From the point of view of a geologist, tens of thousands of years is not very much.
It is useful to an archeologist, but to a geologist that's just the recent past.
It seems, then, that we can't do dendrochronology unless we already have a way to determine the age of a piece of wood.
If this was the case, it would be one of the more useless scientific techniques. Suppose we take a core sample from a tree which grew between 1500 AD and the present; that gives us fingerprints for the past 500 years or so.
It is a well-known fact that many tree genera will produce one new growth ring each year.
In addition a dendro-archaeological study allows to rediscover the old know-how of wood artisans, supplies indications concerning forest management, the selection of raw materials, etc.By continuing this process with older and older samples of wood, we can build up data stretching back tens of thousands of years.This technique is known as crossdating; similar principles can be employed in other absolute dating methods.Observing the identities between these fingerprints, we can now put a date on each of the rings of the dead wood, which allows us to extend our knowledge of what the fingerprints look like back to 1100 AD, four hundred years before the living tree took us.Now if we find another dead sample which runs from 800 AD to 1250 AD, its tree-rings have a 150-year overlap with the known sequence, we can use this to date it, and then we can extend the sequence still further.