The Plan

The general idea is to explore. And what I will explore first is the urban hierarchy in Europe, taken at 50-year intervals and presented in "European Urbanization: 1500-1800" by Jan de Vries. Then I will take some ideas from "Linked" by Barabasi in his book about networks, in particular the idea of using a fitness function in a scale-free network, and use it to see what might pop out of the European data. Using the fitness value might be a good way to help understand the dynamics of the urban network.

We begin first discussing the table of urban places that de Vries uses in his book. It is a listing of 379 urban places that all attained a population of at least 10,000 during the time span 1500 to 1800. This data is duplicated in table 1 below, but as you will quickly notice, the table is incomplete. There are entries that have been marked as unknown (UNK) because de Vries could find no reliable source for that city at that particular time period. And another set marked 0 (zero) where there is also no reliable estimate but it was within a range of 1 to 9 thousand(s). In the tables the UNK are also shaded red and the second set is shaded yellow and in later tables where estimates are given for these entries the same color shading will be used to identify them.

de Vries also created another table that he used for much of his analysis that consisted of the cities listed in population categories or classes, with the first class or category being those cities with populations from 10,000 to 19,000 and five additional categories with the population doubling in each larger category. Since Table 1 only gives population in thousands, Class One is from 10 to 19, Class Two is from 20 to 39, Class Three is from 40 to 79, Class Four is from 80 to 159, Class Five is from 160 to 319, and Class Six is all cities with populations over 320. But even here we are faced with the problem of missing data.

How did de Vries deal with the problem of the unknown data? In his book he writes that even for the cities for which he could find reliable estimates, the dates did not always fall within the bounds he used - every 50 years plus or minus 10 years. If he had two estimates that fell outside this window then he used straight line interpolation to come up with a value. But for some cities even this was not possible. So his solution was to assign a population category that seemed most likely given the history of the place and any good estimates during the entire period. To then finally assign a number for the population, he takes the average for all cities in that category and uses that as the value. (fn: pp. 25-26)

My first goal is to produce a new version of Table 1 that has entries for those marked as unknown. Some of the other tables in de Vries' book give population totals that include those he gave to the unknown values, providing me with a way to check how closely the values that I come up with match those he came up with. It will be an opportunity to try and repeat his work and recreate the population estimates for all of the entries in his database, a table of information that was not included in the book.

Animated GIF 1
fit 1 fit 1 fit 1 fit 1 fit 1 fit 1