The Janitor Who Fixed the Experiment

In the physics building of a large research university, a graduate student was, at half past eleven one Thursday night, sitting on the floor of a laser lab in a state of quiet despair. Her PhD experiment — three years of setup, a grant renewal in two weeks, and a paper she had been drafting for six months — had that afternoon begun to produce a signal that made no sense. Her advisor was at a conference. Her lab-mates had gone home. She had spent the last four hours checking, rechecking, and re-rechecking every part of the optical alignment. Nothing she could see was wrong. Everything she was measuring was wrong.

At about a quarter to twelve the door opened and the night janitor, a man in his fifties named Ray whom she had greeted a hundred times without ever having a real conversation with, came in with his cart. He apologised for the intrusion and began, in his usual quiet way, to empty the bin at the front of the room. He noticed her, sitting on the floor. He noticed she had been crying. He asked, in the way people ask when they mean it, whether she was all right.

She, exhausted, told him. Not the physics — she did not think he would follow — but the shape of the problem. Three years of work. A signal that had been clean for a year. A drift she could not explain. He listened, wiped his hands on a rag, and then, after a moment, asked if he could look at the apparatus. She was tired enough to say yes.

Ray walked around the optical table slowly. He looked, for a long time, at a mount that held one of the mirrors. He looked, for a long time, at the ceiling. He looked, for a long time, at a small fan on a shelf that had been running since before the graduate student had joined the lab. He said, mildly, 'That fan started running slightly slower about two months ago. You can hear it. It's still moving air, but it's about half the speed. This side of the table's a bit warmer than it used to be. Your mount there,' — pointing — 'is aluminum. It'll be about a millimetre longer than it was in April. Is that going to matter for you?' He said all this in the tone of someone remarking on the weather.

The graduate student, who had spent three years learning that a change of one micron in that mount would move her signal by exactly the amount her signal had, in fact, moved, sat on the floor for approximately fifteen seconds in complete silence. Then she stood up, replaced the fan, wrote down the temperature at the table, and — at three in the morning — sent her advisor an email with the diagnosis. Two days later her signal was clean. Her paper survived. Her grant renewed.

It emerged, over subsequent conversations, that Ray had been a machinist in a semiconductor plant for twenty-two years before an injury had ended that work. He had, over those twenty-two years, learned to hear a bearing wearing out from ten metres away. He had learned to see thermal expansion the way a bricklayer learned to see a wall out of plumb. He had never had the vocabulary of academic physics, but he had, in a real sense, been running experiments longer than any graduate student in the building had been alive. Nobody had ever thought to ask him about a signal. He had, for six years, been emptying their bins.

There is a whole class of expertise, in every institution, that is not marked by any credential the institution would recognise. It is carried by people whose job titles suggest they should be invisible to the people they work alongside. The janitor. The receptionist. The nurse who has been in the ward longer than any doctor. The stagehand who has watched two thousand rehearsals. The best move you can make, in almost any professional environment, is to notice these people, greet them by name, and, when you are genuinely stuck, ask them what they have seen. Most of the time they will tell you nothing that helps. Sometimes, at a quarter to twelve on a Thursday night, they will save your PhD.