The Earth's variable rotation, its departures from what it would be if it were a rigid body rotating in isolation, has occupied the interest of astronomers and geophysicists for more than 100 years. The reason for this is quite clear when one becomes aware of the range of processes that perturb the Earth from uniform rotation (Figure 1). A complete understanding of the driving mechanisms requires a study of the deformation of the solid Earth, of fluid motions in the core and the magnetic field, of the mass redistributions and motions within the oceans and atmosphere, and of the interactions between the solid and fluid regions. The discussion of evidence for the variable rotation includes the examination of not only a variety of optical telescope evidence that goes back some 300 years, but also of historical records of lunar and solar eclipses, and planetary occultations and conjunctions for perhaps the past three millenia. The geological record, in the form of fossil growth rhythms in organisms such as corals, bivals or brachiopods or as cyclic organic growth and sediment sequences such as stromatolites or banded iron formations, extend, albeit with considerable uncertainty, the record back through Phanerozoic time and into the Early Precambrian. To this variety of measurement techniques now has to be added the new methods derived from the space-oriented technological developments of the past few decades.

Twenty years ago I published A Catalog of 939 PZT Stars on the System of the FK4. At the time it was well received and indeed it did tighten up various PZT results. As I recall, the catalog, among other things, confirmed an empirical relationship between the Richmond, Florida and Washington, DC instruments which had been derived by Prof. Markowitz. In any event, I planned to follow up on this work, but an assignment to El Leoncito, Argentina intervened. Perhaps it was just as well because other things were happening. Astronauts were soon tramping about the moon and littering up the place with a used car and something called a retro-reflector. Radio astronomers were constantly refining the astrometric potential of their interferometric techniques; strange, high-z, starlike objects had been detected a few years earlier; shiny, new satellites were launched; and shortly the science of determining Earth Rotation Parameters (ERP), or “Orientation Parameters” if you prefer, was to explode with vitality and to reach precisions and accuracies never before achieved. As a matter of fact, the new riches became so numerous that, like the proverbial kid in the candy store, the ERP community was, to some extent, forced to choose among some of them. Thus was born the MERIT campaign.

Present and near future advances in astrometry at radio and optical wavelengths will allow at least an order of magnitude increase in the precision with which fundamental reference frames are defined compared with those available ten years ago. A brief review is given of the present status of fundamental celestial reference frames with a view towards defining the problems encountered in establishing a fundamental reference frame and methods of linking these reference frames. An estimate is made of the progress to be made in reference frames over the next ten years.

In order to give an exact and clear description of the angle of rotation of the Earth, we propose to use, as the reference point in space, a “non-rotating origin” (Guinot 1979) such that its hour angle, reckoned from the origin of the longitudes (or “non-rotating origin” in the Earth), represents strictly the specific rotation of the Earth. The position of this origin on the instantaneous equator depends only on the trajectory of the pole of rotation. We show that the estimation of the deduced angle of rotation is not critically affected by the precision with which this trajectory is known. We give therefore the formulae to obtain the non-rotating origin, at any date t, from a chosen fixed reference, and we propose a definition of the Universal Time which will remain valid, even if the adopted model for the precession and the nutation is revised. We show that the use of the non-rotating origin also simplifies the transformation of coordinates between the terrestrial and celestial reference systems.

Universal time is a commonplace quantity, and yet it has no generally accepted definition. The expression for sidereal time is often quoted as a definition, but this does not convey much meaning to the average person. Also sidereal time is measured from the equinox of date, and so involves the precession formulae of the stellar reference frame, and hence may not be ideal for other modern techniques of observation. Guinot (1979) has proposed a broad definition of UT1:

“UT1 is an angle which is proportional to the sidereal rotation of the Earth, the coefficient of proportionality being chosen so that UT1, in the long term, remains in phase with the alternation of day and night. In some applications, UT1 can be considered as a non-uniform time scale”.

The modern space techniques have achieved such an unprecedented accuracy and precision in position measurements that there is required a rethinking and reformulation of reference frames, both in their concepts and in their realizations. Various reference frames are a natural consequence of the different observational techniques and data types. Not only are the determination and the necessity of these various frames stressed, but also their evolving nature is recognized. The current status of frame ties is highlighted, and the need for improved ties is motivated. We conclude with our concerns and recommendations.

Complete ephemerides of the moon and the four inner planets could be created solely from ranging data alone. Such ephemerides would then be independent from any outside astronomical reference system, and, therefore, would define their own unique reference frame. In fact, this is nearly the case with present-day ephemerides; the accuracy of the ranging data tends to dominate most of the least squares adjustment.

This paper outlines the process of creating the lunar and planetary ephemerides along with the orientation of the ephemerides onto the dynamical equinox. The resulting accuracies of these processes are given and a number of uses for the ephemerides are highlighted.

We have begun a program to establish a dynamical reference frame based on the motions of minor planets. The program will utilize observations from the Hubble Space Telescope, and will ultimately tie the HIPPARCOS reference system to a dynamical base. Thirty-four minor planets, 20 of which are suitable for observation with the Hubble Space Telescope, have been selected. Ground based observations, particularly crossing-point observations with long focus reflectors, have been initiated.

A computer program to simultaneously solve for the corrections of the orbits of the 34 minor planets including the crossing-point observations, was successfully run. The observations are treated by the method of W. H. Jeffreys. Using simulated data, solutions with and without crossing point observations demonstrate the value of those observations to produce a homogeneous and coherent set of results.

Several series of celestial reference frames have been produced during the past few years as part of VLBI geodynamic programs. They consist in coordinates of 20 to 150 extragalactic radio sources with a precision at the level of 0.001″. The relative orientations of these frames and the evolution of each series of catalogs are studied.

Very Long Baseline Interferometry (VLBI) and optical positions of 8 radio stars are compared in the J2000.0 system. The mean differences in right ascension and declination found are +0.02″ ± 0.04″ and −0.02″ ± 0.07″, respectively. These differences show that the JPL radio celestial reference frame is aligned on a preliminary FK5 frame to at least this level.

In order to provide a direct connection between the radio reference frame and the fundamental optical system, based on fundamental stars, a radiostar observing programme has been started. Most of the astrolabe stations in the world are participating following the proposal of one of us (Débarbat, 1980), after Commission 24 of the IAU created a Working Group “On the Identification of Optical/Radio Astrometric Sources”.

A celestial reference frame can be defined by precise positions of extragalactic radio sources using Mark III VLBI data available to the NASA Crustal Dynamics Project for geodynamic research. Seven years of such data have been analyzed to generate a catalogue of 101 sources with formal statistical errors between 0.01 and 0.77 ms in right ascension and between 0.2 and 9.3 mas in declination. In order to achieve such precision it is necessary to adjust the standard IAU nutation model. The rotations and scatter of the positions from year to year are generally less than 1 mas. A comparison of this catalogue with a completely independent catalogue derived from Mark II data shows a weighted average position difference, after a rotation, of 1.9 mas.

The right ascension of 3C273B has been a reference of the celestial VLBI coordinate system. However it seems that 3C273B is inadequate for the reference because 3C273B has the jet component which was detected on long baselines in our experiments. We present the best reference source for celestial coordinate system. Next we estimated the source positions, taking the new reference radio source. The nutation and plate motion were considered in our analysis. We show the change of 3C273B position. The estimated source positions are compared with those given in GSFC source catalog (1985).

In astrometry, celestial radio sources with extended structures are useful. An algorithm to correct for structure in VLBI delays and delay rates is developed. The radio source NRAO140 is mapped from data with limited u-v coverage by the hybrid method and the Maximum Entropy procedure. The magnitude of the structure corrections in astrometry is discussed in the case of NRAO140.

The weak thermal emission from the largest minor planets can be detected in the microwave regime by the Very Large Array (VLA). Signal-to-noise ratios are sufficiently high to permit precise measurement of the positions of these objects at all points in their orbits with respect to background extragalactic sources. We are in the process of obtaining observations of astrometric accuracy for minor planets 1 Ceres, 2 Pallas, 4 Vesta, and 10 Hygeia.

Minor planets have historically served as “test particles” in the solar system, and optical observations of these objects have been valuable in the determination of fundamental astronomical constants. In particular, optical observations of minor planets have played an important role in the establishment of the fundamental optical reference frame by permitting the determination of the orientation of the Earth's orbit relative to the stars defining the frame.

Similarly, radio observations of these bodies can play a corresponding role in the establishment of a fundamental radio reference frame. Our observations will provide a direct link between the dynamical and radio reference frames, and provide important information on the relationship between the radio and optical reference frames.

Three fundamental concepts of reference frames in relativistic space-time are confronted: 1. the gravitational compass, 2. the stellar compass and 3. the inertial compass. It is argued that under certain conditions asymptotically fixed (stellar) reference frames can be introduced with the same rigour as local Fermi frames, thereby eliminating one possible psychological reason why the importance of Fermi frames frequently has been overestimated in the past. As applications of these three concepts we discuss: 1. a relativistic definition of the geoid, 2. a relativistic astrometric problem and 3. the post-Newtonian theory of a laser gyroscope fixed to the Earth's surface.