Ab correlations
 ---------------

 3 analyses each producing a list of events used, charge sign decision and 
 event weights in the Ab fit. 
 
 The event list tabulation instructions were:
      write   Run,event,weight,abs(costh)
      format(1x,I8,4x,I8,4x,F8.4,4x,F8.4)  
   The cos(theta) values is explicitly tabulated, but by absolute value while 
   the weight is still signed as before and with the costh factor taken out.
   So the weight factor should include 
      Weight = Q * Wb * Fpol * (1-D_QCD) 
   where 
    Q  = charge  (charge of jet, vtx, lepton, kaon all turn out to 
                  be -ve for b quark).     
    Wb = analysis event weight, depending on b fraction and various
         analyzing power terms.
    Fpol = (Ae-Pe)/(1-Ae*Pe) 
    1-D_QCD = still evaluated as function of cos(theta) event by event   

 The same data sample of Winter98 was used with results: 
 File     Nevt    wt1-sum    wt2-sum    wt2-cth     First-evt     Last-evt
 Jet-Q    8359   2313.029    893.731     96.652   41112   229   42596  8010
 Lepton   6633   1262.361    474.963     46.959   41060  1709   42596  8010
 Vtx+K    6547   2896.266   1342.100    146.304   41112   229   42596  8010

              Jet-Q             Lepton
              Nevent   wt2-cth  Nevent  wt2-cth      W1*W2
 Non-overlap    5609    67.034   3877    14.443
 Agree          1833    22.586   1833    25.432     18.938
 Disagree        920     7.032    920     7.085     -4.034
 Correlation coefficient =   0.2212

              Jet-Q             Vtx+K
              Nevent   wt2-cth  Nevent  wt2-cth      W1*W2
 Non-overlap    3344    37.278   1532    31.911
 Agree          3545    48.822   3545    83.222     54.965
 Disagree       1470    10.552   1470    31.171    -14.213
 Correlation coefficient =   0.3427

              Lepton            Vtx+K
              Nevent   wt2-cth  Nevent  wt2-cth      W1*W2
 Non-overlap    4546    23.513   4460   104.472
 Agree          1373    18.196   1373    29.207     19.289
 Disagree        714     5.251    714    12.625     -6.111
 Correlation coefficient =   0.1590

 Because the full analyses were covering different data periods, a set of 
 corrected correlation coefficients are made for the final overall 
 combination:
                    Win-98 data    actual analysis data periods  
                    -----------    -----------------------------
  JetQ   vs Lepton      22.1         22.1  (~100% overlap) 
  JetQ   vs Vtx+K       34.3         32.0       (87%) 
  JetQ   vs Ktag        23.1          8.3       (13%) 
  Lepton vs Vtx+K       15.9         14.5       (83%) 
  Lepton vs Ktag        10.4          4.3       (17%)
  Vtx-Q  vs Ktag          0            0   (93-5 separate from 96-98)    
  
  The corrections are based on  JetQ 93-5,96-98  wt=13/87
                                Lept 93-5,96-98  wt=17/83
                                Ktag 94-5 only   wt=100/0
                                Vtx-Q 96-98 only wt=0/100 
  Corr-coef correct ~ sqrt(data period overlap weight fraction)*Corr0