Combining Written and Tree-Ring Evidence to Trace Past Food Crises: A Case Study from Finland

Abstract

The lack of written source material on population and food availability has hindered studies on medieval and early modern food crises in many parts of the world. Examining the case of sixteenth and seventeenth century Finland, this article explores how indirect evidence—so called proxy data—could be used to identify past food crises. The proxies of past climate, grain harvest, storage capacity and population variability were derived from tree-ring studies and early administrative accounts. Evidence from “natural” and written archives supplemented each other. The applicability and limitations of using proxy data to trace past food crises is further discussed by comparing the examples of the sixteenth and seventeenth century to the better documented famine period of the 1860s. It was found that tree-ring data and early administrative accounts provides valuable material to identify past food crises.

Appendix

Annual registers for tithes and deserted farms in Finland can be found in bailiff’s accounts up until 1634 and provincial accounts from 1635 onwards (National Archives of Finland, NAF). The accounts hold, inter alia, annual tax, land and tithe registers. The smallest unit in the accounts is the individual farm, which is grouped by villages and further by parishes. Some data were missing from the accounts, which are detailed in Table 3.2. With the exception of the deserted farm data for 1689–1698, which was derived from literature (Mäntylä 1988), the data for tithe and deserted farms time series was collected from the primary sources.

Table 3.2 Missing data by harvest year

The parish tithes in Ostrobothnia were recorded by a measure of volume, which was commonly a barrel (146.35 L). Registers that marked the tithes in another measure of volume (mainly the 16th century registers), have been converted into barrels. In addition to the missing annual registers (Table 3.2), some sporadic parish accounts were missing from the registers over the period from 1590 to 1619, without any explanation given. The missing records might have resulted from crop failure, changes in local administrations (parishes could have been divided or joined together) or simply the result of a clerk’s mistake. These matters pose a challenge for calculating the tithe time series. The parish series could not be summed up due to missing records. Moreover, the series could not be averaged due to the varying numbers and sizes of the parishes.¹⁰ Thus, the annual tithe time series ($T_{i}$) was calculated as the weighted sum of the paid tithes (in barrels B) per parish ($B_{P}$) as:

$$T_{i} = \sum\limits_{P \in Parish} {B_{P} \cdot w_{P} } ,$$

(3.1)

where the weights ($w_{P}$) are the ratio between the parish tithes and all the recorded paid tithes in this year, $w_{P} = {{B_{P} } \mathord{\left/ {\vphantom {{B_{P} } {\sum\nolimits_{P} {B_{P} } }}} \right. \kern-0pt} {\sum\nolimits_{P} {B_{P} } }}$. To enable interproxy comparisons and coarse comparison between the different study periods, the tithe time series ($T_{i}$) was standardized (z-scores) as:

$$z_{i} = {{(T_{i} - \mu )} \mathord{\left/ {\vphantom {{(T_{i} - \mu )} \sigma }} \right. \kern-0pt} \sigma },$$

(3.2)

where $\mu$ is the sample mean of the tithe time series and $\sigma$ is the sample standard deviation of the series during the study period.

The recorded units in the deserted farm registers varied over the 16th and 17th centuries. On some years the number of the deserted farms were given as mantals (a taxation unit), on some years as number of peasant farms. The relation between mantal and peasant farm varied over time. In the mid-16th century one mantal more or less equaled one peasant farm, but in the early 17th century ½ and ¼ mantal farms were common in the land registers (Seppälä 2009, 63–68). Thus, to avoid bias due to changes in the recorded units, the deserted farm ratio for year i ($F_{i}$) between the number of deserted farms ($F_{D}$) and the total number of farms ($F_{T}$) of the same unit (mantal or peasant farm) was determined as

$$F_{i} = {{F_{D} } \mathord{\left/ {\vphantom {{F_{D} } {F_{T} }}} \right. \kern-0pt} {F_{T} }}.$$

(3.3)

To enable visual comparison between the deserted farm time series and population change data (see below), the deserted farm time series are presented in inhabited farm ratios in the illustrations (see Fig. 3.4). The ratio of inhabited farms for a year i is given by $1 - F_{i}$. The farm ratios are given in percentages.

Provincial population data form Ostrobothnia for the period 1689–1698 was derived from literature (Muroma 1991) and national data for the period 1869–1870 from the Official Statistics of Finland database. From these time series, an annual series of the natural population change (the difference between the number of live births and deaths) was calculated.

Provincial data on seed and harvest for both grains—rye and barley—were derived from the official statistical reports (Suomen virallinen tilasto 1868, 1875). The yield ratio time series were calculated as a weighted average of rye and barley yield ratios for each year as

$$Y_{i} = w_{R} \cdot Y_{R} + w_{B} \cdot Y_{B} ,$$

(3.4)

where the weights are the fractions of harvested rye and barley, respectively. The grain specific yield ratios $Y_{R}$ and $Y_{B}$ were themselves calculated as weighted averages of the provincial yield ratios (the weights being the fractions of harvested rye and barley from a province compared to the total rye or barley harvest from all provinces). To enable comparison with the tithe series, yield ratio series were transformed into z-scores (cf. Eq. 3.2).

The data on Swedish nominal grain prices and Helsinki rye prices was derived from literature (Edvinsson and Söderberg 2010; Vattula 1983, respectively). The adjustment procedures of these series are presented in the original publications. Also the price series were transformed into z-scores (cf. Eq. 3.2).

The statistics for import and export for the period from 1869 to 1870 were derived from literature (Vattula 1983). Here, the net import, the difference between import and export, was calculated. The annual net import is given in tons.

And last, the tree-ring (MXD) data originated from Lapland and northern Finland (Matskovsky and Helama 2014) and southern Finland (Helama et al. 2014). For the purpose of indicating annual temperature and yield ratio fluctuations over a larger part of Finland, the two time series were combined by normalizing the records into z-scores (cf. Eq. 3.2) and by averaging the normalized data into a mean MXD record.