The numerical analysis of the oxygen flash yield Yn sequences, alone, does not allow to choose between two models: equal S state misses with non negligible double hits or unequal misses with nearly no double hits. Nevertheless, the comparison of the sequences in different conditions shows that the equal miss model is unrealistic: in very different experimental conditions (non saturating flash, different batch of Chlorella or chloroplasts), a parallel variation of the homogeneous miss and double hit factors is observed. This correlation seems strange within the equal miss model: misses come from incomplete reaction (i.e. for exemple insufficient light) and double hits i.e. double advancement come, in principle, from excessive light or too long flash; for these reasons, opposite variation of misses and double hits as a function of light intensity are expected. Within the equal miss model the inverse is exactly observed: at low flash light intensity (11%) which increases the misses, 16% of double hits are needed, which is quite unrealistic. In contrast, the unequal miss model explains such result quite naturally by a m athem atical property: any theoretical sequence with only a unique S state miss and no double hit can be fitted with homogeneous misses and double hits, which increase in parallel as a function of the damping. Evidence for unequal misses in oxygen flash yield sequence is provided by the heterogeneous properties of the light saturation curves (M. J. Delrieu, Biochim. Biophys. Acta 592, 478-494 (1980)). At high flash intensity, all, excepted the transition S'2 → S3, are saturated; the transition S'2-→S3 is far from saturation and its very large saturating light intensity is actually not known. A comparative study, in the same chloroplast batch, of the oxygen yield patterns with attenuated flashes and of the experimental saturation curves of S states shows that only photochemical misses (due to non saturation) exist. At high intensity, there is only a unique miss for the transition S2→S3 i.e. the probability for this transition is low. A model involving a second acceptor could explain the slow increase of transition probability of S'2→S3 at high flash intensity