Missing cycle IDs?

It appears that cycle IDs for some changes between 1993 and 1998 were generated but have not been recorded. For example the upgrade to 4D-Var in November 1997 is recorded only by its date, but it may in practice have had a cycle ID given the sequence in which it occurs. On the other hand, some updates never had new cycle IDs (for example some changes to observation usage) so it is not certain either way.