Coarse-grained images are often claimed to suffer from something called “grain clumping,” which in turn has been ascribed to too long a development time, too high a development temperature, too alkaline a developer, and too long a total wet time, among other reasons. Detailed rationales of clumping describe how individual emulsion grains swim together in the emulsion to form these clumps (doesn’t happen); others claim the origin is development by-products from one developing grain diffusing out and rendering nearby unexposed grains developable (also doesn’t happen except with lith films in lith developers). Still others confuse extreme graininess with reticulation (an altogether different “dry lake bed” effect resulting from a temperature shock between solutions that shatters the gel matrix).
The many references to grain clumping do have a valid premise in that they recognize that what we term graininess in a print is not a visual response to individual emulsion grains. These are far too small to resolve in even a very big enlargement. What we are reacting to is indeed aggregates of several grains. So call these aggregates “clumps” and there is surely some driving force responsible for their formation. Right? Nope. The clumps are of purely stochastic origin: The distribution of developed grains in a uniformly and moderately exposed area of a negative is completely random, and this randomness guarantees the presence of clumps.