We have arrived to the end of the story about demography in Venice, through which we gained great new knowledge about historical events, and also we saw some interesting consequences on population based on these historical events. This project we valued as a great opportunity to observe socio-geographical dependencies and to try to reconstruct population dynamics between 1300 and 1797.
We should introduce the term of population dynamics and to explain which part of it we have covered. Population dynamics is the branch of life and humanities sciences that studies short-term and long-term changes in the size and age composition of populations, and the biological and environmental processes influencing those changes. Population dynamics deals with the way populations are affected by birth and death rates, and by immigration and emigration, and studies topics such as ageing populations or population decline. In our works we didn’t have access to any information about age composition, birth ande death rates (except for plagues), immigration and emigration, thus we could only change parameter which is overall number of population of Venice, and the population was function depended on social-influence level of historical events.
Methods and Results
At the very beginning we had a big obstacle to find the data about the population in Venice for period before 1871. After deep investigation and searching for any data before mentioned year, we managed to collect population censuses for different years without the same time-shift between them, e.g. for some there are time-shifts of 10 years and for some around 50 years.
Our resources were from different books and encyclopedias written in Italian, English, Serbian and Croatian. We spent few weeks to find the books and to go through them and extract the information.
We found our data from one resource (Rosina & Rossi, 2000), but we wanted to explore other resources to be sure we had relevant data [1-4]. In some of the sources we found information for many years, the same as from initial source (Rosina & Rossi, 2000), and in the rest we managed to find only for specific years and not rounded numbers as it was a case for initial source.
After data collection, we wanted to apply our data to proposed known models, but for results we got strange representation which cannot be manipulated easily to reconstruct the population dynamics. We tested models such as normal growth model, which is non-appliable at all, since it’s only increasing trade model. Next one was coalition model, which is impossible to modify since it behaves as ‘faster than exponential’ growth model.
Couttsian growth model for which we tought that would be the most realistic one we couldn’t even to apply it, because it asked us for replication and immigration rate which we couldn’t find for our data.
With Nicholson-Bailey model, right from the start it was difficult to try to adaptate the formula. First we tested the model with one range to see how it works, then we noticed that this model’s plots are only linear and then we cannot apply to our data. Here we have two presentations, just to show how it would look like with situation of increment and decrement of population (Fig. 1).
Lotka-Volterra model was difficult for applying, and when we tried to test only for first period, for first plague year, it didn’t work, since value was on the 0, and then it started to go in negative values (Fig. 2).
For Nurgaliev’s law we did adaptation, but it wasn’t good since we didn’t have information about changing rate of male and female and death rate, so we had to maintain the same factor and that was why the representation was not good (Fig. 3).
After realising the fact about proposed models, we decided to reconstruct population dynamics by manual weighting. First, we wanted to do polynomial fitting of existing data and to extract the function and manipulate the function (Fig. 4A). Since we had not so dense data points on large time span, fitting was badly represented, completely unrealistically. The similar thing we also tried with doing a spline method to our data and that representation was a way better than fitting, but still pretty inaccurate (Fig. 4B).
Then we implemented a weighting method, which first we did with no accurate weights for specified years to see only if it was working. Later on, we saw that there is nice, but quite unrealistic representation since we did ‘blind’ testing with badly assigned weights. Years for we had information about population were these: 1300, 1450, 1509, 1540, 1563, 1581, 1586, 1624, 1633, 1642, 1696, 1760, 1780, 1790, 1797. Then we were exploring and looking for years between these known data, e.g. between we know that between 1300 and 1450 there was a plague period, some wars and other events which were possibly affecting the population dynamics. When we found all these years we was reasoning what affecting these events were and based on that we were assigning the weights to initial representation. For few plague cases we had information that population was decreased for half or for one third, so here weighting was assigned by itself, while for other cases we had to take into account if there were wars, fires, also other events that were positively influencing on population by making new selling routes and so on and so forth.
Assigned score for wars were different based on where was the battle field, how much it lasted and whether Venice lost or gained some new territories. Also those territories were significant for population dynamics if they were close to Venice or if they were important as cruising ports. For some periods we didn’t do anything for reconstruction, because during the exploring the events there were not any specific events that had an impact on population. Those periods were or to short (5 years of time-shift) or only thing that was happening was changing Doges (rulers of republic) on regular basis.
We were doing the reconstruction until the year of 1797, since Republic of Venice was conquered in that year and after Venice was annexed to Habsburg monarchy. There is a huge gap about population data between 1797 and 1871, since we didn’t manage to find any information about. For this period we wanted to the same technique of population prediction, but doing kind of extrapolation was impossible, because of sufficiency of data. So this part of the work had to be dropped.
At the end, after reconstruction we did again fitting (Fig. 5) and spline(Fig. 6) methods with purpose of getting better representation of population dynamics.
Although, this project has strong limitations based on having population data only for specific years between 1300 and 1797, reconstruction of population was successfully done. Obtaining more parameters such as number of male and female persons, age pyramid, born and death rates would give with further research better reconstruction of population in Venice and also would be able to conclude more about sociocultural phenomena in this period of almost 500 years.It would be better Since we had only raw numbers of population for specific year we were enforced to do population dynamics only on strict population number information.
We would like to thank to the whole organisation of the Digital Humanities course for providing us idea of working in hands-on project full of multidisciplinarity in which we had to involve our different skills and to learn new ones. One person deserved more than words “Thank you”, since high-school professor and Master of Humanities Science Eva Tatic helped us for data collection. At the end, we would like to thank professor Frédéric Kaplan for being our project leader and person who taught us a lot about new, bright and useful field,the Digital Humanities.
 Bek, K. (1998). Istorija Venecije. Zemun: Plato.
 IX knjiga Povijesti. (2008). Zagreb: Biblioteka Jutarnjeg lista.
 Lane, F. C. (2007). Povijest Mletacke Republike. Zagreb.
 Linderman, M. (1999). Medicine and Society in Early Modern Europe. Cambridge: Cambridge University Press.
 Rosina, A., & Rossi, F. (2000). La popolazione di Venezia, 1633-1797: una ricostruzione delle dinamiche evolutive. Padova.