Fredrik,
At the beginning of using the database, I had filled the tool "Ignore when sorting" with many prefixes ( Le, La, Les, L', Un, Una, Der, Die, Das, and many others )
A first problem was with L' : In french it's attached to the next word ( without space ), sample :
L'affaire Baldwin ( in english : The Baldwin Affair ). I should enter and record the book first, and then re-open the book to edit manually the Title sort to obtain : Affaire Baldwin, L'
Same with different languages in the same database :same words means diffent things :
French:
Il tue son voisin ( in english : He kills his neighbour ) : it's classed under the letter i
Italian :
Il prezzo del silenzio ( in english : The price of the silence ) : it's classed under the letter p
The only way to do it was to edit manually the Title sort
The second problem was when I updated Bookcat to 10.14 ( it never happened before with previous upgrades, and I use Bookcat since 2005 ) :
Most of my manually editions of Title Sort turned back automatically to the default sorting :
Sample with Un, Una,...

I'm really disappointed ! I must now re-edit all the database ( nearly 14.000 books ) one by one and without being sure it will not happen again when future updates.
It did not happened in the sorting order of Series or Persons : another Sample :

Do you have any idea what happened ? and to avoid such things for the future ?