The excerpt below is from an article I wrote and published in PM several years ago:
Back plate 629 first went to press late in 1947 just as the M-A serial block for $5 silver certificates was ending. The early sheets landed on the numbering presses just in time to get those serials. The rest of them received N-A block serial numbers. The same sheets were also being routed to presses having Series of 1928E $5 legal tender and Series of 1934C $5 Federal Reserve note face plates.
Back plate 629 has an interesting history. It was finished in normal fashion on December 29, 1933, as one of the last old-gauge $5 backs made by the Bureau of Engraving and Printing. The gauge referred to the width of the gutter between the two columns of subjects on the plate.
The Bureau began increasing the gauges on small-size plates starting with $1 faces in July 1934. The wider gutters on the new-gauge plates allowed plate printers more tolerance when mating printed backs onto face plates.
All $5 back plates were new-gauge starting with back 630 certified on January 31, 1935. The Bureau typically rushed them to press and earmarked unused old-gauge plates for cancellation. Five-dollar backs 575 to 628 met that fate, but 629 was saved as a model of the former registration. Up to this time, it had never gone to press.
Beginning in 1938, the Bureau started salvaging usable master plates for conversion into production plates. These salvaged plates, formerly known as late-finished plates, also included $1 back 470, $5 back 637, $5 silver certificate face 307, $10 silver certificate faces 86 and 87, and $20 back 204.
To finish them as production plates, plate technicians etched plate serial numbers onto each of the twelve subjects on the plates, certified them, and sent them to the plate vault for eventual use. Because of when they were finished, they carried plate serials that were out of sequence with contemporary plates.
Back 629, though, was not a true salvaged plate. A common thread among the other salvaged plates is that they were new-gauge plates. Because 629 was not shows that it was actually used by mistake. In fact, likely after it was pulled for safekeeping someone had penned in large letters, "Do Not Send To Press," at the top of the plate ledger page for the plate.
Regardless, 629 went to press for one short period from November 17, 1947 to February 2, 1948--14 years after it was certified. Its short press-time shows that someone quickly realized it was an old-gauge plate and promptly removed it from service. It was canceled on February 17, 1948.
Most observed $5 silver certificate 1934C 629 mules have serials within the first 8 million numbers of the N-A block. The other M-A note has serial M99064292A. Considering that 629 sheets comprised an extremely small percentage of $5 backs at the time, and were being parsed among three different classes, it is not surprising that 1934C 629 M-A notes are so rare. Who will be the lucky person to report the third note?