1According to Lawrence Stone, data collecting for its own sake belongs to some kind of slightly queer behaviour as stamp or butterfly collecting1 Therefore, we have first of all to explain why we are preparing a prosopography of the urban elites of the imperial city of Augsburg 1500-1618 and another one of the prelates of the Roman curia at the time of pope Paul V, 1605-1621. Both prosopographies are meant to support a specific theory of social behaviour in general and of elites in particular. And by doing so they should explain better how the urban and the papal systems of government worked and produced their specific results than previous research has been able to do2

2The introduction of a certain kind of ecclesiastical reformation in Augsburg was neither based on converging interests of people with similar social characteristics nor on an incidental confluence of individuals longing for eternal salvation. And the ecclesiastical and secular policy of pope Paul V cannot be explained completely in terms of personality, ideology and state interest. In contrast, we claim that in both cases appointments to leading positions and subsequent decision making have been influenced by regard for personal relationships to a much larger extent than previous research has been willing to admit It is not a certain set of equal social data or an abstract adherence to a common ideology which constitutes elites and produces the necessary consensus for decisions, but the personal connections of the people involved, which bring about, foster and canalize interaction. Therefore we borrowed certain elements of the social anthropologists’ network analysis and started to study our cases as social networks of structured interaction, considering social relations and interaction not as mere indicators of something else, but as the very essentials which constitute human groups and actions.

3However, whereas social scientists are able to observe the working of social networks during actual interaction, historians have to cope with the temporal distance from their objects and the pre-selection of data which characterizes their sources. But we can try to become less dependent on coincidence by basing our conclusions not so much on those real interactions we incidentally come to know about, but on the very relationships which made such interactions possible and canalized them. Thus we emphasize the interaction potential instead of actual interaction itself. And the application of the more sophisticated mathematical techniques of network analysis is no longer necessary, because under such conditions it can only produce illusionary pseudo-exactness. It seems reasonable that we should pay particular attention to four kinds of relationships which have been of primary importance for early modem European society: Kinship, with ritual or artificial kinship by marriage or godparentship included; group solidanty produced by common place of origin, by membership of the same religious order, or the same students’ fraternity; friendship, not as emotional attachment, but as an instrumental alliance of giving and receiving. In the language of the sources "friends" quite often designate kin, but are also used as a kind of euphemism for patron-client-relationships, the fourth essential type, defined as a rather stable alliance of unequals, where the stronger partner can require certain duties of the weaker one in return for his protection.

4However, if we intend to evaluate the weight of informal relationships in history correctly, we have to extend our research beyond those persons we come across as members of "networks" and have to include all potential position holders and decision makers. For this reason we became involved in prosopography The basis of our research on the Roman Curia will be a prosopography of all persons holding the rank of prelate or more between 1605 and 1621, because the Roman power elite is recruited from this group In the case of Augsburg we started with a list of office holders, but discovered that this list has already to be considered as a result of "networking". At the same point we decided to include the whole economic elite to find out to which extent political "networks" coincided with economic partnerships in that New York of the 16^th century. Since economically important people almost by definition were merchants we took the records of membership of the Kaufleutestube, the general organization of merchants (not to be confused with the much smaller merchants’ guild) as our starting point

5The next step had to be the decision which data to include besides basic biographical information. In the case of Rome, when that project was started 20 years ago without any electronic assistance, I decided to include everything which might provide us with information on social relations and interaction of the person in question, even if this information remains hidden unless additional data might provide us with a clue. Of course, every hint concerning kinship, group solidarity, friendship, and patron-client-relationship remains essential. The correspondence of the patron of patrons at the Roman Curia, the pope’s nephew cardinal Borghese, provides us with abundant data in that field. In addition, the career and the benefices of the prelate have to be documented as completely as possible, because everything in this context might indicate a relationship of the kind we are looking for. Finally, other possible indicators have to be included, e g. the habit of sending Christmas greetings and presents quite regularly. All this information is arranged in chronological order, additional data of other members of the family included Sometimes "networking" becomes self-evident from the mere chronological sequence of events (cf. appendix 1). Our draft prosopography has reached letters MAN now. The project is still conducted manually until the end of this phase of work has been reached3.

6When we decided to extend our Augsburg project to include all merchants in the early 1980s, computer technology had made a tremendous progress. Therefore we started this second prosopography on PC. In contrast to the loosely structured prosopography of Roman prelates, which remains always open for additions and variations, in this case we had to fix all possible variables in advance. But two pilote studies on the 73 mayors of Augsburg 1518-16184 and on the external networks of Augsburg’s elites5 had provided us with the necessary information. We knew in advance that Augsburg’s sources, in contrast to the Roman ones, almost never contain explicit statements about social relationship and its working But there is plenty of indirect evidence which had to be included in our coded list of variables (cf. appendix 2). Besides the basic biographical data (L) we had to take into account many possibilities of kinship (E, G, P, K, KK, S, SS, C, V), education (A), economic activities (B, F, Z, T), office holding (AA, PO, VW, ZE) and social rank (ST), property and taxation (BV), membership in fraternities and similar organizations (M), other positive active interaction such as patronage, wardship, godparentship (SP), other positive passive interaction such as clientage, being a ward or a godchild (SK), negative interaction such as being a party in a law suit (SN), finally other kinds of interaction (UE). Let us assume that a citizen of Augsburg had two married brothers, got married once himself, had three married children, three in-laws, two uncles and two aunts, went to school and to some other educational institution, took part in four different economic activities with three partners or employees, held two urban offices and nothing else. An accurate description of such a moderately active person will need 111 variables of our list, if we have complete information, which is very rarely the case. Most Augsburgian social biographies remain as fragmentary as the Roman ones (cf. appendix 3).

7Nevertheless on the basis of the evidence collected so far we have every reason to believe that many social relationships and much interaction will become reconstructable from our prosopographies, enough at least to demonstrate convincingly the ways of networking in 17^th century Rome and 16^th century Augsburg

Reconstructing connections: Technical aspects of the Augsburg project6

8When we began our project in 1983, the state of the computer art did not favour historical analyses of our design Special programs for historians, as for instance KLEIO (Max Planck Institute for History, Göttingen) or PROSOP (C N. R S, Paris I University) were just in the making and available only for mainframe computers Their emphasis often lay on histoneal demography, the study of tax records or collective biographies of a more statistical type, but not, as in our project, on the study of relations, i.e. network analysis7.

9The subject of our analysis is the merchant class of the Imperial city of Augsburg from about 1500 to 1620 – a group of about 1650 individuals. These persons (1.) have to be identified as clearly as possible through a number of personal variables, and (2.) must be reconstructed in their respective connections to other persons inside and outside the documented list of members of this elite group.

10(1.) Our list of variables identifying each and every person (or case, cf. Appendix 2) includes characteristics like the years of birth, death, and marriage, the parents’ names, the place of birth, the father’s occupation or position, church membership and the place of burial. At this stage, however, we are already confronted with a lot of missing data because of inadequate source materials, although the program has to keep in readiness the full range of according variable fields. Moreover, the information which is to be fixed in the according variable fields consists of texts (i.e. alphanumerical values), as well as figures, in this case chronological values (i.e. years).

11(2.) The aim of our project is the reconstruction of social relations, which may occur principally in two forms. First, they can be set up structurally or appear as ascribed relations, as in the case of consanguinity, marriage, or friendship. Second, they can be set up by event or appear as acquired relations, established by attendance of a certain school, membership in a certain circle, business contacts etc. The actual quality of a structural relation is determined by its whole cultural-historical context. Sometimes an acquired relation is merely the expression or the result of a structural relation, while in other cases it is a genuine additional relation. Our classification of network variables or social relations in a narrower sense (kinship, political relations, business relations, etc., cf. Appendix 3) therefore must not be misunderstood as a historically selective classification.

12Every group of network variables is systematically structured yet open to supplementary variables. For example, we can expand the list of spouses and their according sisters, brothers etc. by simply adding new variable fields Wherever convenient, we can furthermore attach "open" (formally structured, but non-defined with regard to contents) variables, or vanable fields, to fix data of "other relatives" or "other business contacts".

13Not only the diversity of value types and variables but also the relatively high number of variables require the use of a powerful, flexible and fast software program. This program also has to support the manufacturing of structured hardcopies (printouts) of all variables and values at all stages of the project, leading to the publication of an according biographical dictionary (cf. Appendix 3) to be distributed to, used by and improved by all historians of Early Modem Augsburg. Finally, the program has to be easy to learn and to use.

14When we started our project, the only standard user program serving our needs was dBASE, now available in its versions III+ and IV8. DBASE is a relational database program because it deals with related information and it enables its user to draw information from different files and databases linked by a common field9.

15Version 1.1. of dBASE IV includes one billion records (in our case: persons with their according personal and network data), 255 fields to be processed in one file, and 254 characters one may write into a field to express the value of a vanable. Additionally, nearly 100 files may be opened simultaneously. In other words, the program enables us to conveniently analyse the main biographical features of the population of an average Early Modem town, at least if we accept the then diminished speed of the data processing

16The capability of dBASE to produce and print comprehensive and selective reports supports our interest to document results (even graphically) and to generate the above mentioned hardcopies. The MEMOry fields of the program provide for footnotes. The data entry methods of dBASE support and secure fast, comfortable data fixing. The system is flexible; one is always allowed to add new fields wherever wished, and to erase others. The format of the dBASE files in the meantime is accepted by a wide range of other programs. Therefore, transfer of data and the transformation of data base information into text files are relatively easy.

17Nevertheless, the program also has its disadvantages. The handling of the query-and processing-procedures, for example, is quite old-fashioned, and the length of each field has to be fixed in advance, which means a considerable waste of space. Here recently developed database packages like PARADOX offer more advanced solutions10.

18Usually a project carried out on the computer includes four steps of work, namely data entry, data manipulation, analyses, and documentation In the foilwing paragraphs we will briefly describe the according approaches

19Data entry. Data entry in our project is carried out directly from the sources, or source reproductions, at a computer work station This data entry system maximizes speed and accuracy. It is also possible to program a variety of logical checks for internal consistency. If one tries to put numbers into alphanumeric fields, for example, the system will refuse to accept them. Consistent data verification is possible by comparing the text of hardcopies with sources.

20Data manipulation. This step serves the aim of generating a consistent data unit. It can be done either by hand or automatically, in any case starting with a checklist of all numeric fields carrying data Another duty to be performed here is the standardization of time data if this was not already done during data entry.

21More complicated is the problem of standardizing last names without distorting historical evidence. Here the most reliable solution lies in maintaining close contact to the sources. Luckily for us, real problems will occur here, as far as we can see, in only a few cases

22Then there is the chronological problem. We need at least a rough chronological orientation to distinguish persons and generations, especially to keep the data of the acquired connections clear. Unfortunately, due to the loss of some important church records, in most cases we do not have the year of birth. Instead, we know, for instance, the year of marriage or of admission to the Augsburg merchant’s chamber and can reconstruct supplementary values based on the averages of the cases we know

23The basic order of the records, however, is alphabetical, arranged with regard to the last names of our listed elite members. In establishing different arrangements by using SORT-commands and creating indices dBASE is specifically strong. We can arrange our records by choosing any field to become a sorting or indexing field. Moreover, we may let the machine re-order the records on multiple fields. This procedure is interesting, for instance, if we decide to compare family relations according to their alleged intensity, that is by distinction of inner and outer circles.

24An efficient way of sorting a database alphabetically, numerically or chronologically, is the use of index files, which let you view or list a database in a particular order without actually rearranging or copying the database records In other words, they are a kind of glasses to look through to your database showing a certain order of the data. Thus, indices can be updated automatically to accommodate new records or other data changes, and indices also allow one to quickly find records.

25Analyses. Having significant sets of data stored within the database at our disposal, we can begin to use certain inquiry procedures to analyze the contents of our data base. In technical terms, the inquiry procedures we employ produce selected information. Selection thereby means isolation of certain records, exclusion of certain values, and compression of records and values.

26DBASE provides for two ways to accomplish according tasks, in the program’s language called queues. The first technique is the "query-by-example" (QBE) approach, regrettably usable only for not too complicated operations.

27The user creates a certain query on the so-called Query Design Screen, which shows a visual model of the data and offers windows with all needed hints to design the query. In a first step, the user has to select the fields he wants displayed in the result. Then he simply has to enter a matching expression in a desired field to receive a list of possible expressions. For example, if we want to select only persons belonging to the kinship of the Fuggers we have to write into all fields containing kinship information "= Fugger" respectively "like *Fugger*". The "like *Fugger*" expression guarantees the inclusion of all cases where "Fugger" is only part of the names.

28The second way to accomplish queries lies in computer programming. Here dBASE also offers relatively good chances of success We are not forced to learn a sophisticated programming language. Instead, dBASE uses its own language or, for more complicated needs, offers the use of the Standard Query Language (SQL).

29Any computer program is simply a senes of instructions to a computer. These instructions are commands which cause the computer to perform specific tasks. With regard to the instructions which constitute the program itself, we have operators at our disposal. Commands, operators and functions shape so-called statements, i.e. tentatively autonomous parts of a programm.

30Documentation. In most cases, the results of our analysing procedures are needed in a form which enables us to include them in our descriptive text In other cases, we would like to have them displayed at our computer screen

31For both demands, dBASE offers some programming opportunities which can be realized comfortably by the use of the so-called Form Design Screen. In contrast to most dBase-fanatics, however, we prefer the transfer of all according information to our usual word processing program. Our working biographical dictionary, for instance, eventually got its form by the use of Microsoft WORD, though its contents are based on a dBASE-program.

Appendix 1