Ed Gaida has inquired about the Ampico Trial Rolls that come around
every once in awhile.  From my understanding (and Bruce Clarke may be
able to shed more light on the subject) the process went like this.

A master recording was made on the recording machine, next the
appropriate coding was added.  Once the coding was added, a number of
trial cuttings were made.  These copies were used to check length of
notes, straighten out flubbed trills (no one's perfect), and before the
spark chronograph (and possibly after) to determine the best way to
replicate the playing of the artist.

From my understanding, there were usually around three trial cuttings
made, with the first, I would imagine, having the greatest number of
corrections and the last, the least number of corrections.  I would
imagine that that could be changed if the performance was absolutely
perfect and they only needed one trial cutting, or conversely, if the
artist didn't like his performance and wanted to "experiment" and make
it better, they could have any number of trial cuttings.  There were
also trial cuttings made for the addition of words.

I have a trial cutting for the "Song of India" by the Original Piano
Trio which was made with "B" coding.  I believe this one to be somewhat
of a rarity in that not many popular recordings with "A" coding were
reissued with "B" coding.  Usually once a popular roll had seen its
moment of fame, the master would be tossed in the furnace and more room
would be made for the new titles.

I'm sure there are other voices who can elaborate more.

Mike Walter