The data is only as good as whoever 'built' it in the first place. Since most of the European losses were sourced from Henk's files, the dates are as recorded by him. Once in a while when corrections come in, I go in and correct it individually. In this case, I am assuming Henk may have used the Jim Halley's RAF Serials or Bill Chorley's BCL losses to collect the data and he used the second date whenever "on the night of" appears. The DB does not support "8/9" type of dates so we have to go with only one.
The POW DB was updated based on AIR81 records, so its no surprise that it reflects the date from the AIR81 file.
The same issue exists with CWGC.. many ac losses are recorded for the night before, and casualties are shown as dying a day later..
In the end the multiple databases that are here are linked via a common theme (in this case serial number) .. so "Your market may vary" in these searches. It gives you a one view to various different data points..
Bookmarks