Is it possible to have a zone so thoroughly cracked that self-similarity fails at scales larger than 1 km? This is most likely the case in the Eastern Tenn. seismic zone. This zone crackles away like popcorn with small earthquakes, and there is no evidence that large earthquakes are possible. As I've mentioned before, fractal self-similarity has a range, from the smallest to the largest. Usually, the smallest is at the scale of the actual rock crystals. The large limit depends on the geometry of the fault zone and determines the maximum magnitude for the largest earthquake.
For a large seismic zone, such as the San Andreas, the top limit is at a thousand klicks or so. For Pacific subduction zone, the limit is thousands of klicks. Interior zones, such as Hamilton or New Madrid have varying top limits. For Hamilton, it is around 20 km, which limits things to M6.5 or so. New Madrid has a limit of around 100-200 km (per section), which gets it up to an M7.5. As I have said before, these zones grow with every earthquake.
But the Eastern Tenn. zone has always fascinated me. It appears totally disorganized, and could never spit out a major earthquake. With this zone, I have always wondered if it could evolve to a standard zone, or if one could possibly determine what the maximum magnitude is.
The Western Quebec seismic zone is somewhat similar, as long as one stays away from the major rifts. It is most likely thoroughly fractured and has a low 'upper limit'.