Index: /trunk/MagicSoft/Mars/Changelog
===================================================================
--- /trunk/MagicSoft/Mars/Changelog	(revision 2323)
+++ /trunk/MagicSoft/Mars/Changelog	(revision 2324)
@@ -1,3 +1,19 @@
                                                  -*-*- END OF LINE -*-*-
+
+ 2003/09/07: Abelardo Moralejo
+   * manalysis/MCerPhotCalc.cc:
+     - removed normalization of array fWeight introduced on 30/06. For
+       now, this array is intended only to select the FADC slices which
+       we want to integrate to get the signal. The use of this feature
+       of MCerPhotCalc is explained in the example macro MagicHillas.C
+       The array fWeight must contain 0's and 1's. If we normalize
+       these values, the signal units is no longer ADC counts (but 
+       "average ADC counts per slice"). As David realized, this was the 
+       reason for the discrepancies in the gamma/hadron separation 
+       results we were observing lately. The meaning of Size was 
+       different among the different people working on the matter, and 
+       hence, when we compared samples with a given Size cut we were
+       actually comparing different energy ranges.
+
 
  2003/08/27: Thomas Bretz
Index: /trunk/MagicSoft/Mars/manalysis/MCerPhotCalc.cc
===================================================================
--- /trunk/MagicSoft/Mars/manalysis/MCerPhotCalc.cc	(revision 2323)
+++ /trunk/MagicSoft/Mars/manalysis/MCerPhotCalc.cc	(revision 2324)
@@ -117,15 +117,9 @@
 
     // Calculate quadratic sum of weights:
-    Double_t sum = 0;
-
     fSumQuadWeights = 0;
     for (Int_t i=0; i<fWeight.GetSize(); i++)
-        sum += fWeight[i];
-
-    for (Int_t i=0; i<fWeight.GetSize(); i++)
-    {
-        fWeight[i] /= sum;
-        fSumQuadWeights += fWeight[i]*fWeight[i];
-    }
+      fSumQuadWeights += fWeight[i]*fWeight[i];
+
+    fSumQuadWeights = sqrt(fSumQuadWeights);
 
     return kTRUE;