Using data from ESSA Research Laboratory bulletins an estimate of the uncertainty for the times of beginning, maximum and ending, and maximum areas of Ha flares reported during 1967 is deduced and discussed. The need for an improvement of the reliability of the solar patrol service is also indicated.
During the last years we investigated the photometric structure of solar flares in Ha light and obtained development curves which showed some rather new aspects of these phenomena (Falciani et al., 1967(Falciani et al., , 1968(Falciani et al., , 1972a, b;, b;Falciani and Rigutti, 1972). As a byproduct, it seemed to us that some earlier criticisms (Falciani et al., 1967;Michard, 1970;Dodson and Hedeman, 1970; de Jager and Jappel (ed.), 1971) indeed be meaningful and the whole matter of solar patrol should be rediscussed and reorganized to be really useful to solar physics and geophysics. For this reasons we thought of some interest to contribute to clarify a bit more this point by analizing also the data in the ESSA Research Laboratory bulletins.
As it is immediately seen from the ESSA tables that the data for a given event are rather different from each other, we first checked if the parameters which may affect the estimates, i.e. the distance 0 of the considered event from the center of the solar disc and the maximum area A of the phenomenon, might play some role. For this, we grouped the events in three classes according to their importance (< 1, 1, ~>2), and from the ESSA tables for 1967 singled out groups with at least five observations. For these we computed three A T differences (the ones for the beginning, maximum, and ending of the events) of the single time measurements from the adopted average group value and considered these values as the y coordinate for correlations with the abscissas x = O, and x = Ag (with Ag = maximum group area). The same was done also for the quantities ~= (A i-Ag)/Ag. The number N of the observations considered for correlations approximately ranged from 900 to 1500 for importance < 1 flares, from 400 to 1300 for importance 1 flares, and from 50 to 100 for importance >_-2 flares. As the obtained correlation coefficients ranged from 0.001 to 0.050 no correlation was possible and we got histograms for the same quantities considered above. Figure I(A) shows some examples of this analysis. One can see that the data relative to the beginning and the maximum phase are very uncertain and those relative to the end are hardly significant if we want to use them for flare models. Some or even several minutes of time seem indeed to be a too large uncertainty. Figure I(B) shows what is the situation about area estimates. Again, an average error of about 50~o seems to be rather large. It comes out from here how much consideration should be given to a clear definition of what must be observed and how. However, our impression was that besides some