There were some interesting papers concerning photographic plate processing for the spectrographic astronomy. Micro samples for GCMS analysis will be very local as the process is active in the gelatin layer. The active chemistry will be depleted adjacent to that surface. The bulk volume of the tank will trace a gradual depletion of the whole active chemistry. Bottom line by selecting your sample locations and tuning your agitation you will have to be very careful to not measure process point time differences rather than chemical process differences. Random timing will prevent time sensitive variations from hiding in a null or amplifying at a peak generated by synchronization with agitation or instance. A too large sample will become averaged by its own general volume draw, thus requiring a far more sensitive instrumentation for measurement. Such a wonderful two edged sword problem in technology. Happy exploration!