Siddham Sa 𑖭

Sa is the dental sibilant, the third and commonest of the three, and it stands second to last in the Siddham order. It is the sibilant of the nominative singular ending, of a large set of prefixes and of a great many grammatical forms, so it is among the most frequently written letters on any page of Sanskrit.

Under stated conditions a rule converts it to the retroflex sibilant, and that conversion is where most retroflex sibilants in the language come from, so the two letters are related as a source and an outcome rather than as two independent sounds.

Its frequency is one reason a Siddham hand is often described by the way this single letter is drawn. A letter written thousands of times is written fast, and the shapes that result vary more between scribes, periods and copying traditions than the shapes of letters a copyist has to stop and think about.

In the Japanese esoteric schools the bare letter is the seed syllable of Avalokiteshvara in the plain form called Sho Kannon, distinguished by that syllable from the multi-armed and multi-headed forms, each of which has a seed syllable of its own.

The letter is also the one most often met at the very end of a word, since the commonest nominative ending closes on it, and in a manuscript that runs its words together without spaces that makes it one of the few reliable signals of where a word stops. Readers have always used it that way, and so do editors working on the manuscripts now.

Siddham was a manuscript and woodblock script for the whole of its working life, and it came late to movable type and later still to digital text. It now has a block of its own in the Unicode Standard, which means a Siddham text can be stored, searched and transmitted as characters rather than as pictures, and that a scholar can quote a syllable rather than reproducing an image of one.

Encoding a script of this kind involves decisions that are not neutral. The alternate letterforms found in the manuscript record have to be handled somehow, and the Siddham block carries several of them as separate characters, because a text cannot otherwise be represented faithfully. The dependent vowel signs, the nasal and aspiration marks and the cancelling stroke are encoded as marks rather than letters, which matches the tradition. Conjuncts are formed by the rendering software rather than stored as characters, which is how the whole Brahmic family is handled.

The practical result is that this atlas can carry the actual character for each letter rather than a description of it. Not every script is so fortunate: a living liturgical script with no Unicode allocation has to be shown as a drawn outline instead, with the character column left empty, because storing a lookalike from another script would assert an identity that does not exist.