# 10.2.1 Source identifiers

The source identifiers have a level 12 HEALPix index encoded in their upper bits. This index may, however, in some cases be slightly deceptive as illustrated Figure 10.4, which shows the distance from each source to the centre of the pixel encoded in its source identifier. Depending on the declination, the distance should normally not exceed about 0.011 ${}^{\circ}$ and never exceed 0.015 ${}^{\circ}$, however there are 1972 sources outside that limit. This is largely due to high proper motion stars having moved to a neighbouring HEALPix either since the epoch for the input catalogue prepared before launch or since their first observation.

The cross match process linking clusters of on-board detections to a list of sources is highly complex and involves identifying - and removing (black listing) - detections that are deemed spurious. If an isolated cluster of detections can be equally well matched to two different sources, these sources are merged in the sense that they are replaced by a single source with a new source identifier. If, on the other hand, we have two clusters and only one matching source, this source is split in the sense that it is replaced by two new sources each with their own new source identifier. From Gaia DR1 to Gaia DR2 the number of detections has increased substantially and the strategy for black listing spurious detections has improved. As a result, many source identifiers from Gaia DR1 are no longer present in Gaia DR2. Figure 10.5 shows the fraction of repeated source identifiers. This fraction is very low for the brighter sources, which are especially influenced by spurious sources generated during the cross match for Gaia DR1, though not necessarily published in Gaia DR1.