Dear Magicians,

I’m an experimental physicist who has spent three decades calibrating scientific instruments.

In 2012, I helped develop a technique called self-calibration for the polarization angles of Cosmic Microwave Background telescopes. It is still used today.

Recently, however, I discovered that the word calibrate may itself need calibration.

The word comes to English through French calibre, meaning the size or bore of something. After that, the etymological trail forks.

The conventional explanation traces calibre to the Arabic qālib, meaning a mold, form, or pattern, perhaps ultimately from a Greek word for a shoemaker’s last: the foot-shaped form around which a shoe was made.

That is wonderfully appropriate.

Long before SI units, human beings used their own bodies as standards. The foot. The hand. The cubit, or amah, derived from the forearm. Measurement began with an extraordinarily literal question: Compared with what?

But there is a rival etymology, probably wrong and almost too perfect to resist.

Some nineteenth-century scholars proposed that calibre came from the Medieval Latin quā librā: “by what weight?” Libra meant both a Roman pound and the balance used to weigh it.

Look again:

ca-LIBRA-te.

The derivation is disputed. The metaphor is not.

Calibration is fundamentally an act of comparison. You put the unknown on one side of the balance and a trusted standard on the other.

The Torah understood the stakes thousands of years before NIST. Deuteronomy forbids merchants from carrying heavy and light versions of the same weight and calls dishonest measures a to’evah, an abomination before God.

Today, instead of biblical weights, inspectors walk into supermarkets carrying certified masses. Put a known weight on the scale. Does the scale report the correct answer? If not, it can be taken out of service.

Modern technology changed enormously.

The epistemological problem did not.

And there is a subtlety even scientists routinely get wrong:

Calibration does not make an instrument correct. It tells you how wrong it is.

A thermometer can be calibrated and still read one degree too high. Calibration establishes the relationship between its answer and a trusted standard. Adjustment comes afterward.

Measurement asks:

What is the value?

Calibration asks the more dangerous question:

Why should I believe the value?

Which brings me back to self-calibration.

Ordinary calibration requires something external: a known mass, voltage, temperature, star, or other reference. Self-calibration tries to infer the reference from the data themselves.

That is extraordinarily powerful.

It is also dangerous.

Suppose theory predicts that some signal should equal zero. You can rotate your instrument until the signal disappears and declare yourself calibrated.

Unless the signal you erased was never an instrumental error.

Unless it was the new physics you were looking for.

After thirty years in experimental science, I find it reassuring—and slightly alarming—that even the word for checking our measurements against reality comes with its own error budget.

Until next time, have a M.A.G.I.C. week.

Brian