From GE Tech Notes ....

   "Among the many users of batteries in both the industrial and consumer
sectors, the idea of a memory phenomenon in nickel-cadmium batteries has been
widely misused and understood. The term 'memory' has become a catch-all
'buzzword' that is used to describe a raft of application problems, being most
often confused with simple voltage depression.

   "To the well informed, however, 'memory' is a term applied to a specific
phenomenon encountered very infrequently in field applications. Specifically,
the term 'memory' came from an aerospace nickel-cadmium application in which the
cells were repeatedly discharged to 25% of available capacity (plus or minus 1%)
by exacting computer control, then recharged to 100% capacity WITHOUT OVERCHARGE
[emphasis in the original]. This long term, repetitive cycle regime, with no
provisions for overcharge, resulted in a loss of capacity beyond the 25%
discharge point. Hence the birth of a "memory" phenomenon, whereby
nickel-cadmium batteries purportedly lose capacity if repeatedly discharged to a
specific level of capacity.

   "The 'memory' phenomenon observed in this original aerospace application was
eliminated by simply reprogramming the computer to allow for overcharging. [Note
that no mention is made of adding an intentional *discharge* to clear the
problem - RLM] In fact, 'memory' is always a completely reversible condition;
even in those rare cases where 'memory' cannot be avoided, it can easily be
erased. Unfortunately, the idea of memory-related loss of capacity has been with
us since. Realistically, however, ' memory' cannot exist if any one of the
following conditions holds:

      1. Batteries achieve full overcharge.
      2. Discharge is not exactly the same each cycle - plus or minus 2-3%
      3. Discharge is to less than 1.0 volt per cell.

   "Remember, the existence of any ONE of these conditions eliminates the
possibility of 'memory'. GE has not verified true 'memory' in any field
application with the single exception of the satellite application noted above.
Lack of empirical evidence notwithstanding, 'memory' is still blamed regularly
for poor battery performance that is caused by a number of simple, correctable
application problems."

End of quote ... Basically memory (loss of capacity) due to discharge is a myth.

But it should be conditioned when new.