On the basis of experience with RA determinations during the polar nights of 1974–1977 on West Spitzbergen and of RA determinations of the Sun, Mercury and Venus at high latitutes one can conclude as follows: At high geographical latitudes the polar night conditions are nearly optimal for absolute determinations of stellar coordinates; high altitude sites near the equator are best for day time observations. It is suggested that observatories be built for absolute position determinations on Spitzbergen, and on McMurdo on the Antarctic Continent. The link between observations in the northern and the southern hemispheres can be established through observations at a high-altitude equatorial observatory, where the Sun and the planets can also be observed.