This page has been archived and is no longer updated - last site update was in 2017.
spacer spacer Go to Kaye and Laby Home spacer
spacer spacer spacer
spacer spacer

You are here:


Chapter: 1 Units and fundamental constants
    Section: 1.1 Units
        SubSection: 1.1.5 Uncertainty of primary standards



« Previous Subsection

Next Section »

Unless otherwise stated this page contains Version 1.0 content (Read more about versions)

1.1.5 Uncertainty of primary standards

Setting up and evaluating a primary standard usually entails a long and painstaking programme of experiments spread over several years before the necessary low level of uncertainty is achieved. A great effort is made to eliminate, or at least to identify and quantify, all possible contributions to the uncertainty of the standard. Statistical (random) contributions may be reduced arbitrarily, e.g. by increasing the number of measurements made, but there always remains the possibility that some unsuspected, non-statistical (systematic) source of uncertainty exists. The application of rules for the statement of uncertainty, as in section 6, is thus less straightforward in the case of primary standards work than it is in routine calibration work.

When the results of measurements are compared by those whose measurements are traceable to the same primary standard, they will be using identical units and any uncertainty in the primary standard may be neglected. Where traceability is to different primary standards, information about any difference in the sizes of the units may be obtained from the BIPM or from one of the national laboratories concerned.

In some scientific work it may be essential to consider the uncertainty in the primary standard when assessing the significance of the measurements. Much relevant information is published by the BIPM, particularly in reports on international comparisons of the units realized and maintained by the various national laboratories. It is always advisable to approach the appropriate expert at the NPL or other national laboratory in order to confirm the interpretation of such material, and to ascertain whether any more recent information is available.

For illustrative purposes only, the table below lists the 1-sigma uncertainties assigned to a number of NPL standards in 1991. Note that standards set up to provide calibrations of very large or very small values of quantities such as pressure, for example, will have much greater uncertainties than those of the base units.

Table of uncertainties



(one sigma)






Time interval


  4 × 10−14




  2.5 × 10−11




  2.3 μg


Potential difference




Electric resistance






  0.000 1 K

  At triple point of water

Luminous intensity


  3 × 10−4

  For monochromatic radiation


  arc second






  For solids




  Up to 1.2 MN







   3 × 10−6

  At 100 kPa

   high pressure


   2 × 10−5

  At 100 MPa




  At 1 μPa

Sound pressure level

  dB re 10 μPa

   0.04 dB

  At 5 KHz in air

Acoustic pressure


   1.5 × 10−2

  At 5 MHz in water

Thermal conductivity

  Wm−1 K−1

   10−2 to 2 × 10−2


Electric power:




   supply frequency


   5 × 10−5

  At 50 Hz




  At 10 GHz

Laser pulse energy


   1.5 × 10−2

  0.4 μm to 1.06 μm

Optical power


   4 × 10−5





  For colour surface

Neutron emission rate


   2 × 10−3

  For radionuclide neutron sources



   1.5 × 10−3 to 10−2

  Depends on radionuclide

Absorbed dose


   3.5 × 10−3

  In graphite

Amount of substance


   3 × 10−4

  For gases, usually as gas mixtures,
     mole fraction

The BIPM suggests the use of ‘category A’ and ‘category B’ instead of ‘random’ and ‘systematic’.

O.C. Jones


spacer spacer spacer

Home | About | Table of Contents | Advanced Search | Copyright | Feedback | Privacy | ^ Top of Page ^


This site is hosted by the National Physical Laboratory © 2018.