In order to obtain very precise positions of the stars on a sphere, it seems absolutely necessary to go out the atmosphere.

1. 1 – We avoid atmosphere refraction with always uncertain corrections.

2. 2 – We avoid atmospheric turbulence. So the quality of the images is limited by diffraction and not by the seeing.

3. 3 – We avoid atmospheric absorption and diffusion limiting the possibilities on the faint stars. Moreover if we observe from an artificial satellite we have no bending of the instruments.

In view of very high precision (0″.001) on the directions from an artificial satellite, the difficulty is to obtain sufficient stabilisation in attitude. Also, it is necessary to measure the angles between the stars, instead of the directions relative to the satellite.

The mapping of a sphere by angular measurement requires precise measures of large angles; Indeed BACCHUS has shown how that is possible to obtain a precise sphere with relative positions and proper motions, but absolute parallaxes only by precise measures of angles near one value, 90°. Those measures are possible by superposition of two fields in one telescope. A special complex mirror (Fig. I) images the two fields at basic constant angle. We have only to make differential measures to add a small value to the constant.