Counted properly, the traditional Mongolian alphabet is thirty-five letters, and the count only means something once it is said what has been counted. Twenty-five are the inherited core, taken from Old Uyghur when the script was adopted around the founding of the Mongol state, and sufficient to write the Mongolian language: seven vowels sorted by harmony, and the consonants of the native system, with the ambiguities the language itself resolves. Ten were added afterwards for sounds met in Chinese, Tibetan and Sanskrit vocabulary, moving from single consonants to a cluster and finally to two syllables.
Beyond those thirty-five lies the Ali Gali apparatus, encoded separately, for writing Sanskrit and Tibetan in full. And beyond that, in the same Unicode block and built on the same shapes, are Todo, Sibe and Manchu, which are separate writing systems and no part of this alphabet however similar their letters look. Any figure quoted without those boundaries is unreliable, and figures for this script are quoted carelessly rather often. The alphabet proper is a finished object with a clear internal history: an inheritance, an extension, and a boundary where other systems begin. That the last letter of it stands for a Chinese syllable rather than for a Mongolian sound is a fitting place to stop, since it says as plainly as anything could that this alphabet has spent eight centuries writing more than one language.