Students of migration usually have less information on the mobile population than they do on the sedentary. This is because data for the study of migration are tied to specific local or regional record-keeping units, so when people cross a boundary they become lost to the observation of the local clerks. Tracing mobile individuals thus requires sources that cover a broad regional, even national, area, plus plenty of time and dedication. No wonder, then, that few historical migration studies based on the micro level of the entire individual life course have been attempted. Much scholarly attention has been drawn to long distance or even international migration, that often had strong impacts on the societies and the indivi-duals involved. However, this kind of migration only forms the proverbial tip of the iceberg. Recent studies disclose that most adults have moved at least a number of times during their lives, but these moves were generally over a short distance.

In this comparative analysis, the Netherlands are represented by the province of Utrecht. We use a database with a detailed life course reconstruction, including migratory moves, of several thousand local-born individuals (see below). These people are followed when they left Utrecht wherever they went, as long as they remained within, or returned to, the Netherlands. The small province of Utrecht is located in the center of the country. In 1850, the total population amounted to a mere 149.380 inhabitants. The surface of the province was 1323 km²and the population density thus 113 inhabitants per km². Almost a third of the Utrecht population (47.781) lived in the provincial capital, named Utrecht as well. Another 12.377 inhabited Amersfoort, the second largest city in the province. The rest of the population lived in the 89 other municipalities, all counting less than 5000 persons. In 1900, the total population of the province had increased by 68% to 251.034, a density of 190 inhabitants per km². This increase was entirely caused by natural growth. The birth rates remained high (still 35‰ in 1891-1895) but the death rates had dropped quickly since 1875 (20,9‰ in 1891-1895). Immigration had hardly contributed to the growth. The province as a whole was not affected dramatically by the agrarian depression of the 1880's and 1890's, but it did experience a net migration deficit of 20‰ in the period 1890-1900. On a regional level, the picture was very diversified. In the western and southern parts of the province, where dairy farming was predominant, the out-migration rates were high. However, residential areas in the central regions attracted many newcomers from outside the province. In particular the city of Utrecht, which was the nation’s railway hub and had attracted metallurgic industries and commercial services, had grown fast to a total of 102.086. In-migration did play an important role here, with large surpluses particular in the 1880's. In 1900, the number of Utrecht municipalities had fallen back to 72, due to annexations and combinations. By then there were four municipalities with between 5000 and 10000 inhabitants.

The sample from the American North is derived from genealogies in which the first settler in America came to what is now Massachusetts before 1650. During the period of interest in this paper —roughly 1820 to 1880—the population was expanding rapidly through natural increase and also spreading geographically as new lands were opened up for settlement. We refer to the sample in what follows and in several tables as “New England”, but that is simply shorthand for the descendents of the founding ancestors who by the time of this study were to be found across the American North, but largely east of the Mississippi (The area West was not yet colonized.). The majority were living in the New England States and New York but there were several already in what is now the Midwest. This area was, obviously, a much larger region than Utrecht province. Fertility was very high and child mortality low.

The comparison we undertake in this paper is most congruent for men born 1820-1840. The Dutch data is constrained before 1820 because of the late beginnings of their system of registration; the American data was constrained by the publication dates of the genealogies, most of which were compiled at the end of the xix^th and beginning of the xx^th centuries, making it difficult to obtain complete life course information for men born after 1840.

The genealogists saw their basic task to be the linking together of everyone with the same surname wherever they might have lived, in order to produce a history of the descendants from a founder in the male line [2]. Thus the genealogy follows migrants and their families as they moved across the American North. In contrast to earlier “town” studies based upon family reconstitution of single communities, a genealogical database allows the study of links between local communities, and the conceptualization of space as consisting of flows rather than as bounded units. Nine published genealogies were chosen on the basis of the quality of the information about migration. They were also selected so that the founding ancestors settled in different parts of what is now Massachusetts.

There is an inherent bias in genealogies to include stayers at the expense of movers, as indeed, there is in any source for migration. Still, it must be said that the migration rates and distances migrated we report are very high. Rates for the colonial period were higher than previous estimates from reconstitutions of the populations of individual towns in New England. The data do provide much more information about migrants than previous studies based upon rates of migration out of, or into, single locations.

For the Netherlands, we make use of data collected as part of a large project called Historical Sample of the Netherlands (HSN). Begun in 1991, this project constructs standardized computer files with historical micro-data in order to facilitate scientific research in such fields as mortality, social mobility and migration. Eventually, the database will consist of the reconstructed life courses of more than 70.000 individuals, whose background, literacy, occupations, eventual marriage(s) and migratory moves are known (Mandemakers, 2000). The first step was to create a file with the yearly number of births in every one of the 1400 different municipalities in the Netherlands between 1812 and 1922. The next step was to let a computer program draw random numbers from the number of births per municipality and per year. These numbers were assigned to specific birth certificates. In this way a sample of a 0,5 % of all persons born in The Netherlands between 1812 and 1922 was taken. All the information that can be found in the birth certificate was stored in the computer. At this point an extensive search started to retrace the life courses of these persons. Firstly, the decennial indexes on death registers were checked to find the certificates of persons who have died in infancy or childhood. Because the death registers always mention birth place, age and the names of the parents, linking individuals is rather straightforward. In a similar way, marriage certificates are retrieved by checking local or provincial indexes. The Dutch population registers, which started in 1850, mention the provenance and destination of all individuals. In principle, the sample persons can be followed from one municipality to another. The Utrecht sample consists of 3660 persons born in Utrecht. This has been reduced in this paper because many of these individuals did not survive to adulthood.

The strength of genealogical data lies in the genealogical connections themselves. It is quite a rich source for the study of the dispersion of families. A second advantage is the coverage of a long span of time, longer than the registration systems instituted in the middle of the xix^th century permit. However, they often fail to reconstruct the movements of women, because they change their surnames. Trajectories based on genealogies depend on vital events. By supplementing with other sources such as censuses and city directories, some of the black boxes can be filled in, but it is not possible to glimpse all the detailed comings and goings that occurred in people’s lives. In particular the timing of leaving home, the movements of adolescents, and moves after childbearing cannot be studied in detail. Also, observation of childless couples tends to be infrequent.

In this section, we concentrate on the rural population of both areas and we compare their migration flows in term of distance and type of destination. We discuss the approximately forty year period, between birth of parents and birth of their last child, using only the men. To begin, we described the migration distances in terms of concentric circles using also the sizes of places in the two national regions. A move under 6.5 kilometers (4 miles) is equated with staying [6]. Table 3 reveals in a single glance that the migration systems were quite different. We notice that about 30% of the New Englanders had their last child within six kilometers of their place of birth. For the Dutch, this percentage was almost twice as high. So, the Dutch tended to stay in or return to their areas of birth far more often than the Americans. Also, many more New Englanders moved over long distances: 49% of the American men migrated over forty kilometers before having their last child, versus only 6-13% of the rural Utrecht men [7].

Now we will re-examine the same data, but from the perspective of the different stages within the life course. Having looked at flows in space, we turn to looking at flows across the conventional “boundaries” of the human life cycle which both sources of data permit: moving as a child, youth migration, and movement during and after the childbearing years. We start with a broad mesh of life cycle stages and work down to the smallest units that can be observed in both sources, birth intervals. In Table 4, we extracted data from the HSN filtered to resemble genealogical data in order to provide summaries of migration at key intervals during the life course. So as to be comparable to the genealogical data, the information from the Netherlands is based upon the whereabouts of men when they experienced the particular vital events that frame the interval. As in the case of the flows we have already discussed, here there are also some stark contrasts: the time of greatest movement in the Netherlands is between the ages of 15 and marriage, while in the American data it is later, both during the child bearing years and the period between the birth of the last child and death [8].

Movements before the child-bearing period were very predictive of movements during that period. In the Netherlands, 26% of persons who had moved earlier, would also move during the childbearing phase, whereas this percentage was only 10% for those who had not moved earlier (N=753). In New England, these rates were much higher but still men who had moved earlier were more apt to move during the child bearing period: 50% of the men who had moved before having their first child would move again and 20% of men who had not moved earlier, did so for the first time during the childbearing stage. Who were these “double movers”, and how can we explain a pattern of frequent moving? In the following table we look closer into the family background, type of residence and occupational characteristics of the heads of families in our samples.

In what follows our mesh becomes very fine, the smallest intervals both datasets are capable of observing. We ask whether or not couples migrated between the births of successive children and our intervals are often only two years long. Figure 3 shows there is a decline in the rate of moving from earlier to later birth intervals in both countries, but the rates at each interval are higher in the American North, as might be expected since the overall frequency of moving during the childbearing years was higher, as we have just seen. There are “bumps” in later intervals in both sets of data, the Dutch remaining flat from 4^th to 7^th child and in the American North there was more moving between 6^th and 8^th child than between earlier births. The rises during later intervals may be an indicator of heterogeneity in the behavior of large vs. small families. Families that have more children seemed to have moved more, as indeed would be consistent with the pioneer pattern of moving after the oldest sons would have been able to help by clearing land.

In our final analysis, we compare the tendency to move between two specific birth intervals, one early and one late in the child-bearing period. The analysis gives a sense of the unfolding of family behavior over time. We have also made a regression to predict moving during the entire childbearing period in the two countries, but, many of the variables which seemed important when we looked at the probability of moving during this larger unit of time were no longer significant when we looked at each particular interval. For example, in both countries farmers were less apt to move during the entire child bearing period than men in other occupations. Smaller settlements (the villages under 5000 inhabitants) were more apt to be left in both countries than towns and cities. Migration was more frequent among later cohorts, also, in both countries [10]. Also important in both countries was whether the person had moved prior to the childbearing period: men who had moved before were significantly more apt to move during the child bearing period, the phenomenon described as double movers. Apparently, these factors encouraged moving, but not at any particular point during the childbearing period.

In this paper we have sought to demonstrate, first, that genealogies, even though coarse-meshed, provide adequate coverage of moves in many contexts, hence that they are a useful and valid data source, especially if they are used with appropriate caution. This is very helpful since the finer-meshed register data do not exist for many countries.