Three small blades compared by what they ask of your hand: grip length, scabbard fit, cast fittings at 60–90 mm, and what a yoroidoshi was built to...
GlobalQuestions
Reading a Mark: Inscriptions on Blades and What They Prove
What a signature, mei or stamp really proves about a blade — and why marks get faked.

Contents
By Hong Kong Miniature Armor Co. Ltd. · Last updated October 2026
A maker's stamp on a Japanese blade is about 4 mm across, with strokes 0.2 mm wide, and the whole attribution can hang on whether the second character has a crossbar. Photograph it badly and you have not recorded a mark, you have recorded a smudge.
The sections below cover the families of marks, what a mark actually proves, and the tells of a fake. The bench section is the imaging arithmetic: how many pixels a stroke needs, why stamping fails at small scale, and what a replica should never carry.

The families of marks
On a Japanese blade the mark is the mei, cut into the tang (the nakago) rather than the blade face. It normally names the smith, his province and sometimes the period — a Met conservator, handling a 17th-century blade, read aloud a mei citing Bizen Osafune, one of the great sword-making centres of Japan. The tang also carries the smith’s file marks (yasuri-me), whose pattern is part of the attribution; a square photo of the tang that hides those marks has missed half the evidence. Reading a mei is a specialist skill in classical Chinese characters and provincial abbreviations, and a rough “translation” from a forum is worth less than a careful transcription.
European blades carry different marks for different purposes. A maker’s mark identifies the smith or workshop; an arsenal stamp records acceptance into a state armoury; a proof mark (common on firearms and some edged weapons) records that an object passed a test; a regimental or unit number records issue to a body of troops. A blade may carry all four, laid down across centuries of use, and the date of the object is the date of its earliest Consistent feature — not the latest stamp someone added.
Reading order matters. A blade with a smith’s mark and a later arsenal stamp is two documents, and the younger stamp cannot date the older blade — yet a casual post will read the pair as one fact. The discipline is to separate layers: what the maker put there, what the state added, what a unit added, and what a later hand may have changed. Each layer has its own literature, and each can be checked against a different set of comparables. Collapsing them is the single most common error in mark-reading threads.

What a mark actually proves
A genuine, correctly placed mark proves one thing cleanly: that the feature it sits on was, at some point, associated with the named maker, arsenal or unit. It does not Prove the whole object is original, contemporary or undisturbed. A 15th-century blade can carry a 19th-century arsenal stamp; a genuine blade can sit in a later hilt; a real mark can be cut onto a fake. The mark proves the mark, and the rest has to be argued from construction and context.
Consider the mei again. A genuine mei proves the tang was signed, normally by or for the smith; it does not prove the blade was forged by that smith, because blades were shortened, re-signed, and re-fitted across centuries, and because some genuine smiths signed blades finished by their school. The signature is a strong clue to origin and period, not a birth certificate for the steel. Treating it as the latter is how attributions drift from “attributed to” into “by,” a slide the careful community corrects.
Marks also degrade and migrate. A mei can be shortened when a blade is cut down, or re-cut by a later hand; a European mark can be partly ground away by centuries of sharpening. Reading what survives accurately — and admitting what is lost — is more honest than supplying the missing letters from hope. When a mark is incomplete, say so; a thread that reconstructs a mark from three comparable examples is good scholarship, while one that fills the gaps from a wish is not.

Why marks get faked, and how to spot the tells
Faking marks is old, organised and profitable, which means a mark demands suspicion before trust. The tells are consistency and context: compare the mark’s letterforms, depth, placement and spelling against documented examples of the claimed maker, because fakes frequently drift in exactly these details. A mark that is too clean, too deep, or in the wrong place for its tradition is a warning. The Metropolitan Museum itself documents historical armour that was later “improved” with a purpose-made punch to falsify a maker’s mark on a genuine but previously unmarked object — proof that even real antiques can carry dishonest marks, and that the mark is always a question, never a full stop.
The economic motive is worth stating plainly, because it shapes how suspicious to be. A plain blade with a famous name added can sell for many times its honest value, and the same is true of a regimental number that implies a desirable campaign. The more a mark would raise the price, the more carefully it should be checked — against comparable examples, against the known spellings and placements, and, for high-value pieces, against expert opinion. Suspicion here is not cynicism; it is the normal caution the market itself demands.
Photographing a mark well is half the reading. Shoot it square and in focus, then shoot it again under raking light and, where possible, under oblique raking light from the opposite side, because incised and etched marks reveal their depth differently and a flat flash flattens both. A macro or a phone’s close-up mode is enough; what matters is that the depth, the burr and the wear of the cut are visible, because those are exactly the details a faker gets wrong. Post both shots and let the thread judge the cut, not just the letters.
Re-engraving is another trap. A later owner might add a famous name to a plain blade to raise its value, or a restorer might “restore” a worn mark into something legible but wrong. The honest reading names the mark as it is and flags any sign that it was added or altered after the object left the maker.

A 4 mm stamp with 0.03 mm strokes
Start with the resolution the stroke demands. A 0.2 mm stroke needs at least two pixels across to be read, so 10 px/mm; on a 4 mm mark that is 40 pixels, which any phone can deliver if the light is right. The failure is rarely resolution and almost always contrast: a burnished blade reflects the light source straight into the lens and the stroke disappears. A single diffused source at 30 degrees, with a black card on the opposite side, gives the relief something to cast a shadow into.
Scale that mark to a 1:6 replica and the stroke becomes 0.03 mm, which is finer than a laser etch resolves cleanly and finer than any stamp can form: the metal flows and closes the groove. A stamped mark at 0.67 mm total width fills with polishing compound within a week and reads as a dent. The only workable method at that size is photo-etched resist and a 0.05 mm etch, and even then the mark is legible only under magnification.
Which is why we do not put marks on replicas at all. A copy of a maker's signature is not a decorative detail; it is the one feature that turns an honest replica into a forgery claim, and the scale argument is not a defence anyone will accept at resale.
The judgement call: on an original, a mark is a lead, not a verdict. Record it, photograph it in raking light, compare the stroke order with a published reference, and then check the steel, the shape and the wear. A correct mark on a blade that is wrong in three other ways is a mark that was added later.
How to read a mark against the record
The discipline that turns a mark into knowledge is comparison. Transcribe it exactly, photograph it square and in raking light, then check it against published mark dictionaries and museum catalogues rather than memory. The Metropolitan Museum’s Arms and Armor department And the Royal Armouries collection database Both publish documented marks and signatures you can set beside your own, and specialist references exist for Japanese smiths, European proof marks and regimental stamps alike.
Transcription discipline is the unglamorous half of the skill. Write the mark exactly as cut, including what is worn away, and resist the urge to complete it from hope; a missing character is data, not an inconvenience. Then check the transcription, not your memory, against the published record — mei dictionaries for Japanese blades, proof-mark guides for European firearms and edged weapons, unit histories for regimental numbers. A thread that opens with a careful transcription and a photo almost always reaches a sound reading faster than one that opens with a translation.
Post a mark as a question with its photo, name the family it belongs to, and ask the community to check it against a specific comparable example. The best mark-reading threads here end with a cited accession number or a mark dictionary entry, not a flat claim. Add your own examples in the comments, and bring a mark you would like the community to read properly.
A Met conservator reads the mei on a 17th-century blade aloud — a clear example of how a signature on the tang is identified, and why the mark is evidence to be checked rather than a verdict.
FAQ
What does a signature on a Japanese blade prove?
That someone cut that inscription, at some point. It does not prove the smith made the blade. Genuine signatures were added by appraisers, and later hands added famous names to ordinary blades, so the mark has to agree with the work.
How do I photograph a blade mark?
One diffused light at about 30 degrees to the surface, a black card opposite to fill the shadow, and the camera square to the mark. Fill the frame with the mark and take a second frame with a rule for scale.
Should a replica carry a maker's mark?
No. A reproduced signature is the feature that makes a replica pass as an original, and at 1:6 it cannot even be executed properly. Mark the base or the stand instead, where the mark identifies the replica as a replica.
The research and bench team behind miniaturearmor.com. We build 1/6 and 1/12 armour and weapons in our own workshop in Zhuhai — assembly, CNC and 3D printing under one roof — and every piece passes through our hands before it is packed. These notes come from the bench: fit, finish, rust, and the compromises scale forces on a design.
Keep reading in the Miniature Armor Library
- Japanese Armour: Ōyoroi, Dō and Tosei Gusoku
- Collection Stories: What a Collection Says About Its Owner
- Display Setups: Light, Dust and Scale
- New Acquisitions: How to Photograph a Piece Properly
- Myth vs Evidence: 'Damascus Steel' and What Metallurgy Says
Choosing your first piece: start with the Roman Helmets, Japanese Weapons, Chinese Armor & Weapons Or European Armor Collections. For long-term care, our care & maintenance guide Covers cleaning and storage; and if a question needs a human, write to us — the team that answers is the team that packs your order.
Join the Discussion
Have a question, correction, historical reference or collector insight? Join the discussion below.
Please keep discussions respectful, relevant and historically focused. Spam, unrelated promotions and unsafe content may be removed.Community guidelines.
No replies yet
Be the first to add a question, a correction or a collector insight about this topic.