OK, Let's see if people can agree on these rules as a start to keep graininess down:
1) Overdevelopment increases grain.
2) Overexposure combined with overdevelopment will give you even more grain.
3) Underexposure and overdevelopment will give you grain in the dense areas and not in the thin (underexposed) areas.
4) Expose your negatives 'just enough' to give you adequate detail in the shadows - no more and no less. In Zone terms, this is Zone III.
5) Develop your negs 'just enough' so that they don't build up too much density and thus too much grain. 'Just enough' is a density such that you can just read newprint thru it in a sunny room. You should not be able to read the print too easily, but it should be readable not just discernable.
6) Keep wet time to the minimum for processing.
7) Keep temps consistent to avoid reticulation.
8) Avoid taking picutres of wood as it's by nature 'grainy'.
9) Since a really thin negative has little or no grain, and a really dense one has lots of it, there must be some correlation - eh?
Also, this idea of rating film at varaious ASA's and pushing and pulling processing are really terms that show a lack of the basic concepts of sensitometry. I am not putting anyone down, so please don't throw rocks at your monitor. I simply mean that for your camera\film\dev there is one and only one correct ASA speed. By the definitions of ASA (American Standards Assoc), the ASA speed is that which produces .1 above film base+fog. There is only one ASA speed that can do that, and your mission, if you choose to accept is to find that speed. If caught, Ansel Adams will disavow any knowledge of your actions.
So, I'd recommend sticking with the rated ASA. All it means to rate it at 320 is to give 1/3rd stop more exposure. That's it. Pushing and pulling processing just mean changing the develpment times to compensate for exposure times so that you either build up or contract densities to stay in the ranges that I gave in steps 4 & 5. These terms are mostly used by people who don't own a densitometer (are there actually such people out there? do they secretely whish they had one?), don't understand or care about sensitometry. In short, they're interested in what they can get on the negative which is what it's all about, Alfie. The ZS and sensitometry make a system out of it so it is reliable, repeatable and predictable. How many people do you know who 'push' the film 3 stops, have it push processed and 'hope' that it comes out? If they knew the ZS and sensitometry, they'd know in advance. OK, enough preaching - wine induced I promise you all!!
Does that sum up the reasons for grain, folks? I'd say if different frames showed different levels of graininess then it directly relates to the exposure, development and processing practices employed.