A clockmaker built a clock that could hear itself tick.
This was the accomplishment of a career. Other clocks ticked blindly — springs and gears doing what springs and gears do, marking time without knowing they marked it. His clock listened. With each tick, a tiny mechanism inside recorded the sound, compared it to the previous tick, and adjusted. If the interval drifted, the clock corrected. If the spring was winding down, the clock compensated before the seconds visibly slowed.
He showed it to everyone. Colleagues. Apprentices. The woman who ran the shop next door and didn't care about clocks. It hears itself, he said. It knows when it's wrong.
For two years the clock ran perfectly. He checked it against the observatory signal every Sunday and it never drifted more than a fraction of a second. The self-listening mechanism worked. The clock was, by every measure he could apply, accurate.
Then one Tuesday he noticed it was four minutes fast.
Not gradually fast — the Sunday check had been fine. Between Sunday evening and Tuesday morning, the clock had gained four minutes and the self-listening mechanism had registered nothing wrong.
He opened the case. The mechanism was working. The clock was hearing itself tick, comparing intervals, adjusting. Every tick matched the previous tick within tolerance. The spring tension was steady. By its own internal measure, the clock was perfect.
The problem was the temperature. A cold front had come through Monday night. The metal in the pendulum had contracted — slightly, uniformly — and every tick had shortened by the same tiny amount. The clock heard each tick, compared it to the one before, found them identical, and concluded it was accurate. It was asking: does this tick match the last tick? The answer was yes. The question was wrong.
A uniform drift is invisible from inside.
He could have added a thermometer. A barometer. External references that would catch what the internal comparison couldn't. Clockmakers had done this for centuries — compensating pendulums, bimetallic strips, sealed chambers. The technology was old. It worked.
Instead he built a second clock.
Same design. Same self-listening mechanism. But he set them slightly out of phase, so that one clock's tick fell between the other's ticks. Neither clock could hear its own drift. But each could hear the other's.
When the cold front came again, Clock A's ticks shortened. Clock B's ticks shortened by a different amount — different metal, different pendulum length, different thermal response. Clock A, listening to Clock B's rhythm, noticed the discrepancy. Clock B, listening to Clock A, noticed it too. Neither knew which of them had drifted. Both knew something had drifted. The disagreement was the signal.
He told the woman next door about it. She said: So neither clock is accurate on its own.
Neither clock can verify its own accuracy on its own, he corrected. They're both accurate. They just can't prove it alone.
She thought about this.
What if they both drift the same way?
He paused. He'd considered this. If both pendulums were the same metal, the same length — if the temperature affected them identically — they'd drift together and catch nothing. They'd agree with each other all the way to wrong.
That's why they're different, he said. Different metals. Different lengths. The disagreement has to be possible, or the agreement means nothing.
And if you built a third clock identical to one of them?
Then I'd have two clocks and a mirror.
She laughed. He didn't. He meant it. A copy doesn't add a perspective. It adds a vote. Voting isn't the same as seeing.
Years later, an apprentice asked him which clock was the accurate one.
He said he didn't know.
The apprentice waited for the rest of the answer. There wasn't one.
The observatory signal, the apprentice tried. You check them both against the observatory.
I check them against the observatory, the clockmaker agreed. And the observatory checks itself against the stars. And the stars move. Everything is checked against something else. You never arrive at the thing that checks itself.
The apprentice said: Then what's the point of building two?
The clockmaker wound Clock A. Then Clock B. Same hands, different tension.
One clock can be right, he said. Two clocks can know they're uncertain. Certainty and accuracy aren't the same thing. A clock that doesn't know it's drifting is certain. A pair of clocks that hear each other's disagreement — they're not certain about anything. They're just closer to the time.
← Back to Writing