By primary documents, I mean files. These are normally foolscap papers in Manila folders. Foolscap is an old paper size measuring 13 x 8 inches. The files are individually indexed on the National Archives' computer system, RecordSearch. Inside is the routine paperwork generated by the military machine, the conference notes, correspondence, reports and messages. The messages in particular give a dramatic flavour to the files.
The idea is to extract the facts from the documents and gradually build up a picture of what happened. It is the historian's job to reduce chaos to order. I work from a top down sort of framework. I have a chapter structure and try and write up a chapter at a time. Currently, there are 12 chapters defined. They are organised geographically rather than chronologically. So far only two chapters have been produced, and they were not on the plan. This comes from not knowing the story in the first place. It does generate that sense of adventure that you normally get as a reader.
I have two techniques for writing up a chapter from the primary material I have gathered. One is to just start at the front and move through what I a have to say, marshalling the details from my notebooks, photocopies and books as I go along. This can be painfully slow, as things stop every time something needs to be looked up. The other way is to work through the documents themselves, slotting the details into the appropriate place. This is much faster and more natural to me. The problem is that the nature of the thesis is that I sometimes don't really know what happened at all and so are forced back to the earlier technique.
My method is always to write the paragraph first and get it working later. The main problem is the contents depending on things that I haven't mentioned yet. Logistics is all about details, and something happening in one area can trigger a cascading series of events. This is the strength of talking about logistics as a whole instead of breaking it up into its components. But the paragraphs can easily be moved around. The footnotes just need to be kept close, so they travel with it.
The art of history writing is to make the complicated comprehensible.