Machine identification learning plan: a weekend drill

What to take away

Saturday morning: observe without searching

Set a rule for yourself: no internet until lunch.

Photograph the twelve-frame set. Transcribe every mark exactly, using square brackets for anything illegible. Describe the mechanism in your own words, including how the thread travels and what the bobbin sits in. Then take the six measurements: bed length and width, shank height, bobbin diameter and height, needle system if you can read it, and presser bar offset.

Write all of it out longhand. The friction is the point; it forces you to look twice, and it produces a record in a form that will still be readable when the phone it would otherwise have lived on is long gone.

Saturday afternoon: form hypotheses

Now list every candidate identification you can think of, with a reason for each and a reason against. Aim for three, not one. The single-candidate habit is what makes people force evidence to fit.

Rank them by how much they explain. A hypothesis that accounts for the shuttle type and bed size, and matches a patent drawing, beats one that accounts only for the badge. A patent date dates the patent, not necessarily the machine.

Write down, for each candidate, the one observation that would kill it. That sentence is the most useful thing produced all weekend, because it tells you exactly what to look for on Sunday instead of browsing at random. If no observation could ever kill a hypothesis, it is not a hypothesis, it is a preference.

Sunday morning: search the public record

Test each hypothesis against sewing-machine-specific references first. Compare the badge, construction, shuttle or bobbin arrangement, and controls with identified examples and histories from ISMACS (the International Sewing Machine Collectors' Society) and NeedleBar. Treat a visual resemblance as a candidate, not a confirmed identification. Keep notes on the exact pages and images you use with the guide to storing materials used for machine identification.

Read the complete serial mark exactly, including any letters and leading zeros, and search for a register or serial-number table for the identified maker. ISMACS publishes Singer serial-number tables; compare the full number with the applicable table and report only the date or factory information the entry supports. Do not use a Singer table to identify another maker's machine. Federal publications can add context: the Government Publishing Office's govinfo service makes United States government information searchable and browsable by collection, date, committee and author, which covers a very large body of official material at no cost.

For patent clues, search USPTO Patent Public Search or Google Patents using an exact patent number, inventor name, or company wording transcribed from the machine. Compare the patent drawing with the machine's mechanism; a patent date does not by itself establish when the machine was made. Archival holdings are another route. The National Archives publishes highlights from its holdings location by location, noting that it preserves and provides public access to billions of records across facilities nationwide, including large photographic and graphic collections at College Park.

Neither of those is a sewing machine database. Both will teach you how to search a real record system, which is a skill that transfers to every catalog you will use afterwards.

Sunday afternoon: write the finding

Three paragraphs and a table.

Then put it away for a week and read it again. Most people find their confidence level was one step too high.

Repeat with a partner

The exercise doubles in value when two people work the same machine separately and compare findings. The evidence-gathering method is set out in our photography and presentation guide, and the tools each of you needs are compared in the identification tool article.

Common questions