Oh, I told you not to read the bits where I was talking rubbish...
I did realise that they were from the same roll - but perhaps didn't make the differences between specifics and generalities clear enough. This is the root cause of all human conflict, I believe. Were the two frames scanned with exactly the same settings? (Is the software doing something behind your back, out of your control?) How does the local contrast differ in the two examples? How does the density differ? These are the kind of questions to ask yourself, I guess.
The local contrast, density change and overall density will affect, to some degree or another, the exact way in which the developer works grain-by-grain. The more agitation, the less the effect, in general.
I've got nothing against scanners (and you don't need to justify using one) but some of them do behave oddly when they are dealing with opaque grains/grain clumps when the grain 'size' is in the same ball-park as the scanner resolution. Just another influencing factor.