I noticed a problem with THnBase::Add(THnSparse* h, Double_t c =1). If c > 0, then all is well. However, if c < 1, then, in my case, for the TH1D objects created from the Projections of the resultant THnSparse, the entries in the statistic box are all set to 0. Only the Integral value is meaningful, but the Mean, RMS etc are all set to 0.
I have uploaded the code (nb-thnsparse-test.py) that I used to reproduce this problem (I wrote my code using iPython Notebook).
Following is the version information of ROOT that I am using (obtained directly from CINT after starting ROOT):
ROOT 5.33/03 (trunk@43806, Jun 14 2012, 07:21:53 on linuxx8664gcc)
nb-thnsparse-test.py (904 Bytes)