If we consider that the granules are different and react diferently, and it seams to be known how a particle that react well should look like, couldn't the testers have picked a biased sample for their expensive analysis e.g. a particle that have had the time to saturate in order to ne sure to get an interesting sample. If the number of such particles are say 25% and the rest is a spectra of non saturated granulles it would not be a surprise of the findings. There might still be plenty of active particles left with a slightly different morphology that still reacts. Also the setup is actively controlled and it is not unheard of getting such systems to run at constant power output although the reaction potential decreases with time until a cutoff happens. Just my 2c. Cheers!
