NOTES

All the maps used in this work had been drawn on the basis of a digital chart in the Laboratorio informatico di Archeologia medιevale of the Dipartimento di Archeologia dell'Universitd di Siena within the project ‘Demografia, economia e societa nella Toscana dell'Ottocento'.

All over Tuscany the small land property was around ’0%, coming in the first position in some communities of the mountainous areas (Detti and Pazzagli 2000).

The results are based on a detailed analysis of the nominative data of the ’84’ census, and from their integration with data from a fiscal source.

What is considered here is a price variation, not its level. Therefore the Florence time series is perfectly compatible with the aim of this study.

‘1 day-’ year', ‘2-5 years', ‘6-10 years', ‘’’-20 years', ‘21-30 years',..., ‘81-90 years', ‘100-years', and finally ‘age unknown'. The age groups, sufficiently detailed from the first synthesis picture (year ’8’8), are even more abundant from the year ’855: ‘from day ’ to 6 months', ‘from 6 to ’2 months', and then in yearly groups until year four; finally, in five-year groups (5-’0, ’0-15,., 95-99, ’00-years, age unknown).

An annual price is published based on an average of monthly wheat prices (Bandettini ’957).

’.

2.

3.

4.

5.

6.

7.

8.

9.

10.

11.

12.

13.

For a recent review of this method and results, see Bengtsson and Reher (1998).

Alternative de-trending techniques are, of course, possible. We considered two other possibilities for differentiating the natural logarithms of the original data and calculating ratios of observed values to a eleven-year centred moving average. However, we opted for least squares regression since the first method is not appropriate for our data (we analyse a thirty-year period) and in the second case we overly reduce the number of observations.

This last borough presents a low absolute value for the number of inhabitants in such a large territory, but they are all concentrated in a single small area. Moreover, the borough's extension formed a separate territorial unit, which could not be assimilated by any of the neighbouring agrarian areas.

Prices show relatively low fluctuations compared, for example, to the Napoleonic years and the crisis of 1816—17 when they increased by about 300—400%. However, during the poor harvests from 1845 to 1847, wheat prices rose by about 40% in the Florence market. The magnitude of this increase does not reflect the scarcity of local wheat production. The government, also under the pressure of popular demonstrations, abolished all duty on grains and promoted special wheat imports (Bandettini 1960: 17—18). A less favourable series of harvests commenced towards the end of the analysed period, with price increases of around 30—40%.

The estimates of the model were obtained using the Xtregar procedure in the STATA7 package. This procedure estimates cross-sectional time-series regression models when the disturbance term is first-order regressive. Xtregar offers the Baltagi-Wu GLS estimator of random effects.

The values of the coefficients are, of course, different from those of Table 13.3. In order to avoid redundancy and not to mislead, we have decided not to tabulate the results obtained with the Xtregar procedure.

In reality, the decennial statistics on deaths refer to the resident population, and therefore ought to contain those deaths of residents, which occurred outside the community. This is virtually ‘certain’ at least up until 1847. Childhood deaths were, however, ‘elusive’ (also due to their elevated numbers) over the course of the complex ‘counting’ operation conducted in the central offices of the Tuscan Civic State (Del Panta 1985; Breschi and Del Panta 1993).

¹⁴ The maps drawn using the coefficient values result are very similar to the ones presented. This is also the reason why we preferred to give more emphasis to the degree of statistical significance rather than to the size of the coefficients (elasticities).

^15. These results are the outcome of new and preliminary analysis we carried out after having inserted two new data sets into the demographic database. One of these deals with house ownership and the other with the level of taxes paid by each household. These initial observations would appear to indicate that a higher degree of fragility in adults and the elderly could be linked in particular to household typologies and to the ‘wealth’ variable.