DISCOVERY of the TOP by BILL CARITHERS &

Two leaders of the experiments that isolated the tell the adventure of its discovery. Fermilab Media Services Visual

4 FALL 1995 ANKIND has sought the elementary building blocks of ever since the days of the Greek philosophers. Over Mtime, the quest has been successively refined from the original notion of indivisible “” as the fundamental elements to the present idea that objects called lie at the heart of all matter. So the recent news from Fermilab that the sixth—and possibly the last—of these quarks has finally been found may signal the end of one of our longest searches. Fermilab Media Services Visual

BEAM LINE 5 But the properties of this funda- combinations of three quarks or a CDF Collaboration mental constituent of matter are quark-antiquark pair. Quarks also bizarre and raise new questions. In seemed to form a counterpart to the Argonne National Laboratory particular, the of the top quark other class of elementary , Istituto Nazionale di Fisica Nucleare, is about forty times that of any the , which then included the University of Bologna µ Brandeis University other—a fact which suggests that per- (e) and ( ) (both with University of California at Los Angeles haps it plays a fundamental role in unit ) and their companion ν ν University of Chicago the question of how the mass of any chargeless , e and µ. The Duke University object arises. leptons do not feel the strong inter- Fermi National Accelerator Laboratory action, but they do participate in the Laboratori Nazionali di Frascati, Istituto N 1964 Murray Gell-Mann and electromagnetic interactions and the Nazionale di Fisica Nucleare Harvard University proposed the quark responsible for ra- Hiroshima University Ihypothesis to account for the ex- dioactive decays. They have the University of Illinois plosion of subatomic particles dis- same as the quarks and also Institute of Physics, McGill covered in accelerator and cosmic- have no discernible size or internal University and University of Toronto ray experiments during the 1950s and structure. The Johns Hopkins University early 1960s. Over a hundred new par- But most physicists were initially National Laboratory for High ticles, most of them strongly inter- reluctant to believe that quarks were Physics (KEK) Lawrence Berkeley Laboratory acting and very short-lived, had been anything more than convenient ab- Massachusetts Institute of Technology observed. These particles, called stractions aiding particle classifica- , are not elementary; they tion. The fractional electric charges Michigan State University possess a definite size and internal seemed bizarre, and experiments re- University of New Mexico structure, and most can be trans- peatedly failed to turn up any indi- Osaka City University formed from one state into another. vidual free quarks. And—as became Università di Padova, Instituto Nazionale di Fisica Nucleare The quark hypothesis suggested that apparent from studies of fundamen- University of Pennsylvania different combinations of three tal theories of quarks and leptons— University of Pittsburgh quarks—the up (u), down (d), and major conceptual problems arise if Istituto Nazionale di Fisica Nucleare, strange (s) quarks—and their an- the numbers of quarks and leptons University and Scuola Normale tiparticles could account for all of the are not the same. Superiore of Pisa hadrons then known. Each quark has Two major developments estab- Purdue University University of Rochester an intrinsic spin of 1/2 unit and is lished the reality of quarks during Rockefeller University presumed to be elementary, like the the 1970s. Fixed-target experiments Rutgers University electron. So far, quarks appear to directing high energy leptons at pro- Academia Sinica have no size or internal structure and tons and showed that these Texas A&M University thus represent the smallest known hadrons contain point-like internal Texas Tech University constituents of matter. To explain constituents whose charges and University of the observed spectrum of hadrons, spins are just what the quark mod- Tufts University University of Wisconsin quarks had to have electric charges el had predicted. And in 1974 ex- Yale University that are fractions of the electron periments at Brookhaven National charge. The u quark has charge 2/3 Laboratory in New York and Stan- while the d and s quarks have charges ford Linear Accelerator Center −1/3 (in units where the electron (SLAC) in California discovered a charge is −1). striking new at the then very The observed hadron spectrum large mass of 3.1 GeV—over three agreed remarkably well with the times that of the . This hadron expected states formed from (called the J/ψ after its separate

6 FALL 1995 Electric First Second Third Charge Family Family Family DØ Collaboration +2/3 up (u) (c) top (t) QUARKS Universidad de los Andes -1/3 down (d) strange (s) bottom (b) University of Arizona -1 electron (e) muon (µ) (τ) Brookhaven National Laboratory LEPTONS ν ν ν 0 electron ( e) ( µ) ( τ) Brown University University of California, Davis names in the two experiments) was chart above). Ordinary matter is com- University of California, Irvine found to be a of a new posed entirely of first-generation par- University of California, Riverside kind of quark, called charm or c, with ticles, namely the u and d quarks, LAFEX, Centro Brasileiro de Pesquisas its antiquark. The c quark has a plus the electron and its neutrino. Físicas Centro de Investigacion y de much greater mass than the first But the third-generation quark dou- Estudios Avanzados three, and its charge is 2/3. With two blet seemed to be missing its charge Columbia University quarks of each possible charge, a +2/3 member, whose existence was Delhi University could be established be- inferred from the existing pattern. In Fermi National Accelerator Laboratory tween the quarks and the leptons. advance of its sighting, physicists Florida State University Two pairs of each were then known: named it the top (t) quark. Thus be- University of Hawaii ν University of Illinois, Chicago (u,d) and (c,s) for quarks and (e, e) and gan a search that lasted almost µ ν Indiana University ( , µ) for leptons, satisfying theo- twenty years. Iowa State University retical constraints. Korea University But this symmetry was quickly SING THE RATIOS of the ob- Kyungsung University broken by unexpected discoveries. In served quark , some Lawrence Berkeley Laboratory 1976 experiments at SLAC turned up Uphysicists naively suggested University of Maryland a third charged , the tau lep- that the t might be about three times University of Michigan Michigan State University ton or τ. A year later at Fermi as heavy as the b, and thus expect- Moscow State University National Accelerator Laboratory in ed that the top would appear as a – University of Nebraska Illinois a new hadron was discovered heavy new hadron containing a tt New York University called the upsilon or ϒ, at the huge pair, at a mass around 30 GeV. The Northeastern University mass of about 10 GeV; like the J/ψ, it electron- then un- Northern Illinois University was soon found to be the bound state der construction (PEP at SLAC and Northwestern University University of Notre Dame of yet another new quark—the bot- PETRA at DESY) raced to capture the University of Panjab tom or b quark—and its . prize, but they found no hint of the Institute for High Energy Physics, Experiments at DESY in Germany top quark. Protvino and Cornell in New York showed In the early 1980s a new class of Purdue University that the b quark has spin 1/2 and a accelerator came into operation at Rice University charge of −1/3, just like the d and s CERN in Switzerland, in which University of Rochester quarks. counter-rotating beams of Commissariat à l’Energie Atomique, Saclay With these discoveries, and and collided with an en- Seoul National University through the development of the Stan- ergy of about 600 GeV. The protons State University of New York, dard Model, physicists now under- and antiprotons brought their con- Stony Brook stood that matter comes in two stituent quarks and antiquarks into Superconducting Super parallel but distinct classes—quarks collision with typical of 50 Laboratory and leptons. They occur in “gener- to 100 GeV, so the top quark search Tata Institute of Fundamental Research University of Texas, Arlington ations” of two related pairs with dif- could be extended considerably. Be- Texas A&M University fering —(+2/3, −1/3) sides the important discovery of the for quarks and (−1, 0) for leptons (see W and Z that act as carriers

BEAM LINE 7 of the unified electroweak , the rely on the production of separate t Production CERN experiments demonstrated an- and t¯ quarks from of in- other aspect of quarks. Though coming quarks and antiquarks in the

π+ quarks had continued to elude direct proton and , with subse- u – detection, they can be violently scat- quent decays into observable parti- d π– d tered in high energy collisions. The cles (see box on the right). The design u– Jet high energy quarks emerging from of DØ stressed recognition of the tell- + u K the collision region are subject to the tale leptons and jets over as large a s– as they leave the solid angle as possible. Meanwhile p s K– – scene of the collision, creating ad- CDF had installed a new vertex de- u d p– d ditional quark-antiquark pairs from tector of silicon microstrips near the u u– g the available collision energy (us- beams intended to detect short-lived d u– 2 – ing E = mc ). The quarks and anti- particles that survive long enough to d quarks so created combine into travel a millimeter or so from the in- ordinary hadrons that the experiment teraction point. This detector was COLLIDING PROTON (p) and can detect. These hadrons tend to particularly good at sensing the pres- − antiproton (p ) bring together cluster along the direction of the orig- ence of the b quarks characteristic of quarks uud and uud. In this inal quark, and are thus recorded as top decay. Another method of tagging A − example, a u and u scatter by a “jet” of rather collinear particles. b quarks, by detecting their decays exchanging a (g), the carrier Such quark jets, previously sensed at into energetic leptons, was used by of the strong . Like all SLAC and DESY, were clearly ob- both experiments. Thus the two ex- quarks, the scattered u quark pos- served at CERN and became a key in- periments, while searching for the sesses the strong interaction “charge” called color. The strong gredient in the next round of top same basic decay sequence, had attraction of the u quark color to the quark searches. rather complementary approaches. other quark color charges prevents With the advent of the CERN col- With the data from this 1992–93 it from escaping freely. Energy from lider, and in 1988 the more powerful run, progress accelerated. First, DØ the collision region is converted to 1800 GeV collider at Fermilab, the published a new lower limit of matter in the form of quark- search for the top quark turned to 131 GeV on the possible top mass antiquark pairs. The antiquark new avenues. At the large masses from the absence of events with the (with anti-color) joins with a quark – to produce a colorless which now accessible, the tt bound state characteristic dilepton or single lep- is free to escape. was unlikely to form and isolated top ton signatures. This paper turned out In this example, the primary quarks were expected. For masses be- to be the end of the line in excluding scattered u quark is accompanied low that of the W , W decay top masses. Up to then, with the − − − − – by (dd ), (uu ), (ss ), and (dd ) crea- into a t and b¯ could predominate. absence of an excess of candidate ted pairs, leading to the formation of − − Some indication of this process was events, the analyses had tended to re- π+, π , K +, and K , all trav- reported in 1984 by the CERN UA1 strict the event sample, so as to set eling in similar directions and form- ing a “jet” of particles that may be experiment, but it was later ruled out as high a limit on the mass as pos- detected in the experiment. (The re- by the CERN UA2 and Fermilab CDF sible. But by summer of 1993, CDF maining d quark will be joined to experiments. By 1990 CDF had ex- noticed that the top mass limits were one of the other spectator quarks in tended the top mass limit to 91 GeV, not improving with additional data, the collision.) thus eliminating the possibility for due to a growing handful of events W decay to top. that passed all preassigned criteria. Naive (in retrospect) estimates of N 1992, the DØ detector joined their statistical significance triggered CDF as a long run began. intense activity to review the results IFurther searches would have to and prepare a draft publication with

8 FALL 1995 Top Production and Decay

OR TOP QUARK MASSES above MW + Mb, top production Fproceeds mainly through anni- hilation of incoming quarks and anti- quarks from the beam particles.

qt the goal of publishing the results by CDF, a few physicists felt that no − gluon the October 1993 pp Workshop in publication should occur until the

Japan. As the analysis was too com- collaboration consensus was for “ob- –– plex for a single journal letter, CDF servation.” The other end of the spec- qt planned a series of four papers in trum felt that the data were already − Letters (PRL)—two in that category. Finally, CDF settled The q and q fuse briefly into a gluon, devoted to the counting experiments on a conservative interpretation and the carrier of the strong force, and − in the two search topologies; one de- was careful to disclose even those re- then rematerialize as t and t quarks scribing the kinematics of the events sults that contra-indicated top. traveling in roughly opposite direc- and giving an estimate of the top The pressure within CDF to pub- tions. The top quarks have a very −24 mass; and a fourth to bring it all to- lish increased enormously as the short lifetime (about 10 seconds) and decay almost always into a W gether with the overall significance, winter 1994 conferences approached. boson and a b quark (t → W +b; − − − conclusions, and top production In addition, pirated copies of the t → W b). The b quarks evolve into . In a heated argument Physical Review D manuscript were jets seen in the detectors. The W + − at the October collaboration meet- circulating widely and rumors (some and W have several decay possi- ing, it became clear both that the astoundingly accurate and others bilities: mass and kinematics sections need- comically off-base) were flying about ± → ±ν ed more work and that the “four the globe. The rumor mill churned W e (fraction = 1/9) → µ±ν (fraction = 1/9) PRL” format was not working well. once again when CDF withdrew its → τ±ν (fraction = 1/9) − CDF decided instead to prepare a sin- top talks at the LaThuile and → qq (fraction = 2/3). gle long Physical Review article, to Moriond Conferences. An April CDF − concentrate on the remaining holes Collaboration meeting was the wa- The q and q from W decays appear in the analysis, and to forego any pub- tershed event, as the group ham- as jets. The final states containing lic discussion of new results. mered out the final wordings. After τ’s are difficult to isolate and were By the next CDF collaboration a virtually unanimous agreement on not sought in the experiments. The final state with both W ’s decaying meeting in late January, the count- the final draft, the collaboration − into qq , though relatively copious, ing experiments were complete, the spokespersons finally notified Fer- are buried in huge strong interaction mass and kinematics analyses had milab Director John Peoples of their backgrounds. The remaining de- made real progress, and the single intent to publish. cays, studied by both CDF and D¯, long draft was in reasonable shape. This CDF analysis found 12 can- can be categorized by the number The big remaining issues were the didate events that could not be eas- leptons (l = e or µ) from the W de- wordings of the title and conclusions ily explained on the basis of known cays: Dilepton channel: section and the “between the lines” backgrounds (which were predicted − − − − − tt →W +bW b → l +l ννbb message that they sent to the com- to be about six events). The odds for Single lepton channel: − − − − − munity. Traditionally, scientific pub- background fluctuations to yield the tt →W +bW b → l+νqq bb − − − − lications about new phenomena use observed events were low—about 1 or l ν qq bb certain key phrases: “Observation in 400—but not small enough to war- of” indicates a discovery when the rant the conclusions that the top had The dilepton channels are relatively case is unassailable; “Evidence for” been found. Under the hypothesis free from background but comprise − − means the authors believe they are that the data did include some tt only 5 percent of all tt decays. The seeing a signal but the case is not events, the mass of the top was es- single lepton channels give larger yields (30 percent of all decays) but iron-clad; and “Search for” implies timated to be 175 GeV±20 GeV. The − have substantial backgrounds from the evidence is weak or non-existent. cross section for tt production was W + jets production processes which Within every collaboration, there is determined to be about 13.9 pico- must be reduced. a wide spectrum of comfort levels in barns, larger than the theoretical pre- claiming a new result. On one end of diction of 3–7 picobarns.

BEAM LINE 9 DØ used the period while CDF was tuning each of the separate elements Top Mass Distribution preparing its publication to re- of the process, but until summer 6 optimize its own earlier analysis, 1994 the intensity of the collider was DO focusing on higher mass top. By April disappointing. During a brief mid- ) 2 1994, DØ was also in a position to summer break, however, one of the 4 show some of its new results, which Tevatron magnets was found to have were subsequently updated in the been inadvertently rotated. With this summer Glasgow conference and problem fixed, beam intensities rose 2 published in Physical Review Let- immediately by a factor of 2. With

Events/(20 GeV/c ters. DØ also had a small excess of the now good understanding of the top-like events but with smaller sta- accelerator, a further doubling of the 0 tistical significance; its analysis was event rate was accomplished by 100 200 based upon the dilepton and single- spring 1995. In a very real sense, the Fitted Mass (GeV/c2) lepton channels with a combination final success of CDF and DØ in dis- 5 of lepton tags and topological sup- covering top rested upon the superb CDF pression techniques to reduce back- achievements of the Fermilab Ac-

) 4 2 ground. Nine events were observed, celerator Division. The improved op-

3 compared with an expected back- erations meant that the data samples ground of about four, giving odds for accumulated by early 1995 were ap- 2 a background fluctuation of about proximately three times larger than 1 in 40. The excess events corre- those used in the previous analyses, Events/(10 GeV/c 1 sponded to a cross section of 8.2±5.1 and both experiments were now − picobarns. The expected yield of tt poised to capitalize on the increase. 0 80 120 160 200 240 280 events was virtually the same for DØ By December, both collaborations Reconstructed Mass (GeV/c2) and CDF. Taken together, these re- realized that the data now on tape The mass of the top quark can be sults from CDF and DØ were not suf- should be enough for a discovery, if reconstructed from the energies and ficient to establish conclusively the the earlier event excess had been ap- directions of its decay products as existence of the top quark. proximately correct. In fact, the ex- measured in the detectors using the The final chapter in finding the periments do not keep daily tallies conservation laws for energy and momentum. Since the top quark has top quark began with the resumption of the number of events in their sam- a unique mass, the data (indicated of the collider run in late summer ples. The physicists prefer to refine by the black histograms) should 1994. The performance of the Teva- their analysis techniques and selec- show a “peak” in the reconstructed tron was the key to the success. The tion parameters in order to optimize distribution. The non-top back- ground (the red dashed curve for Tevatron involves a collection of sev- the analysis on simulated events be- DØ and the red dotted curve for en separate accelerators with a com- fore “peeking” at the data. This ret- CDF) has very different shapes. A plex web of connecting beam lines. icence to check too often on the real simulation is required to provide the Many technical gymnastics are re- data stems from the desire to avoid correspondence between the mea- sured jet energies and the parent quired to accelerate protons, produce biasing the analysis by the idiosyn- quark momenta. The red dotted secondary beams of antiprotons from crasies of the few events actually curve for DØ shows the expected an external target, accumulate and found. At the beginning of January, contribution from top for the best fit store the intense antiproton beams, DØ showed a partial update in the value of the top mass. The solid red curve shows what a simulated top and finally inject the counter-rotat- Aspen Conference using some new quark mass distribution would look ing beams of protons and antiprotons data but retaining previous selection like when added to the background, into the Tevatron for acceleration to criteria; these results had increased and these curves should be com- 900 GeV. Enormous effort had been significance, with only a 1 in 150 pared to the actual data. poured into understanding and chance of background fluctuations.

10 FALL 1995 Its simulations showed that for the before. (Los Angeles Times reporter, large-mass top now being sought, K. C. Cole, called a distinguished Fer- World Wide Web Statistics even better significance could be ob- milab physicist to get an explanation tained. Recognizing that the statis- of statistical evidence presented in OTH THE CDF and D¯ World tics were nearly sufficient, the col- the O.J. Simpson trial. The physicist Wide Web pages can be laboration began working in high used, as illustration, “the recent sta- Baccessed from the Fermilab gear to finalize the selection crite- tistical evidence on the top quark home page at http://www.fnal.gov/ ria and obtain the last slug of data be- from CDF and DØ,” and Cole swift- (click there for top quark discovery). fore a late-January Tevatron shut- ly picked up the chase.) Since the announcement of the dis- down. In mid-January, CDF updated In its paper, CDF reported finding covery of the top quark last March, all of its data; with all of the pieces six dilepton events plus 43 single- the CDF and D¯ home pages have assembled side by side, it became lepton events (see box on the next received over 100,000 hits each. clear that the collaboration had the page for more details); it concluded There were 2279 hits on the CDF necessary significance in hand for that the odds were only one in a mil- page alone on the day after the claiming a discovery. The CDF lion that background fluctuations announcement. Six months after the process was now considerably could account for these events. DØ, discovery, the Fermilab top quark streamlined by the experience of pro- in its paper, observed three dilepton page is receiving about 200 hits a ducing the earlier top quark papers. events plus 14 single-lepton events day. By February, both CDF and DØ rec- and concluded that the odds were ognized that convergence was im- two in a million that these could minent. Both worked around the have been caused by backgrounds. clock to complete all aspects of the The top quark masses reported by the analyses. Even though each collab- two experiments were 176±13 GeV oration knew that the other was clos- for CDF and 199±30 GeV for DØ. And ing fast and in the process of writing both experiments gave consistent re- its paper, there were still no formal sults, although somewhat below − exchanges of results, nor of intend- CDF’s earlier value for the tt pro- ed timing. Some copies did leak duction cross section (see box on the out—for example, drafts left on print- next page for more detailed infor- ers in institutions with physicists on mation). both collaborations and computer The top quark appears to be a hacking from one collaboration to point-like particle; it has no internal another. The one avenue of cross- structure that we can discern. As one fertilization that seems not to have of the six fundamental constituents operated was through couples with of matter, it has properties very sim- one person in each group! ilar to the up and charm quarks, with On February 24, CDF and DØ the exception of its remarkable mas- “Observation” papers were submit- siveness and its very short lifetime. ted to (PRL); The top quark is about 200 times public seminars were scheduled at more massive than the proton, about Fermilab for March 2. By agreement, forty times that of the second heav- the news of the discovery was not to iest quark (the b), and roughly the be made available to the physics same as the entire mass of the community or news media until the nucleus! Surely this striking obesity day of the seminars. In spite of all ef- holds an important clue about how forts, word did leak out a few days mass originates.

BEAM LINE 11 CDF AND DØ RESULTS

HE RESULTS FROM THE TWO COLLABORATIONS were remarkably similar. CDF found 6 dilepton events Jet #2 CDF Twith a background of 1.3; 21 single-lepton events in Jet #3 which 27 cases of a b quark tag by the vertex detector (with 6.7 background tags expected); and 22 single-lepton events with 23 cases of a b tag through leptonic decay (with 15.4 background tags expected). D¯ found 3 dilepton events (0.65 background events); 8 single-lepton events Jet #1 with topological tagging (1.9 background events); and 6 sin- gle-lepton events with b-to-lepton tags (1.2 background events). A particularly striking example of a dilepton event Jet #4 with very energetic electron, muon, and missing ET (due to the neutrinos), plus two jets, is shown below from the D¯ + data. The plot shows the detector unfolded on to a plane, e with the energy of the various objects indicated by the height of the bars. This event has a very low probability to be explained by any known background. The probability 3 meters that background fluctuations could explain the observed signal was one-in-a-million for CDF and two-in-a-million for D¯—sufficiently solid that each experiment was able to claim the observation of the top independently.

DO Missing ET 100 Electron Muon 10 Primary Secondary

JET ET (GeV) Vertex 1 Vertex JET –3.7

0 5 millimeters 6.4 η by the need to identify the correct combination of jets with 3.2 φ 3.7 parent quarks in the decay and to accommodate the ten- 0.1 dency of the strong interaction to generate additional jets. Additional studies helped to establish that the new signal The two experiments obtained consistent results for this was indeed the top quark. Both experiments were able to mass measurement: 176±13 GeV for CDF and 199±30 GeV show that the candidate events were consistent with the D¯ for . − expected presence of b quarks. The single-lepton channel The rate for observing tt pairs is controlled by the strong- − events should have one W boson that decays to a qq final interaction production cross section. Each experiment evalu- state, and the data showed the expected enhancement in ated this cross section at their mean mass value; CDF +3.6 the di-jet mass. obtained 6.8-2.4 picobarns, and D¯ obtained 6.4±2.2 pico- Finally, both experiments made a measurement of the barns. These values are consistent with the theoretical pre- top quark mass, using the single-lepton events. In this diction within the combined experimental and theoretical study, the missing neutrino from one W decay must be error, although the tendency for the measurements to fall inferred and the mass of the parent top quark deduced from above theory is interesting and if it persists could indicate a constrained-fit procedure. This mass fitting is complicated that new physics is at play.

12 FALL 1995 ODERN HIGH energy phys- ics experiments are unique Minstitutions; though large enough to require formal organiza- tions and governance procedures, they remain voluntary collaborations of individual scientists who have banded together to permit achieve- ments that a few physicists could not accomplish by themselves. But lack- ing the usual structures of large organizations—the ability to hire and fire, or to rigorously assign jobs to in- Fermilab Media Services Visual dividual workers—they rely to a very must be given to optimizing the Left to right: Giorgio Bellettini, Paul large degree on the existence of a event selection process (only ten in- Grannis, John Peoples, Hugh shared purpose and consensus for teractions out of every million can Montgomery, and Bill Carithers following their constructive operation. be kept for analysis) and to improv- the seminars announcing the simulta- neous discovery of the top quark at Although about 430 physicists ing the software algorithms for se- Fermilab by the DØ and CDF collabora- participate in each collaboration, the lecting particles in the data. Work tions. Grannis and Montgomery are the number active in the top studies is must continue to improve the over- co-spokesmen for the DØ Collaboration, much less—roughly 50 directly en- all computing and software environ- John Peoples is the Director of Fermilab, gaged at a significant level. This ment. New detector development and Carithers and Bellettini are the co- results from the broad scope of col- is conducted with an eye to future spokesmen for the CDF Collaboration. lider experiments; the range of tasks upgrades, and major design and build- The collaborations presented their needing attention is very large. The ing projects are underway for future results at seminars held at Fermilab on March 2, 1995. physics research being done by the runs. The experiments are so big that collaborations is widely dispersed reviews by the collaborations them- over five major areas (top search, selves, the Laboratory, and external bottom-quark physics, strong- committees continually occur. interaction studies, electroweak bo- Finally, with collaborations of these son measurements, and searches for sizes, group psychologists, financial new phenomena). Within these analysts, and tea-leaf readers are physics groups there are perhaps 50 needed! active analyses under study at any The techniques for the top quark given time. Often these other analy- search and measurement of its mass ses contributed synergistic tech- developed over several years. For niques to aid the top search; for ex- most important aspects (and some ample, studies of jets by the minor ones) several different ap- strong-interaction group entered proaches were developed by rival sub- directly into measurements of the groups or individuals. This multi- top mass. plicity is valuable for cross-checking A diversity of tasks occupy the results and as a tool for natural se- members of the collaborations. In ad- lection of the strongest methods, but dition to running the experiment such competitiveness and rivalry can around the clock and attending to its also be debilitating. In some cases, maintenance, constant attention the winner in these rivalries can be

BEAM LINE 13 established on purely technical for internal use, proliferate and are felt the heat of competition; for each, grounds—but often the basis for de- disseminated electronically across no doubt, the dream of being first cision is clouded, as the alternatives the globe. Electronic mail is an es- to recognize the top was a powerful have both pros and cons. Adjudicat- sential component; colleagues work- incentive to be rather closed- ing among these contenders is a dif- ing a continent apart on related prob- mouthed about innovations and ficult and time-consuming problem. lems can conduct an almost on-line progress. In some cases, the choice is made dialog. Discussion and argument over The history of the top search over through a survival-of-the-fittest fine points of interpretation can gen- the final two years shows evidence process—the approach first brought erate a flurry of productive messages of this competitiveness and friend- to full maturity, with documented among five or six experts. At the oth- ly rivalry at work. As the more se- error analyses and cross-checks, is er end of the scale, there remains an nior collaboration, CDF had devel- chosen. Consensus plays an impor- essential role for formal presenta- oped its techniques over time to a tant role in rejecting some alter- tions of work to the full physics relatively refined state. It started the natives; experts not involved in a group and full collaboration at reg- race for top with the UA2 experiment particular question can wield con- ular intervals. Repetition is usually in 1988 and had been denied success siderable influence in rejecting some necessary to bring the non-experts to only by top’s extraordinary mas- weaker approaches. When the issues a reasonable degree of familiarity siveness. CDF would have been dis- have sufficiently crystallized, special with the issues and methods, and appointed to lose the race to the new- advisory panels drawn from the rest to build acceptance of new work. er DØ experiment. On the other of the collaboration can be very use- To what extent did CDF and DØ hand, the Tevatron performance in ful in helping to define the direction. interact during the time of the top the 1992–1994 era improved so much The last resort of turning the issue search and discovery? At a formal that the data in hand was almost all over to the “managers”—the spokes- level, very little. New results from acquired after DØ’s start-up, so the persons or physics group conveners— each collaboration were periodically two experiments could work on al- is successful primarily when the shown at conferences and public pre- most equivalent data sets. DØ, with rivals have agreed that they are un- sentations, so the broad outlines of its strengths of excellent lepton iden- able to resolve their differences. For- each group’s approach were known, tification and full solid angle cover- tunately, this rarely happens. but beyond that there were no direct age, was able to develop powerful Effective communication on interactions. The grapevine was of tools for suppressing unwanted back- many levels is an essential ingredi- course active, with lunch-table con- ground; it was of course determined ent of successful performance in the versations among individuals, but to make its mark as the brash new- large collaborations. For the top much of the information exchanged comer by converging on the top. But studies, weekly meetings of a half in this way was the anecdotal ac- the competition remained friendly dozen subgroups of the top analysis count of the day and was often out- and supportive throughout; it also discuss problems and status; as many moded by the next. There were prob- promoted a of work and depth additional meetings occur on relat- ably several reasons for this relative of understanding that would have ed analysis topics such as electron isolation of two groups working to- been difficult otherwise. And the ul- identification; and a more general ward a common goal. At one level timate scientific value of two inde- meeting brings together all the both groups realized that the inde- pendent results was enormous. threads. Video conferencing and elec- pendence had real scientific value. How do hundreds of authors agree tronically posted minutes of the Guarding against making a discov- on a single paper in a timely fashion? meetings help keep those not at Fer- ery by subconsciously shaping the This too is a sociological challenge! milab involved, although most ac- analysis is an ever-present worry; by The process for writing the final pa- tive participants travel there fre- keeping the walls between CDF and pers began in late January. DØ ap- quently. Informal documents pre- DØ fairly high, this possibility could pointed primary authors—one for a pared on sub-topics, intended only be minimized. Also, the two groups short letter paper and one for a longer

14 FALL 1995 paper ultimately not used. The top group designated a few individuals as its authorship committee, bring- ing the necessary range of expertise to the authors. The collaboration re- view was conducted in large part through an appointed editorial board of ten physicists including authors, drawn from widely diverse portions of the collaboration. The editorial board spent days carefully reviewing the backup documentation, probing the analysis for consistency and correctness. It also helped to revise the paper’s language to make it ac- cessible to the widest possible au- dience. The completed draft was dis- tributed to the full collaboration by Fermilab Media Services Visual electronic mail, and comments were the basic experimental numbers. Happy and relieved physicists who have solicited from all. Well over a hun- Others favored a more interpretative completed PhD theses using the DØ dred written comments were re- approach. Finalization of the paper experiment assemble in the DØ main control ceived (plus many more verbal ones), itself required the resolution of these room. Students and postdoctoral candidates are the lifeblood of the experiment and have all of which were addressed by the oft-conflicting outlooks on what made contributions to every aspect of it. board. The revised paper was again makes good science. Pictured are Brad Abbott, Purdue; Jeff Bantly, circulated for concurrence of the col- In the end, the chief necessity for Srini Rajagopalan, Northwestern; Dhiman laboration before submission to PRL. the convergence on the top discov- Chakraborty, Terry Heuring, Jim Cochran, Greg CDF appointed a chief author to be ery was the willingness of a collab- Landsberg, Marc Paterno, Scott Snyder, Joey assisted by a group advising on the oration to abide by a majority view. Thompson, Jaehoon Yu, Stony Brook; Regina sections of their expertise. Simulta- Securing this willingness requires ex- Demina, Northeastern; Daniel Elvira, Cecilia Gerber, University of Buenos Aires; Bob Hirosky, neously, a review committee (dubbed tensive attention to the process—of Gordon Watts (CDF), Rochester; Jon Kotcher, a “godparent” committee within being sure that all shades of opinion, New York University; Brent May, Arizona; Doug CDF) was appointed. The draft and reservations, and alternate view- Norman, Maryland; Myungyun Pang, Iowa State. responses to questions were always points are fully heard and under- Others who have completed theses on accessible to the collaboration elec- stood. It is more important perhaps DØ but not pictured are Rich Astur, Bo Pi, Sal tronically, both on the World Wide that each point of view is carefully Fahey, Michigan State; Balamurali V., Notre Dame; Web (with security) and via ordinary listened to than that it be heeded. Ties Behnke, John Jiang, Domenic Pizzuto, Paul files on the Fermilab computer. A fine line in resolving these view- Rubinov, Stony Brook; John Borders, Sarah Durston, Rochester; Geary Eppley, Rice; Fabrice In the final convergence to a re- points must be drawn between autoc- Feinstein, Alain Pluquet, University of Paris Sud; sult, it was necessary to come to grips racy and grass-roots democracy. The Terry Geld, Michigan; Mark Goforth, Robert with rather large differences of sci- process must have the confidence of Madden, Florida State; Ray Hall, Thorsten Huehn, entific style among the collaborators. the collaboration, or its general University of California, Riverside; Hossain Johari, Some could find in any result, no effectiveness can diminish rapidly. Northeastern; Guilherme Lima, LAFEX/CBPF; matter how thoroughly cross- What did the physicists of CDF Andrew Milder, Alex Smith, Arizona; Chris Murphy, checked, the need for further inves- and DØ feel on completion of the top Indiana; Haowei Xu, Brown; Qiang Zhu, New York University. tigation. Some were wary of any discovery experiments? Certainly re- interpretation beyond a statement of lief that the quest was complete. Also

BEAM LINE 15 pride that such an elusive quarry had celebrations, but by a week after the been trapped and that the two inde- public announcements it was back pendent experiments had drawn so to business, both on the top studies similar a profile of the top quark. Vir- and on the dozens of other analyses tually all in both collaborations could in progress. take justifiable pride for their per- sonal contributions to the discov- OES THE DISCOVERY of the ery—an effective electronics con- top quark close the chapter troller for the high-voltage system Don our understanding of the was as essential for the discovery as fundamental building blocks of mat- devising the selection criteria for the ter? Surely not—it is truly just the top quark events. Despite the many beginning. Though we now have strains introduced by the frenetic found the long-awaited sixth quark, pace, the process of discovering the with properties as predicted, much top was for most the scientific event more accurate measurements are of a lifetime. Graduate students, post- needed. Dozens of questions about docs, junior and senior faculty, and how the top is produced and how it scientists worked together on equal decays are clamoring for answers. footing, with the contributions of the Does the extraordinarily large top youngest as significant as those from quark mass hold clues to the nature the oldsters. Though for most scien- of symmetry breaking or to the ori- tists, solving the many small daily gin of mass? Do the regularities ob- problems gives sufficient satisfaction served for the lighter quarks hold for to keep them hooked on physics, the top? Are there signs of new par- reaching a milestone touted as a ma- ticles that decay into the top? Or are jor discovery was a special thrill not there exotic new objects to be found to be forgotten. in top decays? The very massive top After the public announcements is unique among quarks in that it of the discovery, a new challenge does not live long enough to be arose in interacting with the news bound into hadrons, so it provides media, representatives of member in- a testing ground for the study of a sin- stitutions, and other scientists to ex- gle “bare” quark. As happens so of- plain and comment on the results. ten in physics, the latest great dis- The media blitz caused strains of its covery should serve as the crucial own. As each physicist had con- tool for the next advance in under- tributed in some major way to the standing the composition of matter. discovery, most wanted to share in the public limelight. In the unfa- miliar world of public exposure, some strains and disappointments emerged, but the opportunity to ex- plain the significance of the work to the general public was for most a rewarding and educaional experience. How long did the euphoria of the discovery last? There were indeed

16 FALL 1995