The number matters, but so does how the question was asked

A measurement is best understood as an answer to a question asked of the world, not as a bare fact lifted from it. The number matters, but so does how the question was asked. A thermometer reading of 21.4 degrees does not, by itself, tell you exactly how warm a room is.

At Silverbrook, where I was born, the river seemed to offer an easy example. One could ask how wide it was at the bend, and answer in metres. But the answer would depend on where, exactly, one chose the banks to be. After rain, the water reached farther up the shore. In a dry spell, it drew back. Even on the same day, the width depended on whether one measured at the surface or along the riverbed.

The number might be written neatly in a notebook. The measurement still carries the choices that produced it.

What does a reading leave out?

A digital thermometer might display 21.4 degrees Celsius. That decimal place can make the reading look exact. Yet the thermometer may be slightly miscalibrated, the room may be warmer near a radiator, and the sensor may not have had enough time to settle. The display reports what the instrument calculated from its response. It does not erase the conditions under which that response occurred.

This does not make the measurement useless. It tells us what sort of confidence is reasonable. If someone wants to know whether a room is comfortable, a reading around 21 degrees may be enough. If someone is testing a medicine that must be stored within a narrow temperature range, the thermometer, its calibration, and its location matter much more.

The same number can therefore be good enough for one purpose and inadequate for another. That is not a defect in arithmetic. It is a consequence of measurement being used to answer a particular question.

When does more precision help?

Writing 21.437 instead of 21.4 adds digits, but it does not necessarily add knowledge. A number can be displayed to many decimal places even when the instrument cannot reliably distinguish such small differences. The extra digits may describe the device’s calculation, rather than a real improvement in what we know about the room.

Precision is useful when it reflects a method capable of supporting it. It is misleading when it merely makes a result look authoritative. A scale that reports a person’s mass to the nearest gram may still vary by several grams each time the person steps on it. The display is precise; the measurement may not be.

This distinction matters whenever numbers travel. A temperature copied into a chart may later be treated as a definite property of a room, even though no one recorded where the thermometer was placed or whether it had been checked. The number survives. The evidence that gave it meaning may not.

So a useful measurement needs more than a value. It needs enough context to show what was measured, by what method, and for what purpose. Sometimes that context is a full laboratory record. Sometimes it is simply “measured at the centre of the room, away from the heater.”

The number is what we can carry. The measurement is the reason we should trust it.