not lagatâ, but lagtâ; वीजका bîjkâ; an aspirate media preceding a tenuis in बूझकर bûjhkar; and many others which would be inadmissible in Sanskrit. Such a process as altering the final consonant of a verbal base to bring it into harmony with the initial consonant of a termination, is quite unknown to the modern languages.
In those combinations which I have in Chapter IV. called the mixed and weak nexus, Bengali betrays some weaknesses. One of these, in which the semivowel व follows a consonant, has been noticed in § 23. Another is seen in compounds whose last letter is म. In this case the m is not distinctly heard, but gives a subdued nasalization to the preceding consonant, which is pronounced as though double. Thus, स्मरण is in Hindi smaran, but in Bengali it sounds shmoron; लक्ष्मी is not Lakshmî, but Lakhkhî; पद्म is not padma, but podda. It is almost impossible to express the exact sound of this nexus—it must be heard to be understood. In the words कृष्ण, विष्णु, the Bengalis and Oṛiyas in speaking substitute ट for ण, and the former add an anuswâra after the final vowel, so that these words sound in Bengali कृष्टं Krishtaṉ, विष्टुं Bishtuṉ, and in Oriya Krushto and Bishtu. Thus, too, the Sanskrit वैष्णव a "Vaishnava," a sect very common in Orissa, is corrupted into Boishnob, Boishtnob, Bastab, and even Bastam.
§ 27. Sindhi has four sounds peculiar to itself, or, to speak more accurately, it has four characters, ग़, ज़, ॾ, ब़, which are not used in any other language. Dr. Trumpp is of opinion that these four characters represent four simple sounds (einfache Laute); it is, however, evident from his own remarks that they are only methods of expressing ग्ग, ज्ज, ड्ड, and ब्ब, respectively, and the analogy of the Bengali pronunciation in the examples of nexus given above helps us to understand how these letters have come to be written with a single character, namely by the stress laid on the first in the effort to give its full value and
