How can archaeologists tell so much from such small pieces of pottery and other tiny bits of evidence?

There are two basic requirements. First, we must know what an object is, and second, it must be found in context.

Experts have classified and dated archaeological objects over years of recorded excavation. For a classic example of this, have a look at Roman Pottery in Britain by Paul Tyers (Routledge, 1996), in which pots are classified to such an extent that it is like reading an Ikea catalogue! Such research is continually being revised and updated, so what we think we know now may be proved wrong in the future, but it remains an excellent method of dating finds. The best objects for this are coins, although these may have remained in use for a long time after they first appeared.

A find is of limited use unless it is discovered in some form of context; it can then be used to date the structures around it. If a find is part of the stratigraphy – layering – of previously undisturbed archaeology, we can, in theory, date the particular layer in which the find was unearthed. For example, we know so much about Roman Samian ware that a sherd found in context next to masonry can date that masonry. We have to say 'in theory' because a sherd could have been part of a pot that was already 100 years old when it was broken. So the process is not flawless – but it's a far better indicator than nothing at all.