SI units -- observable and derived

[js_dukelow@ccmail.pnl.gov]

To RADSAFE:

The recent flurry of messages about SI units sent me to my dictionaries 
and handbooks to try and dig up some of the history.

When I want a horseback estimate of some geographical quantity, such as 
the also recently discussed volume of the oceans, I usually start with 
the definition of the meter as 1.0E-7 of the distance from the pole to 
the equator along the prime meridian.  Sure enough, my American Heritage 
Dictionary, 2nd College Edition, notes that the meter was defined in 
1790 as 1.0E-7 of the Earth's quadrant along the meridian passing 
through Paris ( ... of course).  Memory is that sometime around 1900, it 
was redefined as an artifact, the distance between two marks on a bar of 
precious metal kept under defined conditions somewhere in Paris ( ... of 
course).  Subsequently, in 1960 it was redefined as the length equal to 
1,650,763.73 wavelengths of the orange-red radiation of Krypton-86 
(probably of an atom kept somewhere in Paris).  The General Electric 
Chart of the Nuclides, 14th Edition, notes that in 1983 the CGPM (the 
General Conference of Weights and Measures) redefined the meter as 
distance traveled by light in a vacuum during 1/299792458 of a second.

Similarly, the Geigy Scientific Tables, 7th Edition, notes that "up 
until 1964, the litre (l) was defined as the volume occupied by 1 kg of 
pure air-free water at its maximum density (approx. 3.98 degrees C) 
under normal atmospheric pressure. ... In 1950 the International 
Commission of Weights and Measures gave 1 litre = 1.000028 cubic 
decimetre (relative uncertainty = +/- 3.0E-6) as the best conversion 
factor between the litre so defined and the cubic decimetre.  In 1964 
the 12th General Conference of Weights and Measures abolished the old 
litre definition of the 3rd General Conference in 1901 and agreed that 
the name 'litre' should be used as a synonym for cubic decimetre ...".  
Thus, a liter is now officially and exactly 1000 cubic centimeters, but 
a cubic centimeter no longer contains exactly 1 gram of water.

The most significant digits I could find for Avogadro's number was

        6.0221367E+23 with a standard deviation of 0.0000036E+23

given the the GE Chart of the Nuclides.

The lesson seems to be that metric (or rather SI) units wander back and 
forth between being defined in terms of physically observable quantities 
or in terms of physical artifacts or being derived from other units.

Best regards.

Jim Dukelow
Pacific Northwest National Laboratory
Richland, WA

js_dukelow@pnl.gov