9/12/97


                 LEP ERROR MATRICES FOR 93,94,95
                 ===============================

(these files under /afs/cern.ch/phys/lep/lepecal/95/err_matrix)

The files:

 l3.dat, aleph.dat, opal.dat, delphi.dat

are 7x7 matrices which each experiment should use in fitting their own data
to get the results of that experiment,  ie. the matrices contain both the
common errors and the IP errors for that experiment.

 cern.dat

is the 7x7 matrix which each experiment should use to fit their data in
preparation for combining with other experiments, ie. it contains the
common errors and the IP uncorrelated errors downweighted by the appropriate
factor.  This matrix serves the same function as that which was provided
for Jerusalem, although its elements have changed very slightly.

 full_28x28.dat

is the full matrix containing all correlations between years and IPs.  It
should be used if a simulaneous fit is made of all cross-sections of all
experiments.  Note that  l3.dat, aleph.dat, opal.dat delphi.dat are merely
the diagonal blocks of this larger matrix.

The rows and columns of each matrix are as listed.  The 7x7 matrices
have 93P-2, 93P, 93P+2, 94, 95P-2, 95P, 95P+2.  The 28x28 matrix has
experiments ordered in terms of IP, ie.  l3, aleph, opal, delphi.

Note that the errors are given for the scan periods of 93 and 95, and for
the main running period of 94.  For the errors on the pre-scan periods,
and the bunch train commissioning please refer to the LEP energy paper.

All values are for ECM in units of (MeV)^2.

The 28x28 matrix has been written with format  6X,28(F6.2).
The 7x7 matrices have been written with format  6X,7(F6.2).


                                                     Guy Wilkinson  9/12/97