Skip to content
The Encyclopedia of Horology

Identification · 3 min

How to Read a Serial Number Table Without Misreading It

Serial tables are the most-used and most-misused tool in watch identification. Understanding what a table actually records — and what it cannot — prevents the commonest dating errors.

Related photographs

Hamilton 992B, identified in the source as a 1944 example — exposed 21-jewel movement before workshop servicing
Hamilton 992B, identified in the source as a 1944 example — exposed 21-jewel movement before workshop servicing — HandyDan / Vintage Hamilton Watch Restoration (Hamilton Chronicles), Copyright workshop photograph — rights pending; reduced-size JPEG copy, permission not established. Hamilton Chronicles — exact 992B restoration record
Dial and exterior case of a Hamilton 992B railroad pocket watch
Dial and exterior case of a Hamilton 992B railroad pocket watch — Jack Stephens / Watchfinder & Co., Rights pending — permission not yet requested. Web page (watchfinder.com)
Hamilton 992B railroad pocket watch dial and case exterior
Hamilton 992B railroad pocket watch dial and case exterior — Jack Stephens / Watchfinder & Co., Rights pending — permission not yet requested. Web page (watchfinder.com)
Hamilton 992B railroad pocket watch movement
Hamilton 992B railroad pocket watch movement — Jack Stephens / Watchfinder & Co., Rights pending — permission not yet requested. Web page (watchfinder.com)
Hamilton 992B railroad pocket watch exterior with Ferguson-style dial
Hamilton 992B railroad pocket watch exterior with Ferguson-style dial — Jack Stephens, Rights pending — permission not yet requested. Web page (watchfinder.com)

No confidently matched photograph of this subject has been found yet. These photographs show a closely related record, and are not examples of the type described on this page.

A serial number table maps ranges of movement numbers to approximate production dates. It is a summary of surviving evidence, not a factory ledger: most published tables were compiled by researchers from observed examples, partial records and trade literature, and they carry error margins that the table itself rarely states. Treat every date a table gives you as 'about this year', not 'this year'.

The first rule is to check what the number you hold actually is. American watch cases were usually made by separate companies and numbered in their own sequences, so a case-back number tells you nothing about the movement's date. Swiss and English practice varied; where a case carries a hallmark, the hallmark's date letter is the better evidence for the case. Use the number stamped on the movement's top plate or bridges, and photograph it next to the maker's signature so the transcription can be checked later.

Read the table's own introduction before its numbers. Good compilations state what their dates mean — the year a run of movements was started, finished, sold or shipped — and these are not the same event. Some makers numbered in blocks allocated to grades before production, so a low number can be finished later than a higher one. A table that does not say what its dates record should be used with extra caution.

Watch for the known traps. Numbers were occasionally duplicated, restarted or skipped. Some makers used the same sequence across very different grades, so the number dates the run, not the model. Replacement plates and swapped movements can put a genuine number on the wrong watch. And a date from a table is a production estimate: the watch may have sat in stock, been cased later, or been sold years afterwards.

Cross-check the result against the watch itself. If the table says 1905 but the movement's layout, jewel count or damaskeening pattern matches documented examples from the 1920s, believe the conflict and investigate: you may have misread a digit, used the wrong maker's table, or found a replacement movement. A serial date that agrees with the case hallmark and the dial style is worth much more than any one of them alone.

Record the result reproducibly: maker, full serial as stamped, the table used, the date range it returned, and what kind of date that table claims to give. If two published tables disagree, keep both and say so. The goal is a defensible 'circa' date with its evidence attached — not a single confident year that the underlying records cannot support.

Archives and marks behind the tables

Waltham's own records, for instance, are held at Baker Library, Harvard Business School, where the finding aid notes two typewritten company histories by E. A. Marsh, of about 1889 and about 1921, available on microfilm. The same finding aid records that one volume of sales and credits for 1915 to 1917 was reviewed and deaccessioned in 2026 because of conservation problems. The NAWCC's index of dating aids lists serial-number-by-year tables for makes including Waltham, Ball, Elgin, Hamilton, Illinois, Rockford, Seth Thomas and South Bend, alongside a separate production list for several US makes.

Hallmarks offer an independent check on a case, and Sheffield Assay Office explains how their dating works: since 1975 all four UK assay offices have used the same date letter, changing on 1 January each year, and it has been a voluntary mark since 1999. Each 25-year alphabet uses a new typeface, case and shield shape so that a letter can indicate only one year, and look-alike letters are omitted, which in the last set meant leaving out i and j.

Written from the guidance and records listed under Sources, credited to each publisher and put in our own words; each statement was checked against the cited page.

What the number can and cannot date

What the number can and cannot date
Number or recordBest useCommon mistake
Movement serialFind a maker-specific production or run estimate.Using it as a case, sale or ownership date.
Case numberResearch the case maker when records survive.Entering it into a movement table.
Grade/run recordCheck whether the serial fits the stated model variant.Assuming all tables record the same factory event.
Hallmark or invoiceAdd a separate date witness.Forcing every component into one year.

Evidence checklist

  1. Identify which component carries the number.
  2. Name the table or ledger used.
  3. Record approximate ranges and disagreements.

A useful serial result states source and date type; it does not silently authenticate the whole watch.