Both wood charcoals and charred seeds and fruits may be found concentrated or dispersed in the sediment, and different sampling and recovery procedures are needed to deal with these different types of deposits. Fragments of roots and tubers are most often dispersed. When deciding which contexts to sample, priority should be given to unmixed deposits of known archaeological phase and context type, so far as this can be established at the time of sampling. In a well-planned excavation, all of these deposits should be sampled for plant macroremains. A policy of maximum on-site recovery (excavation is a destructive process) followed by maximum off-site subsampling (a repeatable exercise) offers the best compromise between the recovery of adequate material for study and minimising the time taken for processing. Hand picking of dispersed wood charcoals and charred seeds directly from the excavated deposits should avoided as the material so collected is likely to be unrepresentative of the deposit as a whole and the results therefore meaningless, for either economic or ecological interpretation. Similarly, sampling only deposits where wood charcoals or seeds/fruits are visible in situ is likely to bias the sampling towards concentrated deposits, under-representing the more dispersed remains.

Storage room at Assiros Toumba, Greece.
Each storage bin was treated as a different context and sampled at multiple levels whilst being excavated.
The floor between storage contexts was also sampled separately and at multiple locoations.

Small deposits, often deposited over a short period of time (short-term deposits), should be sampled in their entirety. Timber wood charcoal is a specific case of this and requires special attention, which is discussed in the recovery section. Larger deposits, often deposited over a longer period of time (long-term deposits), require selective sampling. For example, distinguishable layers in a large pit, midden or occupation deposit should be sampled separately, or at different depths if no visible layers are observable. Similarly, large floors should be sampled in several places. Destruction deposits and large storage contexts require special attention, and it is often best to take numerous small samples rather than fewer larger samples, to preserve the spatial information relating to the separate storage/deposition of different products. These may be processed by hand as described in the recovery section.

For large deposits, a minimum of 40-50 litres of excavated sediment should be collected for each sample. Half of this quantity can be processed on site using a water-separation machine, as described in the recovery section. The coarse flot sieve may be scanned for plant macroremains to determine whether or not to process the rest of the sample. When the whole sample has been processed, the coarse flot sieve (and the heavy residue) may be scanned to decide whether or not to process further samples from the same or adjacent deposits.