Many works attain canonical status in their field and will never be too old. A few examples without going too far back: Adam Smith, Max Weber, E. H. Carr. Perhaps a student of history a generation from now will read Richard J. Evans' defense of history against postmodernism. That doesn't make Carr less relevant to what the postmodernist critique was of, just as von Ranke remains relevant to understanding Carr if time constraints prevent reading von Ranke himself. Books dealing with "facts" are trickier.

A narrative history of Egypt is going to change from generation to generation for a variety of reasons. New research, new perspectives, new tastes in what people enjoy. What remain are landmark studies and books that shaped public perception of a topic enough to be relevant to understanding future paradigms. The Atlantic Slave Trade: A Census is a work from 1969 that came out on the cusp of a revolution in quantitative data and remains relevant to the field. It led to articles like 1985's "Deaths of Slaves in the Middle Passage" that challenged previous assumptions about a phenomenon called tight packing, where slaves were given as little room as possible in what historians believed was a calculated decision to let some of them die in the name of greater profits. The paper showed a negligible difference in slave deaths based on allotted space.

Someone wishing to study tight packing is going to examine that article unless and until someone else unearths new data that makes its conclusions not worth studying. Important distinction: Not worth studying is not the same as coming to a different conclusion. The article remains a valid data point until slavery scholarship moves on with new methods, perspectives, etc. It becomes part of the history of the topic when it's no longer cited except as an example of where the field was at a given time.

Hope that makes sense. How to determine relevancy is usually a matter of reading the current journals and monographs in your area of interest. Check someone's bibliography to get a feel for what's important (and how they used it!).

Great question. I think any book written on history can never be too old. An exact time is different for every case but there comes a point at which the book becomes almost as interesting due to the thought process of the time as much as the subject.

In saying that, I think most history books have their place in the records. Providing they were written with discernible intentions, they serve a better record than word of mouth.

It depends on how specific the subject matter is and how heavily it's been studied. If it's a broad subject (eg Roman Britain), one that's had a recent popularity (eg gendered spaces) or one which is reliant on technology (eg chemical analysis of pottery to determine origin of clay) you want to use the most recent books and journals you can find. (My rule of thumb when I was a student was to only use modern sources that were younger than me for my initial research, though I would add older sources as I dug deeper if they were cited by my newer sources or covered specific topic areas I didn't have much on in the initial stage).

For specific subject areas that aren't covered much (like beekeeping in classical Attica), subjects where scientific analysis doesn't play a large part (like analysis of a text or inscription) or subjects which had their peak of popularity a long time ago (like art-historical looks at scultpure poses in classical Greek art) then older works may be the only thing available, or more recent works might not add anything except to summarise disparate older works.

Generally, though, I'd suggest seeing what there is on a topic before jumping in and reading the first thing you come across.

So, for PhD testing purposes, my advisor's policy is that her students read everything every written on Japanese medieval history in English, ever. This starts in 1911. But part of the purpose of reading all of these really old, incredibly outdated (far fewer documents available, interpretations scholars reject now, etc.) works is to understand them as foundations for the work that built on them. They're still important, just in a different way.

It's not that old books are less accurate per se (sometimes they are) but the biases are different. They're still fascinating though, and taken with a grain of salt, as everything should be anyway, they can be a great read and source of information!

The long and short of that is, in my opinion, information in books is never obsolete - it's function just changes.

One way to look up whether or not the information still holds water with current historians is to search for reviews. You can often read at least the first few paragraphs for free in journal databases. If they're too old for that, you can back search what books have referenced them. Google books is great for this. The citation will often show up.