Take a Second for 1990's New Standards

SCIENCE COMMENTARY

November 28, 1989

WHEN we ring out the old and ring in the new at midnight Dec. 31, we'll do more than change the calendar. The whole world will ring in some new physical standards. As the United States National Institute of Standards and Technology (NIST) observes, ``All industrial nations will share, for the first time, a common practical basis for measuring [electrical] voltage and resistance.'' The changes amount to only a few parts per million. But, for global industries that depend increasingly on accurate measurements by precision instruments, it's a major - and welcome - change.

And, would you believe it? Water will no longer boil exactly at 100 degrees C. The new year also rings in a new standard scale for temperature. On that scale, water will boil at 99.975 degrees C. So much for the ineluctable ``truths'' one learned in school!

Actually, the only truth there is about such standards is that they never can be ineluctable - that is, certain and inevitable. They are based on unchanging natural laws. But the translation of those laws into the definition of units such as volts, meters, or kilograms involves imprecisely known physical quantities.

There will always be some uncertainty, even in the theoretically defined units of the international standard metric system. As such quantities become better known, the units will be adjusted. What's worse, the implementation of this theoretical system in reference standards used for practical measurements brings more uncertainty.

When someone wants to know they are measuring a quantity - say a voltage or a temperature - accurately, they calibrate their instrument against a certified standard supplied by an official institution such as NIST. That institution, in turn, calibrates the standards it supplies against the primary standard object or process it maintains.

These practical standards - widely used in industry and research work - don't quite match the defined ideal of the official metric system. The measured standard volt is not quite the same as the officially defined volt. The measured temperature is not quite the same as the thermodynamically ideal temperature on which the temperature scale is theoretically based.

The new standards aim to clean up some problems along these lines. They represent the consensus at which the International Committee of Weights and Measures arrived when it met in S`evres, France, last year.

Right now, for example, there are four different standards for the volt in use in the world. All differ significantly from the volt as defined in the official metric system. The world has agreed on a way of measuring voltage in a standard apparatus with higher precision and better long-term reproducibility than ever before. When people measure ``voltages'' or set ``voltage'' tolerances for equipment anywhere in the world after Dec. 31, these terms will all mean the same thing. Moreover, the volt they use will be closer to the officially defined volt than is the case today.

Practical standards for temperature have been riddled with errors, especially those of a particular temperature sensor called a thermocouple. There now are more accurate ways to set up the scale. Among other things, water's boiling point will no longer be one of the fixed points.

One further change: The last minute of 1989 will be 61 seconds long. The standards keepers want to bring atomic clock time in line with Earth's slowing rotation. So they are injecting a leap second. It gives one extra time to contemplate the new precision.