A statistical model supports the subunit hypothesis of quantal relsease.

Successive frequency amplitude histograms of miniature endplate potentials (MEPPs) from mouse neuromuscular junctions show peaks at integral multiples of the smallest peak (s-MEPPs). The s-MEPPs are assumed to have normally distributed amplitudes and to represent the subunit size. A model has been derived from assumptions based on the hypothesis of MEPPs are generated by the synchronous release of transmitter subunits. The model was statistically tested on amplitude histograms and in the majority of cases the model produced extremely good fits to observed distributions.