Is there a problem with low energy SN1987A neutrinos?
(abridged) We study the low energy events observed by Kamiokande-II. We focus on the event 6 of 6.3 MeV and also on the other events that at 1 sigma fall below the energy threshold: events 3,4,10,12. The volume distribution is not a uniform distribution at 3 sigma, that suggests the presence of background events close or at the border of the volume used for the analysis, including the events 3,4,10. We checked the expected energy distribution assuming that the signal is due to nubar(e)p-->e+n and that the average antineutrino energy is 14 MeV. The agreement with the observations is not perfect but it is acceptable at the 11 % confidence level if we include the peak of low energy background events; otherwise, we face a 2.9 sigma problem. The expected energy distribution implies that the evidence for supernova neutrinos is at 10 sigma and that 1-3 background events are plausible. This conclusion does not change strongly when we model the time distribution of the signal taking into account the presence of an initial luminous phase of neutrino emission. This suggests however that some of the early events are due to supernova neutrinos and not to background. In summary, we formulate the hypothesis that some of the observed low energy events are due to background and that some among them belong to a peculiar phase of emission, that could be further characterized by low energy neutrinos. Such an interpretation diminishes to a minimum the postulated number of background events and thus improves the agreement between the a priori and the a posteriori expectations on the number of background events. We argue on these grounds that there is no significant disagreement between the average energy of the supernova neutrinos seen in Kamiokande-II and the conventional expectations.