"Great Gravity" is featured every edition of God & Nature Magazine, and tells the story of BNL physicist Bill Morse's journey through the world of muons and quarks, colliders and bubble chambers, with the heart of a committed Catholic and longtime-teacher of Sunday school. Read the first post in this series here.
Running the Data
New particles, created from collisions, leave tracks in a bubble chamber
by Bill Morse
We were still waiting for the NAL construction to be completed. Virgil Barnes and I decided to re-analyze some existing bubble chamber data taken at Brookhaven National Lab. The bubble positions had already been digitalized and were on magnetic tape. The computer program to combine the three camera views into the three dimensional space points of the particles had already been run.
The program we were using was called the “Three View Geometry Program.” We called it "TP." It was the forerunner to three-dimensional medical imaging computer codes, but of course we didn't know that in the early 1970s. The next program tried different hypothesis for which particles are produced in an interaction. This program was called SQUAW. I asked the senior grad student what SQUAW stood for. He said he didn't know. I said maybe after you have a TP, you need a SQUAW. He was not amused. However, when I became the senior grad student and the new grad student would ask why it's named SQUAW, I would tell them "After you have a TP, you need a SQUAW." They were always very happy with that answer.
Anyway, I was signed up for the midnight to 8am shift on the computer to re-run the fifty magnetic tapes with the SQUAW program. Actually, there were fifty tapes for the full energy runs, and one tape for a run at a slightly lower energy. The senior grad student set up the SQUAW program to run at midnight, and I started loading the tapes. It went slower than I thought it would, but when he came in at 8am, I was just finishing my fiftieth tape, and I told him I was done. I was tired, and went to bed. That afternoon I told Virgil that I had finished all the tapes. He said that was great to finish fifty-one in one night. I had forgotten the lower energy tape. However, I said, yes—I’d done all fifty-one tapes. The answer was out of my mouth before I had time to think about it.
I started the analysis. It was great; I found that I loved analyzing data. However, about once a month when I would show Virgil some plots, he would ask if the lower energy data was in the plot. I would say yes, but then he would say that it couldn't be, otherwise the plot should look different. I would say that I left the lower energy data out of that plot. Finally, I just told him the truth: "Actually, I forgot to run SQUAW on that tape, but I didn't want to admit it, because I was so proud of having done all the tapes in one night." He said, "Oh, well, it probably doesn't matter." I learned from that experience, not to lie! It's just way too much work remembering your lies; it ain't worth it! Actually, it's a pleasure to tell the truth. The Bible says to do the right thing, and then trust in the Lord for the outcome.
When I much later became the resident spokesman of E821, I was going over the project cost estimate with our chief engineer. It was clear that we had under-estimated the total project cost (TPC) by two million dollars, and we had a review coming up. I told the engineer that if we said we had under-estimated the TPC by two millions dollars, they might cancel the project, so we would tell them one million dollars this review, and one million dollars the next time. As I was driving home, I remembered when at Purdue I had decided not to lie again. The next morning I told the chief engineer that we would tell the truth, and if I ever told him to lie again, he should just tell me no. I could see that a huge load was lifted off his shoulders! The review committee was quite upset when we told them our TPC had gone up by two million dollars since the last review. However, at the review close-out, they said they realized that we now had a bottoms-up cost estimate, which is how we realized the TPC was too low, and they now had confidence in our project, and they congratulated us. Telling the truth is not only the right thing to do, it turns out that it’s also beneficial. My father was right when he told me, “Honesty is the best policy.”
Finally NAL was ready to deliver beam to our experiment, and we began taking data. 2B was to be after all. However, our run coincided with one of the gas crises of the 1970s. We would get our Purdue University car with just enough gas to get us to NAL. Then it was the job of the night shift to get in the service station line to get gas. However, often the station would give only several gallons of gas per customer, and sometimes they would shut down before we had even made it to the pump. Once, we were driving back to our motel when the post-doc who was driving noticed they were constructing a road which would cut several miles off the trip to the motel. It was late at night, and he decided to try it. It was also raining. Naturally, we got stuck in the mud. I had to get out and push! We ended up using much more gas than we would have by driving around, and I was very wet and muddy.
Usually, NAL closed down for a Christmas break. However, since our experiment had been delayed, they decided to continue the run over Christmas. The professors and post-docs all had small children, so I told them that I would take 36 hours of shifts, and they should all go home for Christmas. Boy was I beat when they showed up again! Anyway, finally we had our bubble chamber film, and took it back to Purdue to start measuring the bubbles, and then for the fun part, analyzing the data. Maybe we would discover quarks, or something very exciting!
In 1973, a narrow resonance (an extremely short lived particle) was discovered at about three times the proton mass at both BNL and SLAC. Professor Shelly Glashow from Harvard University came to Purdue to give a seminar. He said charm—a new quark—had been discovered, and we should look for a K-pion resonance in our bubble chamber data. We asked how many events should be in the resonance. He said about the same as strangeness, maybe a little less. We had lots of strangeness events, so we began looking.
Before I tell you what we found, let me give you some background into the "family problem" of elementary particle physics. In the 1930s everything seemed simple: the proton and neutron were in an isospin doublet, (which just describes the symmetry that these are two identical particles, except that one has electrical charge, and the other doesn’t—and actually, it is a broken symmetry, but I won’t go into that here) as were the electron and neutrino. However, in the 1940s, the “muon” was discovered. The muon is like the electron, only heavier by a factor of about two hundred. When Professor I. I. Rabi of Columbia University was told about the muon, he asked "Who ordered that?" The muon became the second family of leptons. The muon decays to an electron and two neutrinos by the weak interaction, kind of like the neutron decaying to a proton, electron, and neutrino. Actually, it’s an anti-neutrino, and you may ask how the heck do we tell a neutrino from an anti-neutrino? That would be an excellent question, and there are experiments going on now trying to give the answer.
Then in the late 1940s, they started seeing some odd events when one proton hit another proton. Many pions (yet another short-lived subatomic particle) were produced; however, occasionally there was a gap of several inches, and then another two particles would appear. This turned out to be a heavy version of the neutron, which decayed after several inches to a proton and a pion, for example. Also, strange versions of the pion were discovered: the K meson. These K mesons also traveled many inches, and then decayed into two pions, for example. In order to restore the symmetry, it was proposed that a heavy version of the proton should also exist. Glashow and others called it "charm." I don't know why they called it charm, but when a grad student asks me, I tell them that it would restore the symmetry like a charm, ie. then there would be a complete second family: charm with strange, and the muon with the muon-type neutrino, as opposed to the electron-type neutrino. Sound complicated? It is. This is why we call it the "family problem."
Let me now get back to the story. We decided to pause in the analysis of the number of pions produced, and start a crash analysis program for this K-pion resonance. The narrow resonance discovered at BNL and SLAC decayed to an electron and an anti-electron, which was relatively easy to find, because proton-proton interactions produced mostly pions. This meant that the BNL and SLAC experiments had discovered "hidden charm," ie. a charm anti-charm resonance. We would be looking for "bare charm," and this would have a much larger background. The charm anti-charm resonance, named the J resonance by the BNL experiment, and the psi resonance by the SLAC experiment, and ever since simply called the J-psi, has a mass about three times the proton mass. The bare charm mass should be a little over about one half the J-psi mass. I came in over the weekend to run the programs, and there staring up at me from the plot was a 2.6 sigma bump at about the predicted mass!
Let me tell you what a sigma is. Suppose you flipped a coin one hundred times. How many heads would you expect? If you said fifty, you are right! However, suppose you came up with fifty-seven heads, after flipping the coin one hundred times. Would that be unusual? That would be one sigma away from the expected value of fifty, and statistics theory says that your result would be that far away from the expected result about one third of the time, so, in other words, it's no big deal. However, if you came up with seventy-eight heads out of one hundred coin tosses; that would be 2.6 sigma away from the expected result. This is much more improbable. To see how improbable, just try it and see how many coin toss "experiments" you have to do to get 2.6 sigma deviation. However, most physicists would say you need at least a three sigma deviation to claim a "statistically significant" deviation from the "null hypothesis." Vernon Hughes always said you needed at least a five sigma deviation.
On Monday I showed the plot to Virgil. He immediately called a meeting, reviewed what Shelly had predicted, and then asked me to show the plot. There was a lot of excitement! Maybe we had discovered bare charm. We all agreed that 2.6 sigma was not statistically significant enough for an announcement. However, this was from only half of our bubble chamber film. If the effect was real, the new data should also have a 2.6 sigma effect. The probability of two 2.6 sigma events in a row is, well, you'll be tossing coins all day before you see something like that. In other words, it's something you have to take seriously. Our experiment 2B was measuring the bubbles, and also Earl Fowler's SLAC experiment that I had gone to California to work on several years earlier. It was agreed that this result was important enough that we would get two of Earl Fowler's experiment bubble measuring machines for a month, along with our machines.
After a week of measuring, I made the new plots. The plot from our machines did not have a bump, the spectrum was just flat, but the plot from our two "borrowed" machines had a hole at the mass where we had previously found a bump! I thought maybe there was something wrong with their machines, and I made many plots of residuals for all the machines, but there was no difference. At the end, the new data had a hole, not a bump, and our effect was declared to be a statistical fluctuation. Years later, bare charm was discovered, but Shelly had vastly underestimated the difficulties in finding it. Our bubble chamber experiment had accumulated about ten interactions per second during our run. It was finally found by electronic experiments which could accumulate thousands of interactions per second. I went back to analyzing the number of pions produced.
Read the next post in the series here.