^{page 381 note 1} The one exception, mãjār, is only apparent. Here the ã represents a theoretical ã, which has been shortened owing to its position before a long vowel with the stress accent.

^{page 382 note 1} Prakrit grammarians treat these two as interchangeable. See Mārkaṇḍëya, iv, 24, where we have, e.g. both jalaṁ manthaï and jalam maṁthaï. Cf. Pischel, § 269.

^{page 382 note 2} In the above I have omitted other changes, such as k gan, for k kan, or cān, for c d. As regards the former, see note 2, p. 385, below.

^{page 383 note 1} See ZDMG., 1, 1896, pp. 21, 22.

^{page 385 note 1} Bhavisatta Kaha, p. 63*.

^{page 385 note 2} Cf. Skt. kaṅkaṇa-, Hindī k gan, referred to in note 2, p. 382, and similar instances in the modern languages of softening after a nasal.

^{page 385 note 3} Here it is the form with the nasal which is original, and jujjaï which is the by-form. Similarly we have maggaï < maṅgati. From this we learn that Dēśī dialects looked upon ṁj and jj as interchangeable.

^{page 385 note 4} Jacobi, op. cit., p. 63*.

^{page 386 note 1} An edition, taken from the same unique MS. as that from which mine was copied, has lately been published as No. XI of The Kashmir Series of Texts and Studies.

^{page 387 note 1} La Nasalité en Indo-Aryan, p. 67.