# 8.1.1 Completeness

• Completeness has improved from Gaia DR2, as shown by several comparisons:

• Comparison with OGLE (Section 8.4.1)

• More stars in the center of Andromeda and in M32 versus Gaia DR2 (Section 8.4)

• Checks in crowded globular clusters (Section 8.7) show that the completeness is generally higher than in Gaia DR2, but strongly depending on the density. In dense areas of globular clusters, a percentage up to 20-30% of stars with astrometry do not have $G_{\rm BP}$ and $G_{\rm RP}$ magnitudes Open clusters are more favourable cases and in general the percentage of stars missing $G_{\rm BP}$, $G_{\rm RP}$ is of the order of 1%-3%. Some artifacts on the completeness showing the effect of the scanning law are visible but the number is reduced in comparison to Gaia DR2.

• Tiny gain in resolution due to the new criterion for duplicated sources (Section 8.2, Figure 8.1).

• Bright stars environment is clean, no completeness problem found around bright stars (Section 8.2.1).

• Based on star counts, the Gaia EDR3 catalogue seems to be essentially complete between $G=12$ and $G=17$ (Section 8.3). Thus, the source list for the release will be incomplete at the bright end and has an ill-defined faint magnitude limit. Fainter than $G=17$ the completeness is complex, being affected by crowding and strongly depending on celestial position (Section 8.2). In any case, comparison with the GOG simulation shows that Gaia EDR3 completeness has improved with respect to Gaia DR2 at $G=19$, although it is still not as high as expected (Section 8.3, Figure 8.10).

• The combination of the Gaia scan law coverage and the filtering on data quality which is done prior to the publication of Gaia EDR3, can lead to some regions of the sky with source density fluctuations that reflect the scan law pattern. In addition, gaps may exist in the source distribution. This becomes more marked if one uses subsets of different astrometric solutions (2p, 5p, 6p).

• In any case, no significant ‘holes’ are found in the sky (Section 8.2).

• Comparisons to WDS show a high completeness for separation above 1${}^{\prime\prime}$, but a rapid decrease at smaller separations (Section 8.4).