Every page in this catalog is built around a recorded figure. That works because mints and central banks are bookkeepers before they are anything else: they count what they strike and print, and the count survives. But walk any bourse floor and you will hear premiums quoted on things that appear in no ledger anywhere - varieties, die states, overdates, signature changes. The catalog needed a rule for these, and the rule is simple: where the record kept a separate number, the variety gets its own page; where it did not, the variety lives as a note on the issue that produced it. The line is not about importance. It is about what can honestly be counted.
The counted: varieties with their own ledger line
Some varieties were tracked from birth, and those hold full pages here with their own recorded figures.
The 1913 Type 1 and Type 2 Buffalo Nickels are the clean case: the Mint redesigned the coin mid-year to sink the wearing FIVE CENTS into a recess, and recorded each version's production separately at all three mints. Six figures, six pages, one year. The 1878 tail-feather Morgans work the same way, the record preserving the counts as Philadelphia revised the eagle.
Paper does it too. The 1935 $20 Large Seal and Small Seal each carry a printage - 200,000 against 802,700 - so each carries a page. And the new 1937 series delivers the best example in the catalog: when the Bank widened the $1 signature panel by 2.4 millimetres in 1938, the record kept both counts, 10,000,000 narrow against 460,000,000 wide, and the printer destroyed the narrow tooling for good measure. A variety with a birth certificate and a death certificate is not a footnote; it is an issue.
The whole 1937 series is really this principle at scale. The Bank recorded printages per signature pair, so a change of deputy governor - Osborne to Gordon to Coyne, each signing beside Towers - splits every denomination into counted populations, and the short Osborne window makes the keys: 70,000 hundreds against 4,170,000 Gordons, and the 15,000-note $1,000 that anchors the series. Nobody polished a die by accident here; a man changed jobs, and the ledger noticed.
The uncounted: premiums with no ledger line
Now the other side. In 1937 a Denver press operator polished a clashed die hard enough to erase the bison's foreleg, and the Three-Legged Buffalo was born - the most famous die accident in American numismatics, and a variety for which no count exists or ever will. The mintage on that page, 17,826,000, belongs to the 1937-D issue entire. The three-legger's population is whatever the grading services have certified, which is a census of survivors, not a record of production. The 1918/7-D overdate is the same story with a wartime excuse: a 1917 die repunched for 1918, the ghost of the old date surviving, and no ledger anywhere recording how many coins the die struck.
Canada's version lives on the 1947 dollar, where the 7 in the date comes pointed or blunt - two master tools, one recorded mintage, zero official split. And even within series that count beautifully, the counting has edges: the 1937 notes carry prefix sub-groups and other micro-varieties that the record splits only partially, so this catalog rows on the signature, where the figures are complete, and leaves the fragments to the notes.
Why the line matters when you are buying
Because the two kinds of variety fail differently. A counted variety fails by misattribution: a wide panel sold as narrow, a Gordon signature in a bad photograph passed as an Osborne. The defense is a clear look at the diagnostic - or a holder that names it. An uncounted variety fails by manufacture: if no ledger says how many three-leggers exist, nothing but the coin itself argues for authenticity, which is why the leg-grinders work the 1937-D and the mintmark-solderers work the 1926-S. The practical rule follows: for a counted variety, buy the attribution; for an uncounted one, buy the certification, because the certified population is the only census that variety will ever have.
The line, in one sentence
The mintage is the issue's number; the variety is the die's, the plate's or the pen's. This catalog prints the first, records the second, and refuses to blur them - because the moment a die state gets quoted like a mintage, somebody is inventing a number, and inventing numbers is the one thing a record must never do.