As I am revising the book I am writing on the history of the CDF experiment, I have bits and pieces of text that I decided to remove, but which retain some interest for some reason. Below I offer a clip which discusses the measurement of the natural width of the Z boson produced by CDF with Run 0 data in 1989. The natural width of a particle is a measurement of how undetermined is its rest mass, due to the very fast decay. The Z boson is in fact the shortest lived particle we know, and its width is of 2.5 GeV.
------
In late summer of 1989, no longer than a month after the SLAC workshop where CDF had presented their first measurement of the Z boson mass, the 8th Topical Conference on Proton-Antiproton Collider Physics was held in Castiglione della Pescaia, a nice seaside town on the Tyrrhenian sea. The conference was attended by many physicists from the CERN UA2 experiment, the only one of the two SppS experiments still running. They had been measuring the masses of W and Z bosons, and they were not welcoming of the CDF competition.
Among the pack of UA2 physicists who swarmed to Castiglione was a young Joe Incandela, the physicist who would 23 years later lead the CMS experiment to the discovery of the Higgs boson. Joe, who had been a Ph.D. student of Henry Frisch on a magnetic monopole search just a few years earlier, was a member of UA2 but was already planning to join the CDF collaboration (he would officially do so in 1991). In UA2 he was working at the Z mass measurement. As soon as CDF published its preprint, where the Z mass was determined with fivefold increase in accuracy over the previous CERN measurements, he devoured the article, trying to learn all that there was to learn about the competitors’ advanced techniques. And he soon discovered an inconsistency in the CDF measurement.
The CDF paper claimed that a combined fit of mass and width of the Z boson from 65 Z-->ee candidates in the 80-100 GeV mass range produced a width measurement of 3.6 GeV. This result, combined with the corresponding one from the Z-->μμ sample, yielded the final width measurement of 3.8 GeV. However, any physicist with some experience in statistical data analysis would have been surprised by looking at the Z-->ee mass distribution shown in figure 3 of the article. There, the peak appeared visibly narrower than what a 3.6 GeV natural width of the resonance would produce, once smeared out by the experimental resolution on electron energies that CDF was quoting.
Above: the mass distribution published by CDF with its Z-->ee candidates
Joe spent the month of August on his computer, generating pseudo-experiments to figure out what was wrong in the CDF determination of the Z width from the electron-positron decays. He tried different models for the energy spectrum of the detected electrons. The relative energy resolution of electrons would improve at higher energy, and the observed Z peak should consequently get narrower. This was practically the only way to justify the CDF result. However, it turned out to be impossible to reconcile the quoted width measurement with the distribution shown in the article. The measured value was six or seven standard deviations larger than what a correct fit of the histogram could conceivably give.
Incandela was quite unsure on how to proceed. Publishing a short paper with a rebuttal of the CDF measurement did not really look like the right thing to do: it was a clear belligerent act, by which he would certainly create the grounds for enmity with the physicists who would soon become his colleagues. Worse still, among the physicists involved in the CDF measurement was his ex-advisor Henry Frisch!
On the other hand, the ethical scientist must correct an erroneous published measurement. In consultation with his colleagues, Incandela decided to discuss directly with the CDF colleagues the calculations he had produced. During a pause in the conference works, Incandela met privately with the CDF members. To their credit, the CDF contingent acknowledged the flaws in their Z width measurement. CDF then assured their UA2 colleagues that they would publish an erratum themselves. However, after the conference they decided that the best way to proceed was to supersede the summer 1989 result by a new measurement. But that in fact never happened.
Comments