The next stage of analysis involves extracting starch granules from the sample so that they can be analysed microscopically. This process can be as simple as removing a small amount of sample from the specimen and applying it directly to a slide. Often, however, more complex procedures are required to break down the sample matrix and then separate and concentrate the starch.

Starch extraction methods are still being developed and often need to be adapted to suit the type of material under analysis and individual research goals. The main issue to be aware of is that starch granules are much more fragile than other plant microremains such as pollen or phytoliths, therefore aggressive extraction techniques should be avoided. For example, starch granules can be damaged by heat above about 35°C (particularly in the presence of moisture), strong acids and oxidizers, and excessive mechanical processing such as grinding. It is therefore useful to test extraction techniques first on reference materials (e.g. Torrence and Therin 2006). If a number of microremains (e.g. phytoliths, pollen) are to be recovered from the same sample, then a sequential protocol that accommodates each residue type should be designed. Starch is usually extracted first in these cases because it is most easily damaged (see Coil et al. 2003 and Korstanje 2003). Below are some examples of starch recovery methods used for different types of archaeological materials.

Sediments

The most common method for extracting starch from sediments and other bulk samples such as coprolites involves heavy-liquid flotation. First, the sediment is disaggregated and dispersed, and unwanted sediment components such as clays, carbonates and organics are removed, either by chemical dissolution or physical separation. Starch granules are then separated from the sediment matrix by floating them in a heavy-density liquid prepared at a specific gravity (s.g.) higher than that of starch (~1.5) but lower than other inorganic particles (generally > 2.0). Sodium polytungstate is the most commonly used heavy-liquid, as it is non-toxic, does not damage starch and can be recycled (Torrence and Therin 2006). Extractions generally involve the following steps and an example of a full protocol can be seen below.

Artefacts

Samples for starch analysis are removed from artefacts in a step-wise procedure designed to remove granules attached directly to the artefact surface and those present in adhering sediment in separate stages. Qualitative and quantitative comparisons between these samples can be used to assess whether starches derive directly from tool use or from the surrounding sediment. The general process is as follows (see Barton et al. 1998; Pearsall et al. 2004; Perry 2004; Perry et al. 2006; Piperno 2006:98; Zarillo and Kooyman 2006):

1. Dry brushing: Gently brush the artefact with a clean brush to remove loosely adhered sediment, which can then be combined with any sediment that has detached in the sample bag. This sample represents the depositional matrix immediately surrounding the artefact, which is the most likely source of artefact contamination (Barton et al. 1998). Starch can be separated from this sample using the heavy-liquid flotation method described above. Sediment collected from around the artefact during excavation serves as an additional control and should be processed in the same manner.
2. Spot sampling: Next, remove residue samples from various locations on the used and non-used areas of the tool with a small volume of water (10-20 µl) applied with a pipette. The aqueous extract can be mounted directly on a clean microscope slide. Particular attention should be paid to pits and cracks where granules could have been embedded during use. Alternatively, granules trapped in these voids can be removed using a needle probe (Piperno et al. 2004). It can be useful to examine the artefact under a dissecting microscope during this process to locate possible starchy deposits. Don’t forget to record the sample locations on a drawing or photograph of the artefact. Use-wear and residue spatial patterns should also be recorded as these can reveal important functional and taphonomic clues (see Fullagar 2006a, 2006b for more information).
3. Wet brushing: Remove all remaining sediment from the used and non-unused areas of the artefact separately by scrubbing with a clean brush and cool, distilled water. Collect the extracts and concentrate by centrifugation. Separate starch further using heavy-liquid flotation if necessary.
4. Sonication: Finally, remove well-adhered starches (particularly those trapped in micro-pits or -crevices) by sonicating the artefact for 5-10 minutes. Concentrate the extract and separate the starch with heavy liquid if necessary.

Charred samples

Extracting starch from charred residues on pottery or other carbonised food remains usually requires chemical treatment to break down the charred matrix and liberate the entrapped starch granules. Hydrogen peroxide (H2O2), sodium hydroxide (NaOH), sodium hypochlorite (bleach, NaOCl) and heated nitric acid (HNO3) have all been used in the past (Boyd et al. 2006, 2008; Cortella and Pochettino 1994; Crowther 2009; Scott-Cummings 2006; Zarillo et al. 2008), but the effects of some of these reagents on starch have not yet been thoroughly tested and some may be damaging (e.g. Cortella et al. 2001; see also Crowther 2009). The extraction procedure illustrated below is recommended (Crowther 2009).

It can also be useful to examine microstructural features of charred or desiccated food remains with scanning electron microscopy or in thin section to establish whether characteristic alterations induced by different processing methods (e.g. wet or dry cooking, fermentation) are visible. Methods for these types of analyses are described by Samuel (1996; 2006) and Valamoti et al. (in press).

Avoiding contamination in the laboratory

It is important that archaeological samples don't become contaminated with modern (and ancient) starch during the extraction process. Here are some simple controls that should be adopted:

• Monitor and control airborne contaminants as much as possible. Place contamination slides around the laboratory and check them regularly for background starch. Filter the laboratory air supply if possible.
• Wash hands thoroughly before beginning work and don't consume food or drink in the laboratory, which contravenes most safety protocols and is also a potential source of starch contamination.
• Wear non-powdered gloves. The powder in gloves is manufactured from starch, which poses a major contamination risk.
• Use distilled water at all times and prepare fresh reagents as often as possible. Work with small aliquots of reagents so that if contamination does occur, it will not be widespread. If reagents such as heavy liquids are recycled, they should be filtered first at < 1 µm.
• Use sterile, disposable equipment where possible (e.g. pipette tips, weigh trays, centrifuge tubes).
• Clean re-useable laboratory materials as soon as possible after use by scrubbing with detergent. In addition to mechanical cleaning, it is good practice also to wash materials with a starch-destroying chemical (e.g. 5% potassium- or sodium hydroxide, 10% bleach), by submerged in boiling water, or by sonication. Rinse thoroughly with distilled water after cleaning to remove any remaining residue.
• Prepare reference materials in a separate location to archaeological samples and use separate sets of reagents and processing equipment (e.g. mortars and pestles) to reduce the potential for cross-contamination.