Sources
To set up a wealthy ‘Tuscany of the river' region against a poor ‘other Tuscany' region is a schematic device, but one that helps with understanding the complexity and importance of the differences that become broader as we look deeper into the matter.
The simplification proposed above is due to the lack of effective quantitative indicators of the wealth of small population segments or individuals. It is in fact the fringe segments—including those without income (at the time called the‘necessary’ destitutes) as well as some categories with a fixed income, such as farm labourers, casual workers, wage earners, and so on—that are more sensitive to recession and react, from a demographic standpoint, with greater evidence.
Even without a more substantial quantitative correlation to describe the population according to their wealth, we can assert that the tight bond that linked together the rent, the land usage, and the typology of the land workers allows for a division of Tuscany into different territories. Each one of these has a certain explanatory potential, especially with regard to variables that highlight the socio-economic traits of its inhabitants. Therefore, given the characteristics inherent to the separate territories, an approach that takes them as the starting point for the analysis should permit greater insight into the link between price fluctuation and mortality.
This approach fits Tuscany well thanks to the numerous and detailed sources of information available on cereal prices, mortality rates, and the features and usage of the land prevailing in each of the agrarian regions. In order to comprehend this aspect of the research, and better understand the work done so far, a short description of the data is essential.
As far as prices are concerned, only wheat, being the most common cereal, has been considered.
The data relate to the Florence market, which can be considered representative of the whole of Tuscany for the following reasons. From the eighteenth century onwards, the cereal market in Tuscany was one of the most integrated of all Europe (Persson 1999). Already in 1768, in fact, the Grand Duchy had enacted a legislation inspired by the principles of free exchange for the commerce of cereals, a first in Europe (Mirri 1972). Also, of all the regions in the peninsula, Tuscany had the best system of transportation and the most extensive road network (Mori 1986). Finally, when comparing the Florence time series to others relating to the Grand Duchy, where possible, simultaneous price movements can be detected, although the former are slightly higher.4With regard to the information on demography, we relied on documentation which is in some ways unique, kept by the State Archives in Florence; namely, the decennial statistics. These statistics cover the period from the years immediately after the French Restoration in the Grand Duchy government (1815) until the formation of the new Italian State (1861). They recorded the total number of marriages, births differentiated by sex and legitimacy, deaths according to sex, and the total number of deaths (without distinction by sex) according to age group, for each community of the Grand Duchy.5 Despite some limitations, the series are detailed and reliable (Breschi 1990).
The temporal unit used to rebuild both the demographic and economic time series is based on the solar year, since the demographic data were only available on an annual basis. We adapted the wheat prices6 which could instead be gathered every month, to these data. This procedure does not alter the direction of the various price fluctuations; in fact it makes the price series less volatile, since the most striking variations are in proximity of the harvest—immediately before or after—between June and July. This discontinuity is alleviated in our series because it is distributed over two years instead of one.
The statistical model used for the short- and medium-term relationship between the economic and demographic time series is based on a distributed lag model (DLM).7 In order to apply this statistical method some comments are necessary.
The first-order autoregressive process provides the most parsimonious description of the data. The estimation technique, taking into account the problem of autocorrelation, is based on the iterative maximum likelihood method (Harvey 1981: 191—5). With regard to the lagged independent variables, we include values for lags of 0—2 years in the regressions of total deaths, early childhood mortality (q1J, and age-classed deaths, 6-20, 20-60, and beyond 60 years. For the analysis of infant mortality (q0), we instead used two lags (lag 0 and 1). Finally, to obtain de-trended values for all demographic and economic series used, we applied a least-squares regression, a simple technique that provides a good fit to the data without loss of observations. We then calculated the ratio between observed and predicted values. These became the transformed values for all the economic and demographic series.8A further clarification is needed regarding the scale of our investigation. In recent years, efforts have been made to overcome some initial limitations due to an overly aggregated approach. Almost all investigations of European preindustrial societies looked at national entities (England, France, and Switzerland) or large regional areas (almost always corresponding to administrative entities). More recently, attention has moved towards the consideration of populations with specific socio-economic traits, or those characterized by clear social differences (Reher 1990; Galloway 1993). In this work, although the analysis centres on a whole state body, the scale of the investigation has been adapted to the minimum territorial unit allowed by the method employed.
The pattern of land allotment within the Grand Duchy in the first half of the nineteenth century is well known. Many studies on the subject have been carried out for the years 1832-5 and in particular on the land cadastre (Biagioli 1975). For the purpose of this work, the partitions made by previous authors were not judged to be entirely satisfactory.
We were therefore compelled to develop a new set on the basis of uniformity criteria, taking into account an outline of already-existing area divisions in the region (Zuccagni-Orlandini 1832; Biagioli 1975; Pazzagli 1979).Consequently, although we have data available for single communities, we decided to use a slightly larger scale. We have divided the Grand Duchy of Tuscany into agrarian areas, each representing a territory as uniform as possible and at the same time distinct from the neighbouring areas. We have also taken into account that in each area the number of inhabitants should satisfy the minimum prerequisites requested by the econometric method adopted. To summarize, the following are the uniformity criteria taken into account: geo-morphological characteristics, predominant cultivation methods, type of agricultural management, and number of inhabitants.
To avoid distorted results, we decided to exclude from the agrarian regions those communities with a town population that was either too numerous or which clearly exceeded the rural population. Thus the towns of Florence, Pistoia, and Siena were excluded, since their borough territories did not go beyond the town-walls, and so
were Pisa, Livorno, Prato, Arezzo, and Grosseto.9 In conclusion, after this process of aggregation and exclusion, we considered forty-three agrarian regions and eight ‘urban’ areas.
4.