The Experiment That Failed for Six Weeks

For six weeks — forty-one days, I counted — my experiment produced nothing. Not interesting nothing, not puzzling-anomaly nothing that you could one day put in a thesis. Flat, dead, dial-tone nothing. The instrument would hum, the software would draw its axes with bureaucratic optimism, and then a line would crawl across the screen at zero like something that had given up on both of us.

This entry is the diary of those six weeks. I am writing it partly for the tropical scholars who come after me, partly for the version of myself who will face the next long failure — because there will be a next one; I understand the business model now — and partly because the ending deserves a record.

A taxonomy of failure

Week one, I assumed I was the problem. This is the immigrant-student reflex: when the machine fails, check yourself first, apologise second, suspect the machine never. I reran my preparation protocol so many times I could do it today blindfolded, in a power cut, during an earthquake.

Week two, I suspected the software. Week three, the cables — laugh if you like, but every experimentalist I told this story to nodded gravely at the word cables, the way sailors nod at the word weather. Week four, I began to suspect the building itself: vibrations from the construction site, interference from the tram line, the humidity, the phase of the moon. By week five I had constructed a spreadsheet of forty-one variables and I was testing them with a grimness that frightened the master's students.

What the weeks actually taught, condensed for future reference:

• Change one thing at a time, or learn nothing from the change
• Write everything down; memory edits failures into fables
• The error is usually in the step you trust most
• Asking for help on day ten is not weakness; not asking until day thirty is
• Stubborn means trying differently. Stuck means trying identically, with feeling.

Day forty-one

It was Priya, of course. Priya of the winter-clothing lecture, who owed me nothing and gave me an afternoon. She refused to look at my spreadsheet — spreadsheets are where hypotheses go to hide, she said — and instead made me walk her through the physical setup, hands on everything, narrating like a museum guide. And when I lifted the sample holder out of its mount to show her the loading procedure, she took it from my hand, held it up to the window, and turned it forty-five degrees.

A crack. A hairline crack in the ceramic, invisible from above, visible only edge-on against the light, letting the vacuum seal leak by an amount too small to trip the pressure alarm and exactly large enough to kill every measurement I had attempted for six weeks. It had probably been cracked since a mishandling incident before Christmas — before my experiments even began. The machine was fine. The protocol was fine. I was fine. A twelve-pound component was not.

What failure is for

The new sample holder arrived on a Tuesday. I mounted it with the reverence of a priest, pumped down the chamber, started the measurement, and watched the screen the way you watch a phone that might ring. And the line moved. A peak — small, noisy, undeniable — rose out of the baseline exactly where the theory said a peak should live. I made a sound in the empty lab that I will not transcribe. Six weeks of nothing, and then the universe answering, in its thin electronic handwriting: yes, fine, you were right, it does work this way.

Here is what I keep turning over. In forty-one days of failure I learned the instrument down to its bolts. I learned our vacuum system's moods, wrote a leak-checking guide that is now taped inside the lab cupboard, and earned a strange new credibility in the group — because everyone here, it turns out, has a six-weeks story, and mine was my citizenship test. I read later that a researcher had built a whole argument that science advances on precisely this kind of productive failing — the importance of stupidity in scientific research — the daily business of being confused at the edge of what anyone knows. Six weeks ago I would have read that essay as consolation. Now I read it as a job description.

My mother asked, on Sunday's call, what I had discovered this term. I tried to explain that I had discovered a crack in a piece of ceramic, and that this was somehow a triumph, and she was quiet for a moment and then said that when the sewing machine jams, the fault is a bent needle worth two rupees, but only the person who knows the machine can find it. Exactly, I said. Exactly. My doctorate is teaching me to be the person who knows the machine. The peak, by the way, has appeared every day since. It is the most beautiful noisy little bump in the United Kingdom, and it is mine.