Dating a band instrument

Three makers out of 146 publish serial figures for drums, brass or woodwind. All three publish something broken. Not approximate, not vague — internally contradictory, in ways that make a confident lookup return a confident wrong year.

Maker	What they publish	What is wrong with it
Conn-Selmer	18 tables — Conn, Bach, Selmer Paris, Artley, Armstrong, Benge, Leblanc	Two rows have their columns swapped
Ludwig	A drum serial guide, 1961–1979	Two serial ranges are assigned to two different years each
Gemeinhardt	509 rows of flute serials across 104 models	Ten ranges run backwards — last lower than first

All three tables are withheld, and this page cannot date an instrument today. None of these three makers publishes terms of use at all — which is not permission, it is the absence of an answer. This site asks before it reproduces somebody else's compilation, and the letters are outstanding. Until they are answered the 818 rows are withheld rather than deleted, and the tool below will tell you which maker's own page to read instead.

What stays is the part that is this page's own work: exactly what is wrong with each of their three tables, set out below, so you can check their page against it yourself. The day a letter comes back yes the rows return and the tool works again — they are a revert away, not a re-derivation.

This one runs on your own device. It is worked out in your browser, so there is nothing to send and nothing to log. Where a tool on this site does send something out, that page says so and names where it goes.

Ludwig — the same serials, two different decades

Ludwig publish a serial guide covering 1961 to 1979. Two of its rows appear twice, against different years:

Serial range, as Ludwig print it	They assign it to
`450XXX – 548XXX`	1967 and 1978
`549XXX – 665372`	1968 and 1979

That is not an approximation or a range — it is the same numbers listed twice, eleven years apart. A drum with serial 500000 is, on Ludwig's own published evidence, either 1967 or 1978, and nothing in the guide distinguishes them. A tool that walked the table top-down would return 1967 and never mention 1978; this one returned both, until the table came off the page on 14 September. These two rows stay because they are the evidence for the criticism — the duplication is not visible with one of them — and because two years for one serial is the page refusing to answer rather than answering.

And there are holes. Four years in the run have no row at all — 1973, 1974, 1975 and 1977. Three more rows carry no serials: 1961–62 is “No number”, 1963 is “No Number – 7900”, and 1970 reads “(Out of sequence, unnumbered)”. Ludwig are being straight about that: those drums were not serialised in a way the guide can express.

Gemeinhardt — detailed, and ten rows run backwards

Gemeinhardt publish 509 usable rows across 104 models, from 1910 to 2012, with first and last serial for each model in each period. It is the most detailed table any maker on this page publishes.

Ten of those rows have a last serial lower than the first. The range runs backwards:

Model	First	Last	Year
M1	25037	24523	1958
53SB	551215	55258	2006
4S	67130	62179	2008

Those are three of the ten. The 53SB row is the clearest: 551215 to 55258 is not a range at all — one of the two numbers has a digit missing or added, and there is no way to tell which from the outside. This page flags those rows rather than swapping the numbers round, because swapping is guessing which one is the typo, and a guess that looks like data is worse than a gap.

Conn-Selmer — eighteen tables, two rows the wrong way round

Conn-Selmer publish serial tables for their whole portfolio on one page: C.G. Conn cup- and reed-mouthpiece instruments and saxophones, Bach trombones, trumpets and cornets, Selmer Paris saxophones, Artley flutes, piccolos and clarinets, Armstrong flutes, Benge and Leblanc. It is far and away the largest body of maker-published dating data in this family.

In C.G. Conn Cup Mouthpiece Instruments (1917 – 1956), every row up to 1954 reads serial first, year second:

As printed	Reads as
`427,301 \| 1954`	serial 427301, year 1954 — correct
`1955 \| 500,001`	year first — the columns are swapped
`1956 \| 571,850`	year first — the columns are swapped

A decoder that trusts the column heading reads 1955 as a serial number and 500,001 as a year. The parse behind this page decides which value is which per row, by what the values are, and then checks each row against its own table's majority. Any row that disagrees is reported rather than quietly corrected — because a disagreeing row is either their mistake or mine, and those need different fixes.

How the other makers were checked — and one it missed

The pool is 146 makers of drums, percussion, cymbals, brass, woodwind, harmonica and bagpipes, listed in band-makers.txt. Websites came from Wikidata's official-website property rather than guessed from company names.

Outcome	Makers
Site swept — robots.txt read and honoured	74
Refused us, or did not answer	18
No official website recorded at all	54

And read the swept figure knowing how much was there to read. Of the 74 counted as swept, 28 offered this method fewer than 25 pages to look at, and 26 of those offered nothing from a sitemap at all. A host that answers, allows us, and then has almost nothing to search still counts as swept. It is the number of doors opened, not the number of rooms searched.

The sweep missed Ludwig, and that is worth more than the sweep. Every page on this site says “nothing found” means this method found nothing, not that the maker publishes nothing. Here is the proof, on this page: Ludwig do publish a serial guide, the sweep came back empty for them, and their guide is one of the three the tool above runs on. The reason is checkable — ludwig-drums.com serves no sitemap, so the sweep fell back to walking the home page two levels and never reached /en-us/ludwig/about/serial-guide. It was found by hand instead.

So take the 74 as swept, not as cleared. At least one of them publishes something this method would not see.

Refused, or unreachable — named rather than counted

Six told us not to, and that was honoured. Dixon Drums, Hoshino Gakki and Rogers Drums name our crawler in robots.txt. Yamaha, Yamaha Drums and Lyon & Healy return 403 on robots.txt itself, so their rules could not be read and the answer defaults to no. Nothing was taken from any of them.

Twelve more did not answer or served errors, including Pearl Drums, Gretsch Drums, Yanagisawa and Antoine Courtois. That is a failure to reach them, not a finding about them, and it is recorded as unassessed.

Where to go next

Other instruments, dated the same way:

guitar serial numbers — ninety-one makers, and what each one publishes
piano dating — where the name on the front may not be the maker

Wikipedia has no saxophone category. None.

Building the maker list turned up something worth knowing if you ever try this yourself. Categories that exist for this family: percussion, brass, clarinet, oboe, cymbal, harmonica, bagpipe. Categories that do not exist, each checked individually:

Saxophone manufacturing companies · Flute manufacturing companies · Bassoon manufacturing companies · Trumpet manufacturing companies · Drum manufacturing companies · Woodwind instrument manufacturing companies

The saxophone is the instrument in this family people most often want dated, and it has no category at all. Yanagisawa, who make some of the best saxophones in the world, are filed only under “Musical instrument manufacturing companies of Japan”. A pool built from categories would contain almost no saxophone makers and would look complete. So they were added by name, each checked against Wikipedia first.

The list also needed a filter for articles that are not companies: 24 were biographies rather than firms, most of them bagpipe makers, who genuinely are individuals. One needed an exception written down — Vic Firth, where Wikipedia has no article for the drumstick company, only for its founder.