hep-ex0612034/mw.tex
1: 
2: \section{Introduction}
3: 
4: The W boson mass and width results presented here are obtained from
5: data recorded over a range of centre-of-mass energies,
6: $\sqrt{s}=161-209$~\GeV, during the 1996-2000 operation of the LEP
7: collider (LEPII), and correspond to a luminosity of about $700~\ipb$
8: per experiment.  In contrast to previous
9: combinations~\cite{bib-EWEP-05}, the results reported by all four LEP
10: experiments, ALEPH, DELPHI, L3 and OPAL, are now final.  However, the
11: combination and its results are still preliminary.  The final W mass
12: results of the experiments in the \WWqqqq ~channel use an analysis
13: trading statistical accuracy for reduced FSI systematic uncertainties
14: in order to achieve a smaller total error.
15: 
16: The observables W mass, $\Mw$, and total width, $\Gw$, quoted here
17: correspond to a definition based on a Breit-Wigner denominator with a
18: mass-dependent width, $|(m^2-\Mw^2) + i m^2 \Gw/\Mw|$.
19: 
20: 
21: \section{W Mass Measurements}
22: 
23: Since 1996 until 2000 the LEP \epem collider operated above the
24: threshold for $\WW$ pair production. Initially, 10~\ipb\ of data were
25: recorded close to the $\WW$ pair production threshold. At this energy
26: the $\WW$ cross section is sensitive to the W boson mass.  Table
27: \ref{mw:tab:wmass_threshold} summarises the W mass results from the
28: four LEP collaborations based on these
29: data~\cite{common_bib:adloww161}.
30: 
31: \begin{table}[htbp]
32: \begin{center}
33: \begin{tabular}{|r|c|}\hline
34: \multicolumn{2}{|c|}{THRESHOLD ANALYSIS~\cite{common_bib:adloww161}} \\
35:  Experiment &   \Mw(threshold)/\GeVm     \\ \hline
36:    ALEPH    & $80.14\pm0.35$             \\ 
37:    DELPHI   & $80.40\pm0.45$             \\
38:    L3       & $80.80^{+0.48}_{-0.42}$    \\
39:    OPAL     & $80.40^{+0.46}_{-0.43}$     \\ \hline
40: \end{tabular}
41: \caption{W mass measurements from the $\WW$ threshold cross section at
42: $\roots=161$~\GeV. The errors include statistical and systematic
43: contributions.}
44: \label{mw:tab:wmass_threshold}
45: \end{center}
46: \end{table} 
47: 
48: \begin{table}[htbp]
49: \begin{center}
50: \begin{tabular}{|r|c|c||c|}
51: \hline
52: \multicolumn{1}{|c|}{ } & \multicolumn{3}{c|}{DIRECT RECONSTRUCTION } \\
53:            & \WWqqln         & \WWqqqq         & Combined             \\   
54: Experiment & \Mw/\GeVm        & \Mw/\GeVm        & \Mw/\GeVm          \\ 
55: \hline
56: ALEPH \cite{mw:bib:ALEPH}
57:            & $80.429 \pm 0.059$ & $80.475 \pm 0.081$ & $80.444 \pm 0.051$ \\ 
58: DELPHI \cite{mw:bib:delphi}  
59:            & $80.339 \pm 0.075$ & $80.311 \pm 0.137$ & $80.336 \pm 0.067$ \\ 
60: L3 \cite{mw:bib:mwl3}      
61:            & $80.212 \pm 0.071$ & $80.325 \pm 0.080$ & $80.270 \pm 0.055$ \\ 
62: OPAL\cite{mw:bib:O-final}
63:            & $80.449 \pm 0.063$ & $80.353 \pm 0.083$ & $80.416 \pm 0.053$  \\ 
64: \hline
65: ALEPH
66:            & $80.429 \pm 0.059$ & $80.475 \pm 0.082$ & $80.444 \pm 0.051$ \\ 
67: DELPHI
68:            & $80.340 \pm 0.076$ & $80.310 \pm 0.102$ & $80.330 \pm 0.064$ \\ 
69: L3
70:            & $80.213 \pm 0.071$ & $80.323 \pm 0.091$ & $80.253 \pm 0.058$ \\ 
71: OPAL
72:            & $80.449 \pm 0.062$ & $80.353 \pm 0.081$ & $80.415 \pm 0.052$  \\ 
73: \hline
74: \end{tabular}
75: \caption{W mass measurements from direct reconstruction
76:          ($\roots=172-209$~\GeV). Results are given for the
77:          semi-leptonic, fully-hadronic channels and the combined
78:          value. The top part of the table shows the results as
79:          published by the experiments, using their individual
80:          evaluations of FSI effects; these results are final.  The
81:          bottom part of the table shows the results of the experiments
82:          when propagating the common LEP estimates of FSI effects to
83:          the mass (see text), affecting also the $\WWqqln$ results
84:          through correlations due to other systematic uncertainties;
85:          these results are preliminary.  The $\WWqqln$ results from
86:          the OPAL collaboration include mass information from the
87:          $\WWlnln$ channel.  }
88: \label{mw:tab:wmass_experiments}
89: \end{center}
90: \end{table}
91: 
92: Subsequently LEP operated at energies significantly above the $\WW$
93: threshold, where the $\epem\rightarrow\WW$ cross section has little
94: sensitivity to $\Mw$. For these higher energy data $\Mw$ is measured
95: through the direct reconstruction of the W boson's invariant mass from
96: the observed jets and leptons.  Table~\ref{mw:tab:wmass_experiments}
97: summarises the W mass results published by the four LEP experiments
98: using the direct reconstruction method.  The combined values of $\Mw$
99: from each collaboration take into account the correlated systematic
100: uncertainties between the decay channels and between the different
101: years of data taking. In addition to the combined numbers, each
102: experiment presents mass measurements from $\WWqqln$ and $\WWqqqq$
103: channels separately.  The DELPHI and OPAL collaborations provide
104: results from independent fits to the data in the $\qqln$ and $\qqqq$
105: decay channels separately and hence account for correlations between
106: years but do not need to include correlations between the two
107: channels. The $\qqln$ and $\qqqq$ results quoted by the ALEPH and L3
108: collaborations are obtained from a simultaneous fit to all data which,
109: in addition to other correlations, takes into account the correlated
110: systematic uncertainties between the two channels. 
111: %
112: 
113: The precision of the $\qqqq$ channel is limited by the Final State
114: Interactions (FSI) systematic, namely Colour Reconnection (CR) and
115: Bose-Einstein Correlation (BEC). Hence, the final LEP analyses remove
116: low momentum particles from the jets as these particles are more
117: sensitive to FSI effects.  Removing low momentum particles from jets
118: reduces the systematic uncertainties due to possible FSI effects in
119: the $\WWqqqq$.  The systematic errors in the $\WWqqln$ channel are
120: dominated by uncertainties from hadronisation.
121: 
122: 
123: \section{Combination Procedure}
124:  
125: A combined LEP W mass measurement is obtained from the results of the
126: four experiments. In order to perform a reliable combination of the
127: measurements, a more detailed input than that given in
128: Table~\ref{mw:tab:wmass_experiments} is required.  Each experiment
129: provided a W mass measurement for both the $\WWqqln$ and $\WWqqqq$
130: channels for each of the data taking years that it had analysed. In
131: addition to the four threshold measurements a total of 38 direct
132: reconstruction measurements are supplied: ALEPH provided 8
133: measurements (1997-2000), DELPHI provided 10 measurements (1996-2000),
134: L3 gave 10 measurements (1996-2000), and OPAL gave 10 measurements
135: (1996-2000). The $\WWlnln$ channel is also analysed by the OPAL
136: (1996-2000) collaboration; the lower precision results obtained from
137: this channel are combined with the mass determinations in the
138: $\WWqqln$ channel.
139: 
140: Subdividing the results by data-taking years enables a proper
141: treatment of the correlated systematic uncertainty from the LEP beam
142: energy and other dependences on the centre-of-mass energy or
143: data-taking period.  A detailed breakdown of the sources of systematic
144: uncertainty is provided for each result and the correlations
145: specified. The inter-year, inter-channel and inter-experiment
146: correlations are included in the combination. The main sources of
147: correlated systematic errors are: Colour Reconnection, Bose-Einstein
148: Correlations, hadronisation, the LEP beam energy, and uncertainties
149: from initial and final state radiation. The full correlation matrix
150: for the LEP beam energy is employed\cite{mw:bib:energy}.  The
151: combination is performed and the evaluation of the components of the
152: total error assessed using the Best Linear Unbiased Estimate (BLUE)
153: technique, see Reference~\citen{common_bib:BLUE}.
154: 
155: \subsubsection{FSI Effects}
156: \label{mw:sec:cr}
157: 
158: In the old mass analyses all experiments had the same sensitivity to
159: FSI effects.  The final LEP analyses have a different sensitivity to
160: FSI, depending on the method and cuts employed to reduce this
161: systematic. Hence, {\it in contrast to our old preliminary
162: combinations we do NOT equalise the FSI systematic anymore}.  Instead,
163: we use as (energy dependent) uncertainty what the experiments find for
164: exactly the same FSI models and parameter variations, as detailed in
165: the following.
166: 
167: In the case of Bose-Einstein Correlations, the uncertainties are
168: evaluated using the LUBOEI model with full BEC, with the effect
169: rescaled to match the final LEP-2 combined one standard-deviation
170: limit of 30 \% of the full effect, see Chapter~\ref{sec-BE}.
171: 
172: %
173: %
174: %
175: %
176: %
177: %
178: %
179: %
180: %
181: %
182: %
183: %
184: %
185: %
186: 
187: A preliminary study of Colour Reconnection has been made by the LEP
188: experiments using the particle flow method \cite{mw:bib:CRcomb} on a
189: sample of fully-hadronic WW events, see Chapter~\ref{sec-CR}.
190: \footnote{This preliminary combination does not take into account the
191: final particle flow and $\delta M$ constraints on possible Colour
192: Reconnection models, nor the final OPAL Colour-Reconnection analysis
193: published recently~\cite{cr:bib:O-final}.  We expect this to change
194: once the final LEP combinations for CR effects will become available.}
195: The combined results are interpreted in terms of the reconnection
196: parameter $k_i$ of the SK-I model \cite{mw:bib:ski} and yield a $68\%$
197: confidence level range of:
198: \begin{eqnarray}
199:  0.39 < k_i < 2.13 \,.
200: \end{eqnarray}
201: The method was found to be insensitive to the \HERWIG\ and \ARIADNE-II
202: models of Colour Reconnection. 
203: 
204: As indicated above, the final LEP analyses have a reduced sensitivity
205: to CR effects. Some analyses were optimised to give the smallest total
206: error based on the experiment's own value of the SK-I model parameter
207: $k_i$.  For the combination, all CR errors have been re-evaluated for
208: the common preliminary LEP-2 value of $k_i=2.13$, without, however,
209: reoptimising individual analyses for this $k_i$ value. The mass
210: results propagating the common LEP estimates of FSI effects are also
211: shown in Table~\ref{mw:tab:wmass_experiments}.
212: 
213: Early \Mw\ combinations had relied upon theoretical expectations of
214: Colour Reconnection effects, in which there is considerable
215: uncertainty. This new data-driven approach achieves a more robust
216: uncertainty estimate at the expense of a significantly increased
217: Colour Reconnection uncertainty. The \ARIADNE-II and \HERWIG\ models
218: of Colour Reconnection have also been studied and the W mass shift was
219: found to be lower than that from SK-I with $k_i = 2.13$ used for the
220: combination.
221: 
222: \section{LEP Combined W Boson Mass }
223: 
224: The combined LEP W mass from direct reconstruction alone is
225: \begin{eqnarray}
226:    \Mw(\mathrm{direct}) = 80.375\pm0.025(\mathrm{stat.})\pm0.022(\mathrm{syst.})~\GeVm,
227: \end{eqnarray}
228: with a $\chi^2$/d.o.f. of 47.7/37, corresponding to a $\chi^2$
229: probability of 11.1\%. The weight of the fully-hadronic channel in the
230: combined fit, previously 0.16~\cite{bib-EWEP-05} as a consequence of
231: the relatively large size of the estimates for the systematic errors
232: from CR and BEC, increased to 0.22 in this combination due to the
233: experiments' final results with reduced FSI sensitivity.
234: 
235: Table \ref{mw:tab:errors} gives a breakdown of the contribution to the
236: total error of the various sources of systematic errors.
237: %
238: The largest contribution to the systematic error comes from
239: hadronisation uncertainties, which are conservatively treated as
240: correlated between the two channels, between experiments and between
241: years. In the absence of systematic effects the current LEP
242: statistical precision on $\Mw$ would be $20$~\MeV: the statistical
243: error contribution in the LEP combination is larger than this
244: (25~\MeV) due to the reduced weight of the fully-hadronic channel.
245: Compared to older combinations~\cite{bib-EWEP-04}, the final LEP
246: results lead to an increased statistical error in the $\WWqqqq$
247: combination but reduced systematics due to CR and BEC, for an overall
248: smaller total error.
249: 
250: 
251: \begin{table}[htbp]
252: \begin{center}
253: \begin{tabular}{|l|r|r||r|}\hline
254:        Source  &  \multicolumn{3}{|c|}{Systematic Error on \Mw\ ($\MeVm$)}  \\  
255:                              &  \qqln & \qqqq  & Combined  \\ \hline   
256:  ISR/FSR                                    &  8 &  5 &  7 \\
257:  Hadronisation                          & 13 & 19 & 14 \\
258:  Detector Systematics             & 10 &  8 & 10\\
259:  LEP Beam Energy                 & 9 & 9 & 9 \\
260:  Colour Reconnection            & $-$& 35 & 8 \\
261:  Bose-Einstein Correlations  & $-$& 7 &  2 \\
262:  Other                                        &  3 &  11 & 4 \\ \hline
263:  Total Systematic                    & 21 & 44 & 22 \\ \hline
264:  Statistical                                & 30 & 40 & 25 \\ \hline\hline
265:  Total                                        & 36& 59 & 33 \\ \hline
266:   & & & \\
267:  Statistical in absence of Systematics  & 30 & 27 & 20 \\ \hline
268: 
269: \end{tabular}
270:  \caption{Error decomposition for the combined LEP W mass results.
271:           Detector systematics include uncertainties in the jet and
272:           lepton energy scales and resolution. The `Other' category
273:           refers to errors, all of which are uncorrelated between
274:           experiments, arising from: simulation statistics, background
275:           estimation, four-fermion treatment, fitting method and event
276:           selection. The error decomposition in the $\qqln$ and
277:           $\qqqq$ channels refers to the independent fits to the
278:           results from the two channels separately.}
279:  \label{mw:tab:errors}
280: \end{center}
281: \end{table}
282: 
283: In addition to the direct reconstruction results, the W boson mass is
284: measured at LEP from the 10~\ipb\ per experiment of data recorded at
285: threshold for W pair production:
286: \begin{eqnarray}
287:       {\Mw(\mathrm{threshold}) = 
288:   80.40\pm0.20(\mathrm{stat.})\pm
289:           0.07(\mathrm{syst.})\pm0.03(\mathrm{E_{beam}})~\GeVm}.
290: \end{eqnarray}
291: When the threshold measurements are combined with the much more
292: precise results obtained from direct reconstruction one achieves a W
293: mass measurement of
294: \begin{eqnarray}
295:           \Mw = 80.376\pm0.025(\mathrm{stat.})\pm0.022(\mathrm{syst.}) \GeVm.
296: \end{eqnarray}
297: The LEP beam energy uncertainty is the only correlated systematic
298: error source between the threshold and direct reconstruction
299: measurements.  The threshold measurements have a weight of only $0.02$
300: in the combined fit.  This LEP combined result is compared with the
301: results (threshold and direct reconstruction combined) of the four LEP
302: experiments in Figures~\ref{mw:fig:mwgw-pub} and~\ref{mw:fig:mwgw-com}.
303: 
304: 
305: \section{Consistency Checks}
306: 
307: The difference between the combined W boson mass measurements obtained
308: from the fully-hadronic and semi-leptonic channels,
309: $\Delta\Mw(\qqqq-\qqln)$, is determined:
310: \begin{eqnarray}
311:  \Delta\Mw(\qqqq-\qqln) = -12\pm45~\MeVm.
312: \end{eqnarray}
313: A significant non-zero value for $\Delta\Mw$ could indicate that CR
314: and BEC effects are biasing the value of \Mw\ determined from \WWqqqq\
315: events.  Since $\Delta\Mw$ is primarily of interest as a check of the
316: possible effects of final state interactions, the errors from CR and
317: BEC are set to zero in its determination. The above result on the mass
318: difference is obtained from a fit where the imposed correlations are
319: the same as those for the results given in the previous sections. 
320: 
321: The masses from the two channels with all errors and correlations
322: included are:
323: \begin{eqnarray}
324:  \Mw(\WWqqln) = 80.372\pm0.030(\mathrm{stat.})\pm0.020(\mathrm{syst.})~\GeVm, \\
325:  \Mw(\WWqqqq) = 80.387\pm0.040(\mathrm{stat.})\pm0.044(\mathrm{syst.})~\GeVm.  
326: \end{eqnarray}
327: The two results are correlated with a correlation coefficient of 0.20.
328: These results and the correlation between them can be used to combine
329: the two measurements or to form the mass difference. The LEP combined
330: results from the two channels are compared with those quoted by the
331: individual experiments in Figures~\ref{mw:fig-qqlnqqqq-pub}
332: and~\ref{mw:fig-qqlnqqqq-com}.
333: 
334: Experimentally, separate $\Mw$ measurements are obtained from the
335: $\qqln$ and $\qqqq$ channels for each of the years of data.  The
336: combination using only the $\qqln$ measurements yields:
337: \begin{eqnarray*}
338:   M_{\mathrm{W}}^{\mathrm{indep}}(\WWqqln) = 
339:   80.374\pm0.030(\mathrm{stat.})\pm0.021(\mathrm{syst.})~\GeVm.
340: \end{eqnarray*}
341: The largest contribution to the systematic error arises from the
342: hadronisation uncertainties, $\pm13$~$\MeVm$.  The combination using
343: only the $\qqqq$ measurements gives:
344: \begin{eqnarray*}
345:  \Mwindep(\WWqqqq) = 80.389\pm0.040(\mathrm{stat.})\pm0.044(\mathrm{syst.})~\GeVm.  
346: \end{eqnarray*}
347: where the dominant contributions to the systematic error are from CR
348: ($\pm 35$~$\MeVm$) and hadronisation ($\pm19$~$\MeVm$).
349: 
350: 
351: \section{LEP Combined W Boson Width}
352: 
353: The method of direct reconstruction is also well suited to the direct
354: measurement of the total width of the W boson. The results of the four
355: LEP experiments, as published and when propagating the common LEP
356: estimates of FSI effects, are shown in Table
357: \ref{mw:tab:wwidth_experiments} and in Figures~\ref{mw:fig:mwgw-pub}
358: and~\ref{mw:fig:mwgw-com}.
359: \begin{table}[htbp]
360: \begin{center}
361: \begin{tabular}{|c|c|c|}\hline
362: Experiment & \Gw\ (\GeVm) & \Gw\ (\GeVm) \\ 
363: \hline
364:            &    published &       common \\ 
365: \hline
366: ALEPH      & $2.14\pm0.11$&$2.14\pm0.11$ \\ 
367: DELPHI     & $2.40\pm0.17$&$2.39\pm0.17$ \\
368: L3         & $2.18\pm0.14$&$2.24\pm0.15$ \\
369: OPAL       & $2.00\pm0.14$&$2.00\pm0.14$ \\ 
370: \hline
371: \end{tabular}
372:  \caption{W width measurements ($\roots=172-209$~\GeV) from the
373:          individual experiments.  The column labelled ``published''
374:          shows the results as published by the experiments, using
375:          their individual evaluations of FSI effects; these results
376:          are final. The column labelled ``common'' shows the results
377:          of the experiments when propagating the common LEP estimates
378:          of FSI effects to the width (see text); these results are
379:          preliminary. }
380: \label{mw:tab:wwidth_experiments}
381: \end{center}
382: \end{table}
383: 
384: Each experiment provided a W width measurement for both $\WWqqln$ and
385: $\WWqqqq$ channels for each of the data taking years (1996-2000) that
386: it has analysed. A total of 34 measurements are supplied: ALEPH
387: provided 8 results (1997-2000), DELPHI 8 measurements (1997-2000), L3
388: 10 measurements (1996-2000), and OPAL provided 8 measurements
389: (1997-2000).
390: 
391: %
392: %
393: %
394: %
395: %
396: %
397: 
398: The BEC and CR uncertainties supplied by the experiments were based on
399: studies of phenomenological models of these effects, using the same
400: estimates of FSI effects as for the mass (see text) and propagating
401: them to the width.  Note that the final W width results of the
402: experiments do not use the techniques introduced to reduce sensitivity
403: to FSI effects as used for the mass analysis.
404: 
405: A simultaneous fit to the results of the four LEP collaborations is
406: performed in the same way as for the $\Mw$ measurement. Correlated
407: systematic uncertainties are taken into account and the combination
408: gives:
409: \begin{eqnarray*}
410:       \Gw = 2.196\pm0.063(\mathrm{stat.})\pm0.055 (\mathrm{syst.})~\GeVm,
411: \end{eqnarray*}
412: with a $\chi^2$/d.o.f. of 37.4/33.
413: 
414: \section{Summary}
415: 
416: The results of the four LEP experiments on the mass and width of the W
417: boson are combined taking into account correlated systematic
418: uncertainties, giving:
419: \begin{eqnarray*}
420:       \Mw & = & 80.376\pm0.033~\GeVm, \\
421:        \Gw & = &  2.196\pm0.083~\GeVm.
422: \end{eqnarray*}
423: The statistical correlation between mass and width is small and
424: neglected.  Their correlation due to common systematic effects is
425: under study.
426: 
427: \clearpage
428: 
429: \begin{figure}[p]
430: \begin{center}
431: 
432: {ALEPH, DELPHI, L3, OPAL as published.}
433: 
434: \mbox{\epsfig{file=mw_ffff_pub.eps,width=0.495\textwidth}}
435: \hfill
436: \mbox{\epsfig{file=gw_ffff_pub.eps,width=0.495\textwidth}}
437: \vskip -1cm
438: \caption{\label{mw:fig:mwgw-pub} The combined results for the measurements
439:           of the W mass (left) and total W width (right) compared to
440:           the results obtained by the four LEP collaborations (as
441:           published). The combined values take into account
442:           correlations between experiments and years and hence, in
443:           general, do not give the same central value as a simple
444:           average.  The individual and combined $\Mw$ results include
445:           the measurements from the threshold cross section. }
446: \end{center}
447: \end{figure}
448: \vskip -1cm
449: \begin{figure}[p]
450: \begin{center}
451: 
452: {ALEPH, DELPHI, L3, OPAL as published.}
453: 
454: \mbox{\epsfig{file=mw_qqln_pub.eps,width=0.495\textwidth}}
455: \hfill
456: \mbox{\epsfig{file=mw_qqqq_pub.eps,width=0.495\textwidth}}
457: \vskip -1cm
458: \caption{\label{mw:fig-qqlnqqqq-pub} The W mass measurements from the
459:           $\WWqqln$ (left) and $\WWqqqq$ (right) channels obtained by
460:           the four LEP collaborations (as published) compared to the
461:           combined value. The combined values take into account
462:           correlations between experiments, years and the two
463:           channels.  The ALEPH and L3 $\qqln$ and $\qqqq$ results are
464:           correlated since they are obtained from a fit to both
465:           channels taking into account inter-channel correlations. }
466: \end{center}
467: \end{figure}
468: 
469: \clearpage
470: 
471: \begin{figure}[p]
472: \begin{center}
473: 
474: {ALEPH, DELPHI, L3, OPAL re-evaluated propagating the common LEP FSI estimates.}
475: 
476: \mbox{\epsfig{file=mw_ffff_com.eps,width=0.495\textwidth}}
477: \hfill
478: \mbox{\epsfig{file=gw_ffff_com.eps,width=0.495\textwidth}}
479: \vskip -1cm
480: \caption{\label{mw:fig:mwgw-com} The combined results for the
481:           measurements of the W mass (left) and total W width (right)
482:           compared to the results obtained by the four LEP
483:           collaborations, propagating the common LEP estimates of FSI
484:           effects to mass and width (see text). The combined values
485:           take into account correlations between experiments and years
486:           and hence, in general, do not give the same central value as
487:           a simple average.  The individual and combined $\Mw$ results
488:           include the measurements from the threshold cross section. }
489: \end{center}
490: \end{figure}
491: \vskip -1cm
492: \begin{figure}[p]
493: \begin{center}
494: 
495: {ALEPH, DELPHI, L3, OPAL re-evaluated propagating the common LEP FSI estimates.}
496: 
497: \mbox{\epsfig{file=mw_qqln_com.eps,width=0.495\textwidth}}
498: \hfill
499: \mbox{\epsfig{file=mw_qqqq_com.eps,width=0.495\textwidth}}
500: \vskip -1cm
501: \caption{\label{mw:fig-qqlnqqqq-com} The W mass measurements from the
502:           $\WWqqln$ (left) and $\WWqqqq$ (right) channels obtained by
503:           the four LEP collaborations, propagating the common LEP
504:           estimates of FSI effects to the mass. The combined values
505:           take into account correlations between experiments, years
506:           and the two channels.  The $\qqln$ and $\qqqq$ results are
507:           correlated since they are obtained from a fit to both
508:           channels taking into account inter-channel correlations. }
509: \end{center}
510: \end{figure}
511: