In 1967, a group of Italian physicists led by Edoardo Amaldi came to the Daresbury Nuclear Physics Laboratory to propose an experiment and a collaboration. Amaldi was already a legend. He'd played a major role in founding Cern, the European Organisation for Nuclear Research, and was a powerful voice in European science. Daresbury's lab and its 5 GeV electron synchrotron were only a few years old, and in the absence of a lecture theatre, everyone packed into the canteen to hear what he had to say.

Introduced by our director, Alec Merrison, Amaldi outlined what he called a simple but important experiment - the electro-production of pi+ mesons (pions) at threshold. Particle physics is magical, of course, and one can create new things out of pure energy. Slamming a sledgehammer on to an anvil may not create a spanner, but similar things happen every day in particle physics. Since the pion decays in about 10-8 seconds, the proposal was to detect the scattered electron and the neutron. Unfortunately, the neutron carries no electric charge, so one must employ indirect techniques to spot it. The usual way is to use plastic scintillators. Neutrons colliding with protons in the scintillator (protons have an electric charge) create tiny flashes of light that photomultipliers can pick up. By measuring the pion reaction when it's just energetically possible - at threshold - one can extract valuable information. In particular, we could measure the so-called axial-vector form factor, at that time an unknown quantity. It should all have been very simple.

Merrison was keen for the collaboration to go ahead, and I was eager to join the lively group of Italians. However, science politics intervened and I was unable to join them. It's a long story, but two years later, the highly attractive collaboration had barely advanced, and indeed was about to collapse in acrimony. Merrison was not pleased. He asked if I'd be interested in trying to get it back on course, and I leapt at the chance. One problem was the theory behind the experiment. The UK theorists disagreed with their Italian counterparts about precisely what we'd be measuring, and we wouldn't get expensive accelerator time unless we could sort it out.

I asked when the two groups had last met, and received only blank stares. Astonishingly, they'd never met, although they had corresponded extensively. So I immediately arranged a working weekend in Trieste for the theorists to resolve their differences. We started at 10am on a Saturday for what I anticipated would be a long, hard theoretical slog. Claudio Verzegnassi wrote an equation on the blackboard, which Sandy Donnachie immediately queried. My heart sank as I feared we were in for some heavy-duty nitpicking. I was wrong. The theorists gathered at the board as if something spectacular had just happened, and there was a lot of arm-waving. Some minutes later, they were all smiles. It seemed that the warring camps had unknowingly been using different notations, but once that was clarified, they were in complete agreement!

ÖйúAƬ

We could hardly contain ourselves. The path to the new experiment was now open, and we could celebrate in full-blooded physicists' style with lots of wine and little sleep in the now-free weekend ahead. But as experimentalists, our troubles were only just beginning.

Electron accelerators have special difficulties. Electrons are by far the lightest of the charged particles, which means that they emit copious amounts of so-called synchrotron radiation when they're accelerated. Scientists in other fields make good use of this trait, but in our case the noise it created was an infernal nuisance. We'd calculated that we could "see" neutrons as long as the noise was no more than about 100 million times the expected signal rate. The scale of our experiment was large (today's are even larger). We would measure the scattered electrons using an existing 20m magnetic spectrometer weighing about 60 tonnes. For the neutrons, we would gauge their time-of-flight to a new spectrometer comprising 2m-long plastic scintillators arranged like oversized venetian blinds. Encased in thick lead shielding, it would weigh about 40 tonnes. We also had to be able to move both detectors and precisely measure the angles of particle production - an engineering tour de force.

ÖйúAƬ

When the manufacturer delivered the first of the scintillators for approval, we measured the actual noise rates we would have to deal with, a task that had fallen to the Italians in a night-shift run. It was to be a moment of truth. Adalberto Giazotto delivered his report in my office early the following morning. It was dreadful - the noise rate was ten times what we'd expected. We could have handled a factor of two by tweaking, but ten was virtually terminal. The news fell like a tonne of bricks. Our high-profile, expensive and international experiment would not work. The horrible spectre of failure and ridicule was opening up before us, and some four years of industrial strikes, science politics and the real work of preparation was about to come to nothing. Moreover, as things stood, our doomed juggernaut was rapidly gathering momentum - the scintillators were about to roll off the production line and our engineers were finalising the now-useless set-up only a few metres away.

This is the sort of situation when good groups come into their own. If there is mutual trust, they can do almost anything. Indeed, the resourceful Giazotto and his colleagues had not wasted the night. He said that there may be a way forward, but it would involve radical changes. The scintillators should each be cut into six (thereby reducing the noise per channel sixfold) and instead of a venetian blind, we would have 156 blocks arranged in a honeycomb matrix. There was then the little problem of photomultipliers. We had 52 5-inch ones, which we could swap for 156 2-inch ones; these were much cheaper, but would require some adjustments. But photomultipliers need amplifiers. They were expensive, and we only had 52. Silence ensued. Giazotto suggested that each photomultiplier did not need a dedicated amplifier. Photomultipliers could share them, provided that the threefold sharing was arranged so that signal channels were as far apart from each other as possible. With some clever programming, we should be able to sort out which counters had actually fired using our powerful computer. But would it all work?

We had to take action quickly, but it had to be right. So we gathered the group together to thrash out the plan for averting the looming disaster. Fortunately, our fate was in our own hands - we wouldn't be second-guessed. When we'd finally agreed, I went to persuade our engineer to make the changes. The heavy weight of responsibility felt remarkably light, which was worrying. There was no doubt that the sky was about to fall on us, but we'd chiselled out an escape route that not only seemed viable, but actually seemed better than the original. Luckily, our engineer loved a challenge and quickly concluded that his team could just about make the huge number of changes in time. All these discussions had taken only two hours!

The new noise rate proved tolerable, and we calibrated our highly unorthodox set-up using neutrons from the Harwell Laboratory. The completed detector worked exceptionally well and turned out to be perhaps the biggest and most sensitive neutron spectrometer in the world. Most important of all, we got a good measure of the axial-vector form factor that has not only stood the test of time but has not been bettered.

ÖйúAƬ

I doubt if such a story could be told about research being conducted today. Flexibility seems to have disappeared. We were all young, and none of us had run a substantial project before. We'd redesigned a major experiment at a national lab without consulting anyone. The director implicitly trusted us (and everyone else) to do what we'd said we'd do so long as we didn't violate our budget. Nevertheless, there would have been a lot of egg on important faces had we failed. Where has all that trust and confidence gone?