Back to EWK
With 20 events from 1987 CDF obtained Mw=80.0 +/-3.3 (Run -1) as
documented in
PRL 62 1005 (89) .
With 4pb-1 of run-0 data (1988-1989), CDF obtained : Mw = 79.91 +/- 0.39
GeV, as documented in
Phys.Rev.D43:2070-2093,1991.
Setting the momentum scale we use the Psi (or for recidivists, J/Psi) mass . After you align the r phi view of beam constrained tracking, you have to worry about how much material the tracks go through doing dE/dx. This was determined from the W electron E/p tail and confirmed using conversions normalized to the central tracker (CTC) inner wall or some wires. You need to worry, among other things, about residual B field nonuniformity, and even though z scale is not directly relevant it effects the extrapolation for psi momenta to 38 GeV/c for Ws. We take the unscaled result to contribute to a 50 MeV uncertainty from momentum scale. It is checked with upsilons and with Zs. The momentum resolution is unfolded from the Z width and its (0.081 +/-0.0085 % x p) uncertainty gives (60 MeV) in the W mass.
After tuning (CDF2487) the response of the central EM calorimeter, the momentum scale is transfered to the calorimeter by matching the radiative shift in the E/p peak . The width of the peak acts as a check on p and E resolutions; as for p, the E resolution comes the width of the Z peak. Resolution uncertainty (13.5 % stochastic, 1 +/-1 % constant) contributes (80 MeV) and the matching, including a lot of systematic paranoia, contributes (110 MeV) to the W mass.
The recoil ("u") is reconstructed without scaling up the calorimeter. Lepton towers are knocked out, including adjacent phi for near edge e, and 30 MeV is put back according to the Sarah Eno legacy. Associated systematics are (25 MeV) e and (10 MeV) mu. Pt W raw = |u| < 20, and no jet <30, and no extra track >10 clean up the basic 25/25 Et/missEt selection. Track iso (e) and cosmic removal (mu) the leave 5718 e and 3268 mu events with transverse mass between 65-100. The tiny background left in the e sample gives (10 MeV) and the slightly bigger crud level in the mu sample gives (25 MeV).
Rather than modelling calorimeter response to recoil, Z ee events are used as a lookup table for backround events for appropriate pt as measure using the ee. The overall resolution is set by starting with a model input W pt distribution given by the observed Z pt distribution and scaling it to match the width of the u projections perpendicular to the electron or muon . Uncorrected pt (u) distribtions for electron and muon events are fairly well reproduced. Associated systematics on the input pt distribution are (45 MeV). The lepton parallel ones for e and mu look ok as do trends in u parallel with about any relevant variable, see fnald::cdf$w_z_data:[ana.wmass.run1a_blessed]uparavs*.ps. The calorimeter u underreconstruction is well illustrated looking at ee vs u for Zs projected onto the bisector eta direction of the 2 e directions. The finite number of Zs and associated systematics contribute (60 MeV).
The rest of the MC model used to generate transverse mass shape 65-100 is just leading order production with NLO PDFs with the pt W put in by hand. The measurement of the W lepton asymmtry constrains PDF uncertainties for e and mu . We take a two sigma sigma for PDFs of (50 MeV). We think we understand QED radiative corrections to (20 MeV) and we messed with QCD higher order and the W width and they are both comperable. The fits for e and mu give (145 MeV) and (205 MeV) statistical, which are as expected. One can vary cutoffs and subdivide the sample in various ways to check and other that CPT - we are statistically entitled to one clunk - it all looks fine. Lepton Et distributions for e and nu(e) and mu and nu(mu) look fine. The 80.49 +/-0.145 +/-0.175 for electrons can be combined with 80.31 +/-0.205 +/-0.130 for muons, accounting for common stuff, giving 80.41 +/-0.18. This unfortunately fits in well with expectations and previous measurements .