Sources · who’s who in the ledgers
The records behind the dates
Every date on this site is cited to a named source, and the same names keep coming up: the Polishing Room ledgers, “CFM” or “Cody,” and Madis. If you’re wondering who or what those are — and why they can put the same serial number in different years — this page is the answer.
The factory ledgers at Cody
Winchester numbered its receivers in sequence, and clerks in the factory’s serialing department — the Polishing Room — logged the day each serial number was applied, month, day, and year, gun by gun. Those ledgers are the closest thing to a primary record of when a Winchester existed that survives today.
And they did survive. When Winchester’s museum collection left New Haven for Cody, Wyoming in the mid-1970s, the surviving factory records went west with it. Today they are held at the Buffalo Bill Center of the West — home of the Cody Firearms Museum — in its McCracken Research Library. When our tables cite “CFM,” “Cody,” or “Polishing Room records,” this archive is what they mean. Collector organizations like the Winchester Arms Collectors Association (WACA) have published revised date tables drawn from these books, and some of our series cite those too.
Two caveats keep the ledgers honest. A serial-application date is not a completion or shipping date — most receivers were assembled soon after they were numbered, but some sat for months or even years. We measured the gap from the letters: the median lettered gun shipped 45 days after numbering, and 13% took more than a year. And the books don’t survive for every model or every serial range; where they stop, later compilations take over.
For a specific gun, the Cody Firearms Records Office will search the original books for your serial number and issue a factory letter with the recorded dates and configuration — the documentation auction houses and appraisers actually accept.
George Madis
For most of the twentieth century the factory ledgers sat in archives, and the standard reference was one man’s work. George Madis was a Texas arms historian who spent decades researching Winchester — working from factory sources and interviews with former Winchester employees — and published The Winchester Book in the early 1960s, followed by The Winchester Handbook and The Winchester Model Twelve.
The serial-number tables in his books became the dates collectors used. They have been reprinted in value guides, forum posts, and nearly every serial lookup site since. If a website has ever told you the year your Winchester was made without naming a source, the number very likely traces back to Madis.
What his tables are: estimates of annual production, compiled decades after the guns shipped — not transcriptions of the Polishing Room ledgers. And his coverage ends where his books do: for most models his tables stop in the early 1960s, and nothing extends past the early 1980s. Models introduced after that era have no Madis dates at all — they are dated from other records entirely, and our tables say so. For some models and eras his figures track the factory books closely; for others they drift apart by years. The famous case is the Model 1894, where Madis ends 1898 at serial 147,684 and the factory ledgers end it at 53,941 — a gap of nearly 94,000 guns that decides whether a rifle counts as a pre-1899 antique — the federal line that makes an antique of any gun made before 1899, drawn for every pre-1900 model on the pre-1899 cutoff page.
None of this is a knock on Madis, who did the work when the archives weren’t open. But it is why we label his figures as his, keep them separate from the ledger figures, and never average the two.
Winchester’s own lists
The third recurring source is the company itself: typed year-by-year serial lists that Winchester’s customer-service desk kept on hand to answer “when was my gun made?” letters. They read as official — they came from Winchester, after all — but they are a mix of genuine records and after-the-fact arithmetic, and some entries are marked as estimates in the documents themselves. We cite them as customer-service lists and carry those caveats through to the tables.
How this site treats them
We don’t arbitrate. Each source is transcribed separately, rendered side by side, and where they conflict the range is flagged in red with the stakes spelled out. The Model 1894 page is the worked example — three record systems, one serial number, and answers that differ by years.
The disagreement has a shape of its own. Madis vs. factory dates lists every model where the sources part ways, with a real serial number from each, and explains why one book estimates while the other transcribes.