A doubt about histogram classes

Dear ROOTers,
A matter I have never well understood is the way that histogram data are recorded in the ROOT TH1,etc…, such as in HBOOK/PAW histograms (I imagine the implementation is the same).
In the structure of the histogram, is recorded the number of event per bin, OR the data as it is, filling then the histogram every time I plot it?
It is different, because e.g. if I rebin the histogram with a larger number of bin, in the first case some bins will remain empty. Also when I save the histogram to file, in the first case I would lose the exact values of the registered event.
I think the correct answer is the second one (it seems to me more correct, and I made some tests), but I want a confirmation. If anyone knows that HBOOK/PAW works in a different manner (I apologize if I use this forum) about it, please tell me


In both PAW /HBOOK and ROOT each histogram bin contains one value only, not the whole collection of events which filled that bin.
When you need to rebin with more bins you should start again from the original Tree or Ntuple.
I think there is plans to have histograms with the list of events, but I am unsure about it. Wait Rene wil be back April 13th to get an answer