, we found to be larger for day-laborers than sharecroppers, confuting what supposed by Laslett.

51These results are far from being definitive and many analyses have to come to illuminate and provide robust evidences about the influence of the kinship network on the mobility of individuals in rural Italy. Much more effort will have to be addressed in a better and more precise reconstruction of the kinship network, not only in quantitative terms, by widening the time period, but also in qualitative terms. This will entail much more detail in specifying the various relatives featuring each kin network since mobility could be influenced not only by the non-coresident kinship size but also by its structure.