As I recall, it took some time before Scott recognized three date sizes. Originally, only two were listed. I have PS5s of all three, making an extremely difficult measurement unnecessary. On S111, the first 1 of 1995 appears under BULK ... on S222 & S333, the first 1 of 1995 appears to the left of the B in BULK. On both S111 & S222, the 9s appear close together ... on S333, the 9s appear further apart. I may have scans of these together ... if I do, I'll post them later today. Also, the date on S333 appears to be just a bit further below the B in BULK than the other two. If these date characteristics are true of all 2905s, then your photo shows two S111s followed by an S333.
Added: While I could not locate a scan of these on my computer, I did find that
ebay seller tbrstamps has such a scan. Something I noticed on his scan of the three varieties together is that the 1995 gradually appears a bit lower below the B in BULK as the plate number increases. My PS5s seem to confirm this fact. His scan also confirms my observations above about the location of the initial 1 and the closeness of the 9s.
Scan of my 3 strips with S111 on top, S333 on bottom:
![]()
Added: I could be wrong, but I think Bardo's medium date is actually another small date. I enlarged his scan and could find no difference in the date sizes. The 2905a date is about 6% wider than the 2905 date. The 2905b date is about 9.5% wider than the 2905 date.