.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

While this paper was at the end-stage of the editorial process, the launch has been posponed to 2008. This will not affect the results in this paper.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... that

The remaining pixels will have an accuracy worse than 1% (the worst accuracy recorded being 2.5%) in the case of a displacement of the spacecraft and of the Sun at the limits of the region for which the expansion is calculated.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... noise

We consider in this work instrumental white random noise. P LANCK receivers are affected by 1/f-like noise that introduces long term correlations appearing as offsets in TODs (and as stripes in maps). On the other hand, destriping algorithms (see, e.g., Burigana et al. 1997) accurately remove this effect in the TODs (and in the maps).

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... allowed

This is not true for a map since in this case $\sigma_p^2$ is a function of p even for stationary white noise.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... LANCK

In this calculation it is assumed that P LANCK is orbiting about the L2 point according to the nominal orbit expected for a launch in February 2007. Changes in this orbit will only slightly change the results discussed above.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... estimate

In addition, while finalizing this paper, a significant improvement in the P LANCK star trackers was achieved leading to an expected pointing accuracy of few arcsec (see, e.g., Harrison & van Leeuwen 2005).

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.