You would think the manufacturer would be the last word on its own history. The files on this site say otherwise, repeatedly, and the reason is structural: a maker is a witness to its history, not the judge of it. Marketing departments round dates to clean years because anniversaries sell. A watch has at least three birthdays, the trade-show appearance, the catalog entry, the first documented sale, and the house is free to pick the flattering one. Renumbering eras blur handoffs. And corporate archives, assumed perfect, are frequently thinner than the serial-dated evidence sitting in collectors' hands. So when a family file here marks a date circa and records a dispute, read the fine print, because a surprising number of those disputes are between the watch and its own maker. Here are the best ones in the files.
The birthday party, held a year late
The crown jewel lives in the Grand Seiko vintage file. The 62GS, the first automatic Grand Seiko, is dated by the house itself to 1967; Grand Seiko's own materials say so plainly, and its anniversary reissues count from that year. The record says otherwise: the earliest 62GS references, the ones wearing the lion medallion caseback, were manufactured and sold in 1966, and the scholarship now dates the debut there, with one prominent publication drily suggesting the house's 1967 may simply be more convenient for issuing anniversary pieces. The disagreement has become so established that general reference works now carry it as a footnote, the house's date in the text and the record's date underneath.
What makes this the perfect specimen is that the dispute has a mechanism, and the mechanism is a fight about a word. The watch began life in 1966 as the Seikomatic Chronometer, self-certified to chronometer standards, until the Swiss certification authorities objected to a Japanese maker using the word on its own say-so. Seiko's answer was to rebrand the very same watch as a Grand Seiko partway through 1966. So the house counts from when the watch wore its name cleanly, and the record counts from when the watch existed, and both are describing the same object. A naming dispute in 1966 created a dating dispute that is still running sixty years later, which is the kind of sentence this site was built to write. The file records both years, marks the date circa, and, per the doctrine, takes no side. It merely notes that the casebacks were there.
The dial that outbids the paperwork
The Seamaster 300 file holds a rarer species: a maker disagreeing with its own product, in writing, at the moment of sale. The CK2913 of 1957 says Seamaster 300 on the dial; Omega's guarantee backed it to 200 metres, the company attributing the gap to the limits of its own testing equipment rather than the watch. Both numbers are the record, the dial's claim and the paper's caution, and the file quotes them side by side, because the argument between a watch and its own warranty card is not one an outside historian could have invented.
The crown's fuzzy founding
Rolex tells the GMT-Master's origin as a 1955 story, and the one unambiguous paper date agrees in part: the trademark filing of April 21, 1955. The watch's public debut, however, is placed by the record at the 1954 Basel fair, a year before the house's own telling, and the family file records the pair without flinching: shown 1954, papered 1955, narrated by the maker from the flattering end. The same fog rolls over the family's transitions, where the maker's clean dates meet the record's untidy ones: the 1675's end is 1980 or 1981 depending on the source, the Fat Lady's start is 1982 or 1983, the 16710's is 1988 or 1989. None of this is scandal; it is production overlap, dealer stock, and catalog lag, transitions being processes that official histories compress into moments. The file marks them all circa, which is not imprecision. It is precision about what is actually known.
What the guides inherit
The Seiko files also expose how these errors propagate, because the mechanism showed itself twice in one family. The founding 3180 ends in August 1963 by the serial record, while the guidebooks say 1964, and its successor explains why: the 43999 is documented entering production in August 1963 but not reaching the market until 1964. Production dates and marketing dates are different facts, and sixty years of literature has mixed them freely, each guide inheriting the previous guide's choice without knowing a choice was made. When a family file here disagrees with a respected guide by exactly one year, this is usually the machinery underneath: nobody is lying, two calendars are simply being read as one.
The doctrine, stated once
So the files run on three rules, and this dossier is the place they get stated in full. First, every date carries its precision honestly: exact means documented, circa means disputed or transitional, and a clean year in a maker's telling earns no exemption. Second, where claims conflict, the file records all of them, the house's version included, because the argument is part of the reference's history and settling it is not the file's job. Third, where a tiebreaker exists, it is the serial record, which is why this site keeps the lookups it keeps: serials are the one witness that was present at production and has no anniversary to sell. The trilogy dossier ended on the site's founding premise, that the references outlive everything. This one adds the corollary: the references also outargue everyone, their own makers included, and the file's whole job is to keep the receipts until the argument ends. It rarely does.