Introduction

Charred seeds are the most common source of analysable plant material in archaeological excavations for studying the origin and spread of cultivated plants (Zohary and Hopf 2000). The identification of charred, desiccated, or waterlogged wood, wild plant seeds, fruit pips, nut shells, and remains of cultivated plants has contributed substantially to our knowledge of human diet in ancient times, past subsistence strategies, and the history of plant domestication.

Cereals have been the principal crops of most civilizations. Their grains represent the main source of carbohydrates for mankind. Their nutritive value is generally high, and the seeds can be stored for long periods. Pulses, annual legumes cultivated for their seeds, accompany the cereals in most regions where cereal agriculture can be found. In traditional agricultural communities pulses served, and still serve, as a main meat substitute. Pulses seem to have started their role as companions of wheat and barley very early in the agricultural history of the Old World. The available archaeological evidence indicates that Pisum sativum (pea), Lens culinaris (lentil), Cicer arietinum (chickpea), Vicia ervilia (bitter vetch) and Lathyrus sativus (grass pea) were taken into cultivation more or less together with the principal cereals (Zohary and Hopf 2000).

Carbonization occurs upon exposure to high temperatures due to fire. Such heating, under a limited supply of oxygen, converts the plant’s organic compounds into charcoal. Since charcoal is not affected by bacteria, fungi, or other decomposing organisms, the carbonized plant remains survive in most environments.

Up to now, many carbonization experiments involving seeds and other plant components have been carried out. These experiments have been studied principally to identify the variables that affect the charring process, such as temperature and water content (Boardman and Jones 1990; Castelletti 1976; Costantini 1979; Gustafsson 2000; Hather 1993; Hopf 1955, 1967; Renfrew 1973; Wilson 1984).

Substantial information about the effects of heating the seeds of various plants has been gained experimentally by charring simulations in laboratory ovens. At fairly high temperatures, carbonization causes characteristic deformations in cereal grains. The intensity of the deformation depends, among other things, on the amount of water present in the seeds, the kind of heating, and the temperature reached.

Experiments were carried out to study the nature of distortion and size changes under various conditions. Stewart and Robertson (1971), for example, investigated the effects of moisture on degree and type of size change in Triticum and Hordeum species. They found less size change in dried grains (11% moisture) than in moist specimens (15% moisture). Renfrew (1973) conducted similar experiments using Triticum, Hordeum, Avena and Secale. Both studies were concerned primarily with size changes, since to compare charred archaeological grains with modern cultivated ones, the nature and magnitude of size changes caused by charring must be understood. Similar experiments have been carried out with Zea mays (maize) (Cutler and Blake 1976; Pearsall 1980a). The most comprehensive study on the effects of charring on seed size and shape was done by Wilson (1984). Rather than examining the effects of charring on individual seeds, Wilson investigated the effects of carbonization on a seed assemblage. She was interested primarily in what effect the loss of taxa during carbonization would have on the interpretation of seed assemblages. Her study contains much information about effects shown by seeds of varying chemical composition and moisture levels during carbonization, as well as charring temperature, and burning/heating duration and how the presence or absence of soil affects seed distortion and destruction. Wilson concludes, “It is not yet possible to deduce the original size of fossil carbonised seeds, because the heating environment cannot at present be deduced” (1984, 206).

The experiments presented in this paper have been carried out to reconstruct the carbonization conditions that may have occurred in a protohistoric house, different species of the families Poaceae, Fabaceae and Vitaceae are considered. Comparing experimental results with those obtained in the field allows a better interpretation of fossil assemblages.

Materials and methods

Hearth experiments

The method used by Gustafsson (2000) was changed a little. Hearth experiments were carried out on four 1×1 m surfaces with a depth of 30 cm. The base consisted of fine sand (hearths 1 and 2) or topsoil (hearths 3 and 4). In hearths 1 and 3, the seeds were mixed with the sand/soil, and in hearths 2 and 4, they were placed on top of the sand/soil (Fig. 1a). The surfaces were covered with twigs, branches of Quercus ilex L. (holm oak) and Castanea sativa L. (chestnut) wood (Fig. 1b) which were burnt. After burning the charred debris was floated in water to separate it from inorganic material.

Fig. 1
figure 1

a Hearths 1 and 3: seeds mixed with the sand/soil; hearths 2 and 4: seeds placed on top sand/soil; b the hearth surface covered with twigs, branches of Quercus ilex L. (oak) and Castanea sativa L. (chestnut) wood; c the reconstructed protohistoric house of the 8th century b.c.

Full size image

The experiment was repeated ten times to get a statistically significant data set. At the beginning of the experiment the total number of seeds for each species was 100.

The following species have been used: Hordeum vulgare L., Triticum aestivum L., Triticum turgidum conv. durum Desf., Avena sativa L., Secale cereale L., Cicer arietinum L., Vicia faba L., Lens culinaris Medik and Vitis vinifera L. subsp. vinifera. They have been eaten by the native populations of Italy since the Protohistoric Age (Costantini 1989; Costantini et al.1994, 1997). Because of their different anatomic and chemical compositions, they can give much information on the effects of charring on seeds.

House experiments

For this experiment, a protohistoric house of the 8th century b.c. has been reconstructed, a rectangular structure with a single floor of 4×2 m. The roof was made of hay from local native grasses. The walls were of Castanea sativa (chestnut) wood (Fig. 1c). The floor of the house was divided into eight equal-sized areas covered with sand/soil, each with a different content:

  • AREA A: Cicer arietinum, Vicia faba, Lens culinaris (Fabaceae, legumes) seeds mixed with the sand/soil, AREA B: seeds placed on top of the sand/soil.

  • AREA C: Vitis vinifera subsp. vinifera (Vitaceae, grape) pips mixed with the sand/soil, AREA D: seeds placed on top of the sand/soil.

  • AREA E: Hordeum vulgare, Triticum aestivum, Triticum turgidum conv. durum, Avena sativa, Secale cereale (Poaceae, cereals), caryopses mixed with the sand/soil, AREA F: caryopses placed on top of the sand/soil.

  • AREA G: Brassica oleracea, seeds mixed with the sand/soil, AREA H: seeds placed on top of the sand/soil.

  • The total number of seeds for each species used was 1,000. Fires were placed at two points of the floor: FIRE 1 between areas A, B, C, D and FIRE 2 between areas E, F, G, H. Both fires burned for two hours, reaching the maximum temperature of 750 °C.

  • Each area of the floor was further divided into two sections; one which had been intensely affected by the fire and the other one less affected (Fig. 2). After the house had burned down, the charred material was floated in water. The amount of carbonized material was 53.3% of the volume of the total amount of debris.

    Fig. 2
    figure 2

    The section of the floor of the house. Area 1 intensely fired; Area 2 less fired

    Full size image

Results

The hearth experiment

The greatest variations and losses of seeds were found in hearths 2 and 4 where the seeds were placed on the surface. The non-carbonised grains were lost. Fig. 3 shows that the greatest losses of seeds (75% on average) were of the carbonized cereal grains. These results agree with those by Boardman and Jones (1990). They recorded that barley was charred and destroyed at lower temperature than other grains, such as speltoid wheat.

Fig. 3
figure 3

The amounts of carbonized Poaceae seeds found in hearths 2 and 4 (average), repeated ten times

Full size image

The results of the carbonized legumes are shown in Fig. 4. As for cereal grains, the loss in hearths 2 and 4 is significant (55%). The same holds true for Vitis vinifera subsp. vinifera (60%, Fig. 5).

Fig. 4
figure 4

The amounts of carbonized Fabaceae seeds found in hearths 2 and 4 (average), repeated ten times

Full size image
Fig. 5
figure 5

The amounts of carbonized Vitis vinifera seeds found in hearths 2 and 4 (average), repeated ten times

Full size image

In hearths 1 and 3, the loss varied between 20% and 40% (Figs. 6, 7, 8). Hordeum vulgare more than 60%.

Fig. 6
figure 6

The amounts of carbonized Poaceae seeds found in hearths 1 and 3 (average), repeated ten times

Full size image
Fig. 7
figure 7

The amounts of carbonized Fabaceae seeds found in hearths 1and 3 (average), repeated ten times

Full size image
Fig. 8
figure 8

The amounts of carbonized Vitis vinifera seeds found in hearths 1and 3 (average), repeated ten times

Full size image

The total loss in all hearths considered was between 38% and 60%.

The differences depend partly on the sensitivity of specific seed types to heat, but many other factors are also involved, such as moisture content and weather, which are difficult to control and measure.

The depth at which the seeds lie below the surface is also an important factor, because the temperatures vary at different soil depths (Fig. 9), as shown by the results from hearths 1 and 3. This model of the temperature pattern at different soil depths during the burning was carried out at the Department of Mechanical Engineering to the Energetics, Faculty of Engineering of the University of Naples “FedericoII”.

Fig. 9
figure 9

Temperature pattern at different soil depths during the fire

Full size image

Figure 10 shows the different amounts of the carbonised seeds from Cicer arietinum, Avena sativa and Vitis vinifera subsp. vinifera according to the various depths below the surface.

Fig. 10
figure 10

The different amounts of carbonised seeds from Avena, Cicer and Vitis found in Hearths 1 and 3 in relation to the depths below the surface

Full size image

The house experiment

The house experiment showed a significant losses of seeds in areas where they were placed on top of the sand/soil and, in particular, in the areas intensely fired (Figs. 11, 12, 13, 14). As for the hearth experiment, the differences depended on the sensitivity of specific seed types to heat as well as on many other factors.

Fig. 11
figure 11

The amounts of carbonized Poaceae seeds found in the house experiment. Area E1 intensely fired; Area E2 less fired; Area F1 intensely fired; Area F2 less fired

Full size image
Fig. 12
figure 12

The amounts of carbonized Vitis vinifera seeds found in the house experiment. Area C1 intensely fired; Area C2 less fired; Area D1 intensely fired; Area D2 less fired

Full size image
Fig. 13
figure 13

The amounts of carbonized Fabaceae seeds found in the house experiment. Area A1 intensely fired; Area A2 less fired; Area B1 intensely fired; Area B2 less fired

Full size image
Fig. 14
figure 14

The amounts of carbonized Brassica oleracea seeds found in the house experiment. Area G1 intensely fired; Area G2 less fired; Area H1 intensely fired; Area H2 less fired

Full size image

To conclude, both in the hearth and house experiments the loss by burning is significant in all species considered, particularly if the seeds were placed on top of the sand/soil and the areas were intensively fired (in the house).

Discussion and conclusions

Comparing the laboratory results (hearth experiments) with those obtained in the field (house experiments) showed that the carbonization processes of different plant species and plant components can be affected by a variety of factors, such as moisture content, weather, temperature and fuel, duration of heating, even when the conditions in the two experiments were very similar. Additional experiments should therefore be done, in particular with factors such as moisture content, time and temperature better controlled.

The results of the experiments showed that there was only little variation in the degree of carbonization within the same family (Poaceae, Fabaceae, Vitaceae). In contrast, the variation of the loss of seeds between the different families considered is remarkable. These differences depend mainly on the sensitivity of specific seed types to heat. In fact, the cereals are very sensitive to high temperatures (between 200 and 400 °C), while the grape pips become completely carbonized at 450 °C and the pulses at 500 °C because the outer seed coat of the pulses is thick and very resistant to high temperatures. It is evident that these differences depend on the structural and anatomical characteristics of the seeds. In general, seeds with similar structure and contents seem to char to similar extents. However, to estimate the heat sensitivity of the different seeds, it is necessary to consider the size of the seeds. Of course, within the same family the smaller seeds are more sensitive to heating, for example lentils compared to chickpea and common vetch.

As for the final results of these experiments, it has to be taken into account that they could have been affected by excavation, sampling and flotation effects. These effects cannot be measured, and anyway the fragile seeds and cereal grains that would not have survived in an archaeological situation, probably also did not survive our experiments. The results are thus somewhat more comparable to protohistoric archaeobotanical material than those of many laboratory experiments.

In a recent paper (2000), Gustafsson has shown the results of two field experiments involving the carbonization of cereal grains and weed seeds—the proportions of charred grains of various types of cereals varied between each experiment. Taking the average number of charred cereal grains found in repeated experiments, it was found that similar quantities of each type of cereal grains were carbonized. These results can be used to interpret the archaeobotanical material found in prehistoric houses in Scandinavia, as these are carbonized and have been deposited throughout the period during which the site was occupied. The experiments also show that different types of seeds, from different cultivated plants and weeds, are carbonized to different extents. This depends upon the structure of the seeds.

In conclusion, while Gustafsson (2000) managed to recreate the carbonization conditions that may have occurred in a prehistoric house in Scandinavia to interpret the archaeobotanical material found there, both our hearth and house experiments provided important information about the carbonization process that may have occurred in a protohistoric house.