Archives are messy: they consist of lots of different types of material, and that material has been worked through in different ways by different kinds of analytical method. This will become important when you discuss the state of the field in your area. But it's also important for building
hypotheses, and for beginning to talk about methods.
So first you need to establish the details of your archive. Say it's agriculture: I'm using an example I talked about recently with one of my Ph.D. students. There are lots of different “mine shafts” here: there's agricultural history, there's plants and animals, there's agricultural tools and instruments and technology, there's studies of agriculture. There's art and music and literature and myths about agriculture.
To start sorting some of this you can think about the different databases available that map agriculture. So for instance you could open up one of the anthropology databases, put in a search term like “agriculture” and select a cut-off date. The ideal is to at least have a knowledge of everything ever written about it, and every aspect of agriculture, everywhere, on Earth. Yes
that's right! Of course you won't do this but approach each archive with that kind of wide-angle view.
Once you've figured that out for each archive, you can create some hypotheses. Just study your archives until some hypotheses emerge. These will be the cutting edges of the machines you will use to work on the materials you find in your archives. A hypothesis converts your archive material into dissertation material: material that you can use to explore your big topic(s).
A rigorous hypothesis is something you can train a C grade undergrad to do with no difficulty. Yes—your life will shortly be that of a C grade undergrad as you trawl through thousands of texts. You need to make it simple. A hypothesis that could result in a yes or no answer is best. You'll need about three per archive to get a good enough model, at this stage at
“There is x in agricultural text y.”
“Agricultural instruments change from y to z at time a.”
Anthropological studies of agriculture contrast with historical studies
because of x.”
“Agriculture x was not practiced in location y.”