How to Read Amulet Inscriptions: Script, Language and Dating Evidence

Unread writing on an amulet is often labelled simply ‘prayer’ or ‘yantra’. It may instead record a personal name, donor, dedicatory wish, religious stanza or commemorative event; sacred use does not make every line the same kind of statement.

Read in the order of script, language and content before asking what the inscription says about date. Reverse the order and an old-looking script becomes ‘an ancient object’, or a named master becomes ‘the maker’. The Sirindhorn Anthropology Centre database keeps these fields apart and offers a useful model.

Script and language are different questions

Chinese characters can record names or transliterations, and Latin letters can write many languages. Ancient-looking letters therefore do not reveal whether the language is Pali, Sanskrit or Old Mon. U Thong tablet 7 records Pallava script, Pali language and number 16/2513; the Ban Khu Bua fragment records post-Pallava script, Sanskrit and number รบ. 7. ‘How written’ and ‘which language’ remain separate.

Calling every unfamiliar form Khmer can confuse script, language and history at once. Preserve the image, location, line count and losses, then consult published readings. ‘Two lines of old writing on the reverse; language undetermined’ is better than a confident guess.

A familiar stanza can exist in different languages

The dependent-origination stanza is often referred to in Thai sources as เย ธมฺมา. Objects may present it in different languages, scripts and layouts; a complete modern printed version cannot be projected onto every ancient tablet. The Ban Khu Bua fragment lacks its upper part and retains two lines. The database reads the surviving text as a Sanskrit fragment. ‘Two lines’ describes survival, ‘Sanskrit’ the language, and ‘fragmentary’ the limit of the reading.

The same stanza can be rewritten or reproduced in different periods. Familiar syllables identify a class of content, not necessarily an issue. Dating still requires letterforms, layout, object format and issue records.

One tablet can carry more than one writing system

Si Thep entry 1000 records Sanskrit in post-Pallava script on the front and Chinese on the back. Only part of the Chinese survives; a later study used a more complete comparator to propose a monk’s name. That is a comparative reconstruction, not a claim that every character remains visible. It also shows that a Thai find need not bear only Thai writing.

Keep three layers: characters directly visible in the photograph; tentative readings from comparable forms; and missing text conjectured from another version. If they remain separate, later revision will not destroy the original observation.

Dating evidence is not always a four-digit year

Metal tablet 107/22 in Chao Sam Phraya National Museum carries Chinese. On the basis of a reference to the Ming, the database proposes a range in the twentieth to twenty-second Buddhist centuries; it separately lists discovery in B.E. 2500. The former is a dating interpretation, the latter a modern event. Era names, titles, temples and persons can set boundaries but not automatically a manufacturing year, and reproductions can preserve old wording.

Do not convert a number on sight. ๑๐๐ may mark an anniversary, ๘๐ an age, or either may belong to an inventory. Surrounding words often determine meaning better than the digits themselves.

Thai numerals and Buddhist Era dates in practice

Thailand adopted 1 January as New Year from B.E. 2484; the change left B.E. 2483 with only nine months. Earlier documents, boundary dates and lunar notation cannot all be handled by mechanically subtracting 543. Preserve the original digits, original wording and conversion on the object record.

Do not fill losses with wishful readings

U Thong tablet 7 is cautiously described as possibly naming a disciple, but wear prevents certainty. Expectation of a famous master makes blurred strokes resemble the desired name. Transcribe visible characters, leave gaps, propose alternatives and test each against grammar and context. A sales title cannot translate the inscription without creating a circle: the title names him, therefore the letters resemble him; the resemblance then proves the title.

How a published Old Mon reading is built

In section one of ‘Notes on Mon Epigraphy II’, Christian Bauer studies three Old Mon votive tablets reportedly from Saraphi near Chiang Mai and reads names of three arahants. Object 24/2523 is read as Bhaddiya, while the spelling remains problematic. The paper publishes readings, compares forms and preserves unresolved points: a supported identification can coexist with uncertainty about a letter.

Votive figures should not all be assumed to be Shakyamuni. A named inscription may refine a broad visual classification; if a reading conflicts with the composition, recheck the photograph, number and surface. The transferable lesson is the argument: state what is visible, how it was compared and what remains unresolved.

Make inscriptions evidence that can be questioned

A useful record preserves the face of the object, line count, reading direction, visible characters, losses, source of the reading and reasons. Keep both sides. When readings differ, locate the disputed character and ask whether the newer proposal uses a clearer image, comparator or linguistic material. Later is not automatically better, but new inspectable evidence deserves reconsideration.

Collectors need not master every ancient script at the outset. Avoid equating script with language, a commemorative date with manufacture, or losses with desired names. The advanced question is simple: where exactly are these characters, and why are they read this way?