Start with construction, as in reading any unknown movement: plate architecture (full plate, three-quarter plate, bridge construction), the escapement type, and the winding and setting system each narrow the field to a broad period and region before any decoration or marking is considered. See the companion article on reading an unknown movement for this first pass.
Move to material and finishing evidence. Gilt (fire-gilt or later electroplated) finish, nickel or rhodium plating, the style and quality of damaskeening or Geneva stripes, the shape of the balance cock and whether it is pierced, engraved or plain, and the presence or absence of a regulator index or a free-sprung balance are all features whose common styles shifted over time and by region, though none is precisely dateable alone.
Examine jewelling: count, placement and whether jewels are set in plain holes, screwed chatons, or friction-fitted settings. Screwed gold chatons are strongly associated with higher-grade nineteenth and early twentieth-century work; friction-jewelled settings became common in mass production from the twentieth century onward. Treat this as supporting, not conclusive, evidence, since grade and market varied enormously.
Look for hairspring and balance clues: a bimetallic compensation balance with timing screws suggests nineteenth to early twentieth-century high-grade practice, while a monometallic balance (often described, where legible, as Glucydur-type or similar) with an Elinvar or Nivarox-type spring points to twentieth-century manufacture, generally interwar or later. See the companion article on hairspring metallurgy for the reasoning behind this indicator.
Check for any unsigned but still informative marks: numbers that may be caliber references, workshop assembly marks, import or duty marks (such as Swiss import marks used in some markets), case sponsor's marks even where the case is otherwise unmarked, and repairer's scratch marks inside a case back, which at minimum give a latest-possible service date.
Finally, weigh the evidence together rather than seizing on the first plausible answer, and write the conclusion as a range with its supporting evidence, for example: 'Swiss bar movement, cylinder escapement, gilt finish, friction-jewelled, consistent with export production of the 1880s–1900s; no maker's mark identified.' A precise-sounding single date without a signature, serial or hallmark to support it should be treated as a red flag rather than a finding.
Where real doubt remains after this process, record it as unresolved rather than guessing to fill a catalogue field; an honest 'not established from features examined here' is more useful to a future researcher than a confident date that later evidence contradicts.