As an update to Xeramic former post he has completed the list below a bit after screening his stuff (and added also some net findings with source):
pin connector, large notch
3/60
3/60
3/60
4/60
1/61
1/61
2/61
bar connector, large notch
1/62
1/62
1/62
2/62
4/62 cda.chronomania.net
4/62 http://www.chrono-shop.net/en/1359-rare-7206-rolex-tudor-1962-vintage-20mm-buckle-clasp-from-submariner-5512-5513-1680-1665-and-1675-6542-5508-1016.html
1/63
1/63
1/63 http://www.network54.com/Forum/207673/message/1349172687
2/63 (only notch visible, last pic shows other bracelet)http://www.network54.com/Forum/207593/thread/1424690750/last-1424954133/View+All3/63
bar connector, narrow notch
2/63 eBay
2/63 (only notch visible) http://www.network54.com/Forum/207673/message/1465380840/FS-+Rolex+5512+-62+Gilt+chapter+ring+excl+dot
2/63 (6636, only notch visible) http://www.network54.com/Forum/207673/thread/1418772001/FS-+--------+Rolex+6636+Rivet+Stretch+bracelet+80+ends+---+Date+2.63+---+12+Links+-----
3/63
1/64
1/64 http://www.network54.com/Forum/207673/thread/1431376407/FS-+1964+Rolex+7206+20mm+rivet+bracelet+with+58+ends+%26gt%3B%26gt%3B
2/64
2/64
3/65
1/66
2/66
1/67
2/67
4/67
2/69
He mentioned in the last post a 3/63 still with large notch, but this was a misreading (was 2/63).
Regarding the cut off date of the reference stamp: I've found in 1963 only one (1/63, the network sample mentioned above) within ten pieces (not all listed above), so that one has a changed bracelet side or stamped piece for sure. In contrary I've found in several 7206/6636 from 1964 only one without the stamp (4/64), so that's obviously a swap too and earlier information about starting with the reference stamp in 1964 (1/64) seems correct, and all other combinations are very likely swap results.
As an overview (based on my observations only, of course - the question marks show missing information or, in parentheses,possible transition not yet excluded):
Rolex 7206 buckle dates
I didn't mention transition possibilities for the reference stamp because my finding matches earlier information around. Anyway it is not 100 % safe that the start was in 1/64, and it may also exist transition quarters. Others may add their own findings.
Below I'd like to show that the initial topic, the 61 vs. the 67 readability, is only a problem with the 4/67.
Best,
Xeramic
