I focus on an important recent study that employs a large database and manages to produce significant results, despite formidable source problems. Paul-André Rosental’s Invisible Paths: Space, Families and Migrations in the France of the Nineteenth Century, draws on genealogies of family groups based primarily on civil status records.

Paul-André Rosental studies the French by selecting family lines that originate throughout the country. Over the course of the nineteenth century, Rosental found, there was a steady trend toward living farther and farther from birthplace before marriage. The countryside was an active and even “agité” arena in which sedentary folk were in the minority and in which migration increasingly focused on cities. Rosental did not find people displaced by events (fleeing the countryside on the heels of the phylloxera epidemic that killed the vines of Burgundy, for example) that created the famous “rural exodus”; rather he saw individuals moving within two kinds of family models that I will discuss below. Thus Rosental disputes the idea of the sedentary countryside disrupted by the industrial revolution and attendant rural exodus.

Rosental reveals a mobile French population. In Sentiers invisibles, family lines and tendencies are under scrutiny, and reveal a striking heterogeneity in migration patterns. Family data are precisely the kind of information that is rich in Rosental’s Sentiers invisibles. Founded in the 3000 Families study initiated in the early 1980s by Jacques Dupâquier and Denis Kessler (1992), this study followed families with patronymics beginning with the letters “Tra”. Rosental’s major resource is some 45,071 marriage records for couples married between 1803 and 1832 and their descendants who married before 1903; the couples originate in twelve departments—from Flanders to Provence and from Brittany to Alsace—that were chosen for the quality of their civil status documentation. Marriage records are extraordinarily rich documents, for they include the birthplace of the bride and groom; the occupation, present residence, age and capacity to sign one’s name of the bride, groom, their parents, and four witnesses; the family relationship between witnesses and the bride or groom. Rosental uses the records of 40,389 grooms to make fundamental observations about migration between birth and marriage. The source for the heart of Rosental’s study is 97 family genealogies based on patronymics; these were constructed for lineages that included at least one male to marry after 1860 with data from the birth, death, and marriage acts, augmented by auxi-liary archival material on successions and twentieth-century marriage records in Paris (Rosental, 1999, 41 and Appendix A). Minimally, date and place of birth, marriage, residence at marriage, residence and profession at marriage and death were collected for each individual; for some, residence and profession at the birth of each child were available, and as well as relational information about wedding witnesses. Omitting those who died before age 18, Rosental’s sample of genealogies consisted of 1119 individuals in 97 lineages. (Also omitted were those in a second or third marriage, those with information only before 1803 or after 1914, and all Alsatians).

There are important lacunae in this data set. Women are missing from much of the record, although not completely nor systematically. Because the record gathering was carried out for patronymics that begin with “Tra”, descendants of women in the study were lost to the genealogies (unless their children were born out of wedlock or they married someone with a “tra” surname). The loss of descendants is not recorded systematically. Moreover, as Jan Kok contends (this issue, below), certain biases are built into a study that has for its database genealogies that have been fortunate enough to survive for six generations.

Perhaps more important, Sentiers invisibles is constructed around hypotheses about the family that are introduced through accounts of exemplary families that readers come to know in detail. These families are from nine of the 12 departments studied (no family from Calvados, Côtes-d’Armor, or the Haute-Rhone is discussed); two each from the Nord and the Sarthe are discussed. Rosental’s interest is in family dynamics, not in regional patterns, thus there is not a sense of representative information by region.

Nonetheless, each study can, and does, address key questions about migration in nineteenth-century France, such as what proportion of French remained in their birthplace. Of 40,389 men from the 3000 Families study, over half of those born in villages remain in their village of birth (52.7 percent) and the proportions are higher for small-town and city men (Rosental, 1999, 41-42). Rosental’s data suggest that one’s migration itinerary varied substantially by the size of one’s birthplace a village, a bourg of some 2500 residents, a town, or a city of over 30,000. Village men were more likely to move, but not to move far (rather, they were likely to remain in their home department).

Rosental demonstrates that literacy affects migration patterns. (Similar observations have been posted by others, for example, Sewell, 1985, chap. 4; Heffernan, 1989, 32-42). He mea-sures the ability to sign one’s name from marriage certificates and reports that of the 643 members of his sample who moved before marriage, those who left the department and those who headed for greater Paris were more likely to be literate than those who stayed at home or migrated locally (Rosental, 1999, 190-94). Finally, Rosental asks what proportion of people ever move to Paris. He finds that of those who moved in the last twenty years of the nineteenth century, 49 percent of those who left their home region went to greater Paris (Rosental, 1999, 30-32, separates out data by period to show patterns for 1881-1902).

The interest here, however, is not in a focus on the kinds of questions that any study of French migration would hope to answer; after all, Rosental’s data base is unique, as are his goals, as his subtitle indicates: “Space, families, and migration”. He engages the rich information at his disposal about 97 lineages to reframe our thinking about migration and the family. Understandably impatient with administrative categories for local areas, Rosental (1999, chaps 3-4) finds a solution in considering the “territory” in which a family moves. In pondering connections between the family and mobility, he finds a spectrum of what Americans might call “connec-tedness” among family members; this is embedded in the occupational cohe-rence of family members over time, the frequency of relatives’ appearance in the role of witness at family weddings; and in like literacy patterns within the family. (Literacy, as recorded on marriage acts, can only be measured as the ability to sign one’s name). Rosental suggests that family dynamics are less the result of patterns of dispersal or migration than they are part and parcel of what makes people stay near home or leave—the social and professional stock and connections on which generations can draw. By way of testing his propo-sals, Rosental characterizes families as either “exocentric” or “autocentric,” and investigates who moves. It is not surpri-sing that he finds, among other things, that some occupational “pioneers” ope-rate through their elder sisters, who had moved and married into a professional group new to the family; or that people from large families are more likely to leave home. Yet family dynamics are not a hard and fast determinant of mobility or stability, Rosental argues. Rather the strength of family ties creates a more or less important location for mediating and cus-hioning the large-scale changes that affected the nineteenth century. Mobility depends not on the individual, but also on whether one is from a large or small family, a differentiated or homogeneous group, a wealthy or modest household, an autocentric or exocentric (close-knit or loosely knit; connected or disconnected) family (Rosental, 1999, Parts II and III).

On each dimension, Rosental (1999, 190-93) begins with genealogical observations about one to three family lines that lead to a generalizable hypothesis, and thence to the larger data set for the 3000 Families project which is summoned to respond. In some cases, the link between hypothesis and information seems to be tenuous. For example, Rosental looks to heterogeneity within the family as a way to explain migration away from an autocentric family—and turns to the capacity to sign marriage records as a signal of heterogeneity, lacking other measures. In autocentric families, he writes, migration would be a resource to fight the unbea-rable unevenness in family fortunes, and Rosental links migration from autocentric families with uneven literacy patterns in those families. For exocentric families, the logic is, the family is less important as a reference group and so being among the less fortunate in such a family carries less a sting (Rosental, 1999, 192-95). As tortured as some of these points are, Rosental’s overall pre-sentation, seated in the literature on the social organization of migration, does remind the reader that migration is a social phenomenon that, however connected to large-scale changes, runs on human ties. And he is convincing on another point: sedentarity is neither a “natural” nor a “conservative” state, but rather is linked with being part of a connected, coherent family group.

Rosental has produced a study with important findings about human migration behavior. Given unwieldy and problematic sources, this should be understood as a triumph of intelligence over archival information. Les sentiers invisibles is rooted in French population and mobility history, the offspring of indefatigable research and a deep know-ledge of French mobility patterns. The study, however, is unable to yield systematic findings about gender or data about female migration, in which I and many other scholars are vitally interested. And although the insights from Rosental’s study are stimulating and intriguing, it will be some time before scholars are able to compare such a uniquely-constructed set of observations to stu-dies of family and migration elsewhere.