The pollen record is retrieved from land. Soil is collected from "today" back 10,000 years, just after the glaciers retreated. It isn't worth collecting samples below glacial till because it's likely that soil has been churned, and no longer sits in the correct, reverse chronological, order. The soil can be collected through a number of methods depending on the environment. As there are no naturally occurring (long-running) lakes on Dartmoor, core samples need to be taken from mechanical boring, or from cuts in the land (e.g., from streams, or dug holes).
The samples of soil are processed in a lab, and the pollen grains are counted. Normally, 500 grains are counted for each "zone" collected. A zone is a visually cohesive length within the core. This might be based on colour, texture, density, or compactness. e.g., sedge peat vs. gravel.
For a grain of pollen to be located in the soil, a nearby tree needed to produce pollen, which was then blown by the wind into the exact spot where the soil was collected from. Obviously, the more there is of a specific type of plant, the higher the chance of the pollen landing "everywhere" and thus turning up in the pollen record. This is why we don't see insect-pollinated plants appearing as frequently in the pollen record.