4 Conclusions

The DE405 ephemerides of all the outer planets Jupiter to Pluto are accurate to better than 50mas in the period 1984-1997.

The Hipparcos and Tycho optical positions show that the DE405 ephemerides for Jupiter (Fig. 2) and Saturn (Fig. 3) are accurate to within $\sim$20mas in the period 1991-1993. The Carlsberg observations agree with this conclusion in that period, but outside this period they show a positive bias in right ascension and a fluctuation in the declination of Jupiter with the 12-year orbital period of Jupiter. The correlation between the fluctuations of the residuals for Uranus and Neptune (Fig. 4) leads to the conclusion that the Carlsberg observations of all the planets are subject to systematic errors of a quasi-periodic nature which complicates the detailed interpretation of the residuals. Nevetheless, the common positive bias in the right ascension Jupiter, Saturn, Uranus and Neptune suggests that there is a frame-tie problem in DE405 for the outer planets which requires further investigation.

Using a CCD camera to measure the positions of Solar System objects relative to the ICRF over a field of $\sim$0.3$^\circ$ is one way of reducing the systematic errors of the optical observations, as demonstrated by Stone (1996) and Réquième et al. (1997). A CCD camera is to be fitted to the Carlsberg telescope in 1998.

Acknowledgements

We thank Daniel Hestroffer for the Tycho observations and Michel Rapaport for supplying the Bordeaux observations.