According to [Draguhn et al], the prevailing view in 1998 was that gamma oscillations are for synchronizing information transmission related to a given object. For example, when thinking about a lemon, information burts corresponding to "round" and "yellow" might fire in synch at 40 Hz. If multiple objects are under consideration, they can be kept separate by spacing them at different phases.
However, perhaps a better supported view is that the theta rhythm is the point of reference for phase. In this view, each 40 Hz cycle represents one "clock tick", and synchronized events can fire anywhere within the same 40 Hz cycle. The "phase" then refers to which 40 Hz cycle is chosen within the 5 Hz theta cycle (apparently, there are about 7 different 40 Hz cycles nested within each 5 Hz cycle (yeah, I know those numbers don't add up, but that's what the following article says)). Evidence for this view is cited in the forth paragraph of [Lisman's 1998 Nature article].
They occur during rest or sleep. They are spacially coherent up to 5 mm.
There is evidence that the 200 Hz synchronization is mediated through gap junctions (see [Jefferys' web page] or the [Draguhn et al] paper).
One theory is that memories are first temporarily stored in the hippocampal CA3 region during exploratory behavior and the theta rhythm; then, later on, during rest or sleep, the 200 Hz rhythm facilitates more permanent learning ([Bibbig and Wennekers]).
Lisman, John. [What makes the brain's tickers tock]. Nature 394, 132 - 133 (July 9, 1998).