Knowledge Has a Date

Picture a laboratory bench and a weight marked “1 kg”. Today is 19 May 2019. Under the definition in force, the kilogram equals the mass of the international prototype — the platinum-iridium cylinder kept near Paris.

The next day, 20 May, a new definition takes effect. Now the kilogram is set through the Planck constant: its exact numerical value in SI units is fixed. The cylinder stays in its vault. Nobody touched the weight. The definition change made it neither heavier nor lighter.

And yet the question “what defines the kilogram?” wants a different answer yesterday and today.

Already adopted, not yet in force

The decision was taken in advance, on 16 November 2018, at the General Conference on Weights and Measures. The resolution named both the new definition and the changeover date — 20 May 2019.

So months before the changeover one could know how the kilogram would be defined. But that knowledge did not cancel the old definition ahead of the appointed day.

Suppose that on 19 May someone reads the news of the Conference’s decision and says: “The kilogram is now defined through the Planck constant.” He found a genuine decision and grasped the change’s substance. The mistake fits into one word — “now”.

That day the exact answer would run: the new definition is adopted but takes effect tomorrow. Saying “no such definition exists” would be wrong too. It already existed as an adopted decision about an upcoming change.

A rule’s text and its time of effect are different pieces of information. Finding the first does not yet establish the second.

Three dates easily confused

To take such a case apart, three questions help.

Which time is at issue? We may ask what defined the kilogram on 19 May 2019, or on 20 May. To answer, the picked date must be matched against the new definition’s effective date. In the first case the old one applied, in the second the new one.

Which information do we allow in? The past can be taken apart against everything available to us today. Or another task can be set: to recover what a person could have concluded from the information available to him then. In the second case later publications must not be quietly added. Nor is “was published” the same thing as “was known to this person”.

When do we answer? A historical question can be asked a week or many years after the event. In June 2019, say, we may work out which definition applied on 19 May. The answer is given in June but concerns May. A new date on our wall clock does not carry the new definition into the past.

These distinctions do not require three calendar stamps under every phrase. They matter when the answer’s meaning depends on time. Mistakes come easiest when a decision is already adopted but not yet in force, or when news of an event arrived much later than the event.

The definition changes, the weight stays

The changeover was arranged to keep the kilogram’s old size within the precision the best measurements allowed. Scale users did not have to recalculate all results into another unit. What changed was the definition’s ground: instead of one object’s mass — a fixed value of a constant of nature.

A material sample’s mass may slowly drift, from contamination or cleaning, say. A constant-based definition does not depend on one cylinder’s safekeeping. That does not free measurements from uncertainty, but lets unit-realisation methods improve without changing its definition.

The second and the metre were revised differently. Their physical ground stayed: for the second, a fixed transition in the caesium-133 atom; for the metre, the speed of light in vacuum. The wordings were brought to the common form of definitions through fixed constant values. So to the narrow question “which physical ground does the unit rest on?” the answers match on both sides of 20 May. But the definition texts are not verbatim identical.

So a date helps see both change and continuity. It does not oblige us to hunt a difference where none exists.

Knowing then and knowing now

Back to the 2005 atlas from the Pluto chapter. It lists nine planets — under the then-accepted classification. The 2006 decision changed that classification. It did not turn the old atlas into evidence that its makers could not count.

But with claims about nature’s workings it is sometimes otherwise. Suppose observations back a general rule, and later an exception turns up. The old conclusion may well have been grounded in the available information. From this it does not follow that nature obeyed it with no exceptions before the discovery.

We return to this distinction in the genetic-code chapter. For now two thoughts suffice: a person may have had serious grounds for his conclusion, yet the conclusion itself may still have proved too wide. To assess grounds, recover the then-available information. To check the claim, allow in what was learned later too.

A date explains why a person may have reached a certain conclusion. It neither excuses a mistake nor makes the false true.

One more question for a claim

We already hold five questions: what exactly is claimed, which information backs it, which conditions are accepted, how the conclusion was reached, and what would make us revise it. Add one more: as of which date do you know it?

Sometimes it needs refining: are you speaking of how things stood then or now? Do you use the information available at that moment, or allow in later findings? When was the conclusion itself made?

An answer grows no weaker from these refinements. We see better what it stands on and where its scope ends.

On the bench lies the same weight. What changed is the answer to where we take the unit for measuring its mass from. And that change has a date.