Preserved vegetative storage parenchyma can be extracted from archaeological contexts using established archaeobotanical methods for collection of macrobotanical remains. These usually involve flotation, or wet or dry-sieving of excavated material, alongside collection of material in situ (Hather 1994a; Pearsall 2010; Thompson 1994:14). Sieving involves the manual agitation of material to loosen sedimentary aggregates and allow smaller particles to fall through the various sized meshes, retaining a portion that contains botanical material (Hageman and Goldstein 2009:2846).This separation is more effective and less biased than collection of botanical material by eye during excavation. Dry-sieving is not always effective when processing clay or silty sedimentary matrices, and so wet-sieving or flotation is more commonly used in these circumstances. Wet-sieving utilises water to further breakdown tough aggregates or sediments that have high clay content. Limiting factors affecting the effectiveness of dry or wet-sieving techniques are the size of the mesh used which can bias the types of remains collected, and the amount of force used to push material through the mesh can also damage or fragment preserved botanical remains (Pearsall 2010:13).
Water flotation utilises differences in density of organic and inorganic material to separate organic material such as charred botanical remains from the soil matrix. When carried out on a large scale, this technique can separate much higher quantities and range of botanical material than sieving alone (Pearsall 2010). Mechanical bulk flotation devices can be employed to process large amounts of sediment where water is freely available or where equipment is available for water recycling (Hageman and Goldstein 2009; Pearsall 2010); however, these
53 machines can increase the risk of charcoal fragmentation. These systems use flowing water to break up sediments and release any charred botanical remains that have a specific gravity or are less dense than water to float to the surface. The remains are caught in sieves with varying mesh diameters to create flot fractions of different sizes (Pearsall 2010:50-52). Mechanical systems are designed to limit the need for extra personnel to operate them, but must be overseen for maintenance and to collect samples (Hageman and Goldstein 2009:2847-8). Bucket flotation is often used where equipment and access to water is limited (Fairburn 2005b; Mason et al.1994; Oliveira 2008; Paz 2001). These tub or bucket systems involve manual agitation of sediments, with muslin cloth pegged into a bucket replacing sieves to catch botanical remains that are decanted from the surface of the diluted sediment. Heavy residue remains in the bucket, which is then either discarded or wet-sieved, to collect any remaining botanical or artefactual material. Comparisons of the flotation techniques have demonstrated the effectiveness of the mechanical systems (40-100%) over the bucket systems (6-100%) in terms of recovery rates (Wright 2005).
It has been argued by some archaeobotanists that dry-sieving is the most appropriate technique for the separation of vegetative storage parenchyma due to the fragility of these soft tissues when preserved in charred, desiccated or mineralised forms (Hather 2000:74). Other studies have demonstrated that the utilisation of both sieving and flotation can result in the recovery of increased quantities and also greater taxonomic diversity (Fairburn 2005b; Hageman and Goldstein 2009). Both techniques have advantages in the separation of particular types of botanical remains and specific taxa. Flotation enables the collection of material with low specific gravities such as many seeds, while sieving can facilitate the collection of endocarp and wood that is often too dense to be recovered through flotation. A combined approach is essential when consideration is given to the fact that that different environmental conditions during carbonisation can cause some botanical remains to vary in density, and thus affect the odds of recovery through flotation (Wright 2005:24).
The selection of a sampling strategy can also influence the recovery of botanical remains using techniques highlighted here. It is most often impractical to process all excavated material from within large-scale projects for all size-classes of preserved botanical remains, and an appropriate sampling regime must be chosen that enables the particular research questions to be answered. Lennstrom and Hastorf (1995) caution against taking the ‘feature bias’ approach that targets only features or areas of concentrated carbon, as it is just as relevant to discover areas that do not contain botanical material. The most common macrobotanical sampling techniques are the ‘blanket sampling’ strategy, whereby a sample is taken from every excavated context and feature (Fairburn 2005b:10; Pearsall 2010:66-68). These include ‘scatter’ or composite sampling where small amounts of matrix are gathered throughout a context and combined in the same collection bag, where samples derive from a column or sequence of deposits, and point bulk sampling of precisely located areas within a context (Pearsall 2010;
54 Wright 2005:20). Comparison of these techniques indicate that composite samples tend to recover higher quantities and diversity of charred material, but had a smaller range of variability within the density of material compared to point bulk sampling. Pearsall (2001:71) argues that this indicates that scatter samples captures a greater range of actions and places within an occupation level, but may also be a biased approach to sampling.
Sampling strategies involve the collection of a small amount of material considered representative of a particular context, but the way and form in which the sample is measured can also vary, and influence archaeobotanical interpretation. Sample size is usually measured by weight or volume, although these can both be influenced by the amount of moisture within the sedimentary matrix (Wright 2005:20). Even very small amounts of water can increase the weight of soil, for example 1ml of water at 4ºC weighs 1g. Volume is often measured using a calibrated bucket, but wet matrix can be harder to pack into these buckets and thus more difficult to accurately measure. Wright (2005) demonstrated this difference in volume by also comparing the types of sediment being measured when wet and dry or partially dried. Clay sediments had an average increase in volume of 25%, while silty sediment only increased 9% and sand was almost exactly the same, with an increase of only 1%. These results indicate that consistency is required to ensure that sample sizes can be considered representative and are comparable (Lennstrom and Hastorf 1995:705). Despite this, repeated wetting and drying of samples is not recommended, as this can weaken the cellular structure of charred remains and increase rates of fragmentation (Greenlee 1992; Hather 2000; Paz 2005).