Table of contents

Fifteen national laboratories participated in the comparison, using standard lamps of two different types for the comparison of luminous intensity and of a single type for that of luminous flux. The results, the first obtained since the adoption of the new definition of the candela, show only slight improvement in the agreement between the laboratories. But, since the number of participating laboratories has doubled since the last comparison and the new methods of realization used have differed more widely in technical detail, the mean representations of the candela and of the lumen are better established, which constitutes important progress.

Results of a theoretical study concerning the measuring principle of the interferometric fringe counting method are used to show ways and means to suppress the quantizing error by an integrating measurement of a stationary sinusoidal vibration. There are two counting methods presented which enable one, under different conditions of use, to reach a lower amplitude limit which with a comparable accuracy is by 3 powers of ten lower than that of the conventional interferometric fringe counting method. Results of experimental investigations including international comparison measurements have shown that with the outlined methods and equipment the range of frequencies in which absolute calibration of accelerometers can be performed while keeping the recommended acceleration range 1 to 100 m s^-2 has been extended to 20 kHz.

Measurements of the Newtonian gravitational constant yielded the value G = 6.667 × 10^-11 Nm² kg^-2 with a 1σ uncertainty of ± 7 × 10^-4. The report describes the set-up and function of the measuring equipment. The basic principle underlying the experiment is the classical principle of the torsion balance. New components are the mercury bearing of the balance beam and the control system with a quadrant electrometer according to Maxwell, which compensates the gravitational forces. The uncertainty of the measurement results and the intended improvements are discussed.

The effect of replacing the boiling point of water with the triple point of gallium in several interpolation schemes for platinum resistance thermometers below 273.15 K is investigated. The sensitivity of the schemes to calibration errors in the gallium point is about five times higher than for calibration errors in the boiling point of water. This disadvantage is offset, however, by the greater accuracy of realization of the gallium point.

The method of clock comparisons using GPS satellites in common view is now widely used in the time laboratories which participate in the international unification of time under the coordination of the Bureau International des Poids et Mesures (BIPM). We report here the results of a campaign of calibration of time delay in GPS receivers under the auspices of the BIPM with the assistance of the National Bureau of Standards (NBS), Boulder, CO. This trip in the United States and in Europe was performed from September 29, 1986 to October 27, 1986.

This paper summarizes a report by the Bureau International des Poids et Mesures (BIPM) of an international comparison of measurements of line scales which was instigated by the Comité Consultatif pour la Définition du Mètre (CCDM). Eleven national laboratories participated in the comparison. The reproducibility of reading a line using various microscopes is characterized by a standard deviation of approximately 0,02 μm in the best case, i.e. when focusing is good. It became evident that one of the causes of the dispersion of the results is the presence of soiling which may be more or less adhesive. For the measurement of an interval L, the standard deviation, s, of the result from any laboratory with respect to the mean of all laboratories is s = 0,03 μm + 0,08 × 10^-6L. This standard deviation is approximately three times that which each laboratory gives to its results; it appears that certain components of the uncertainty have been underestimated and require further study.

The essential need of reliability of atomic time scales requires that these scales be based on the data of numerous clocks. But it is also expected that the algorithms used for processing these data optimize the stability by statistical prediction of the clock frequency. We have investigated the role of the frequency prediction and shown that, in the most common algorithms, frequency perturbations occur when the clock ensemble is modified, especially when clocks are entered or removed.

The frequency-shifting effect that neighbouring lines have on the clock transition of the PTB's frequency standards CS 2 and CSX is experimentally and theoretically investigated. The measurements confirm the model underlying the calculations, which allows for the consinusoidal field distribution in the end parts of the Ramsey cavity, the velocity distribution of the Cs atoms and the population of the hyperfine sublevels.

At the normal value of the static magnetic field usually called the C-field of CS 2, a reliable estimate of the frequency uncertainty contribution due to Rabi pulling is below 1 × 10^-16.