Elements of South-Indian Palæography, from the Fourth to the Seventeenth Century A.D. c5.s1: § 1. Telugu-Canarese Alphabets

Elements of South-Indian Palæography, from the Fourth to the Seventeenth Century A.D. c5.s1

Language: English

c5.s1.t1 A. The Veṅgi Alphabet — 14 parts
c5.s1.t2 B. Western Cālukya — 20 parts
c5.s1.t3 C. Eastern Cālukya — 25 parts
c5.s1.t4 D. Transitional — 25 parts
c5.s1.t5 E. The Old and Modern Telugu-Canarese Alphabets — 34 parts

§ 1. Telugu-Canarese Alphabets

c5.s1.1

Of the South-Indian alphabets, the most important from every point of view are the Telugu and Canarese. The parts of the Peninsula where these characters have been developed have been of the greatest importance in the political and literary history of the South, and chronologically they are the first.

c5.s1.2

The earliest documents existing belong to the Telugu country comprising the deltas of the Kṛishṇā and Godāvari, where also, at Amarāvati, the most important Buddhist remains in the South, have been found. The origin of this kingdom does not probably go back beyond the second century A. D., for it is not mentioned in Ptolemy or by the Periplus of the Red Sea by the name found in the inscriptions—Veṅgideça—or even by the later name Āndhra used by Hiouen-Thsang (7th century¹). The names and dates of the kings are quite uncertain, for only two grants of this dynasty appear to be in existence, and one of these is almost entirely illegible²). The dates they bear, are also, like those of all early inscriptions, merely the year of the king's reign, and this is not referred to any era. This dynasty was supplanted in the beginning of the seventh century A. D. by a branch of the Cālukyas already established at Kalyāṇa about the beginning of the fifth century A. D. and which is the first *historical* dynasty of the Deccan. It appears that the Pallava kings of Conjeveram belonged to the Veṅgi family; probably Conjeveram was a dependent province which became their chief place after the conquest of Veṅgi by the Cālukyas.

c5.s1.3

¹ There is not the least mention of any Telugu kingdoms in the Açoka Inscriptions. Probably that part of India was not then civilized at all, but inhabited by wild hill-tribes.

c5.s1.4

² Mr. Fleet, I regret to say, also gives up all hope of reading the second inscription.

c5.s1.5

Taking Fa-Hian's account of the Deccan (400 A. D.) it is excessively improbable that the history of that part will ever be traced back to an earlier date.

A. The Veṅgi Alphabet

c5.s1.t1.1

Compared with the Cave character the Veṅgi alphabet presents little development, and I think that this fact justifies the date I have assigned to the specimen given in Plate xxiv.¹

c5.s1.t1.2

¹ That the dynasty, to which the inscription given in Plate xxiv. belongs, preceded the Cālukyas was first pointed out by Sir W. Elliot in the *Madras Journal* (Vol. xi. pp. 302-6). The capital (Veṅgi) appears to have entirely vanished; it is said to have been the place now called Pedda Veṅgi or Vegi in the Kṛishṇa District, but there are several places of the same name in the neighbourhood. As in the Telugu Mahābhārata which belongs to the twelfth century A. D. Rajahmundry is called the Nayakaratnam of Veṅgideça, the old capital must have been deserted long before that time. Hiouen-Thsang (iii., pp. 105-110) calls the small kingdom that he visited '’Ān-ta-lo' (Andhra) and the capital—'’Ping-k’i-lo'. It appears to me that this is intended for Veṅgi; the 'lo' being merely the locative suffix '-lō' of the Telugu nouns, naturally mistaken by the worthy Chinese pilgrim monk for a part of the word. So the Portuguese called Çālayam—Chaliatta, using the inflected form of the name. Julien's suggestion 'Viṅkhila' only fails in there not being the slightest trace of such a place. The *-i* in Veṅgi is uncertain; it occurs both short and long in the Sanskrit inscriptions. In Canarese it is certainly short; in Tamiḻ the name appears as Veṅgai (great inscription of Tanjore), and this indicates a short vowel, Veṅgi seems to be a Sanskritized form of Veṅgi. cfr. Kāñcī for Kañći etc. Veṅgī seems impossible as a Dravidian word. Veṅgŏ also occurs. 'Āndhra' is properly the name of the country between the two rivers, and only became synonymous with 'Telugu' owing to that kingdom being the native place of the writers in and on Telugu in the twelfth and following centuries.

c5.s1.t1.3

In **ā** the curl at the foot which distinguishes this letter from the short **a** is extended, and this is a peculiarity which appears only in this character.

c5.s1.t1.4

The perpendicular strokes on the left sides of **j** and **b** are here curved, as are the top and bottom lines of **ṇ**.

c5.s1.t1.5

**v** in the second inscription to which I have referred, is represented by a triangular form disproportionately large compared with the other letters, and thus very near the Cave form.

c5.s1.t1.6

The suffixed forms of the vowels differ somewhat from those in the Cave character.

c5.s1.t1.7

**i** which is in the last represented by a semicircle open to the left is here open towards the top of the consonant which it follows or is united to it; **ī** which was originally represented by a semicircle open above and attached to the consonant, or by a semicircle open to the right is here represented by a curl which marks the long vowel very clearly.

c5.s1.t1.8

**ū** which was originally marked by a semicircle open at the bottom, and under the consonant it follows, is here represented by a highly characteristic curved form which does not appear in any other alphabet.

c5.s1.t1.9

In the compound consonants the second and third letters still retain their complete original form. The superscript **r** still preserves the straight line of the original **r** of the Açoka inscriptions.

c5.s1.t1.10

**r** is here represented by a form that occasionally occurs in the inscriptions of the W. Cālukyas up to the end of the sixth century, viz., with a short loop turned to the left. In the E. Cālukya deeds the loop is generally turned to the *right*, if it is not complete.

c5.s1.t1.11

Final **m** is represented by a small **m** less than the other letters, which is also peculiar to the Veṅgi character. The existence of a distinct sign for upadhmānīya (≍) etc. is especially worthy of notice, as proving that the Sanskrit alphabet was in the fourth century A. D. already adapted to suit the niceties of the grammarians. This character has also a sign ⧖ for the vajrākṛiti (*i.e.* *h* before **k** and **kh**) as has been indicated by Mr. Fleet.

c5.s1.t1.12

As in the Cave inscriptions, so also here, we find that a small cross-stroke or thickening of the top end of the line is made in all cases where the letters begin with a perpendicular stroke downwards. The character of the Veṅgi inscriptions is angular like that of the Caves, whereas the Açoka letters are rounder.

c5.s1.t1.13

The cross stroke has, no doubt, arisen from the necessity of marking clearly the end of the line, especially in inscriptions on stone, but, developed in the course of time, it has become the angular mark ˇ above some Telugu and Canarese consonants which has been strangely imagined to be the short vowel **a**. This error was started by the first Telugu Grammar by A. D. Campbell[^1), but has been constantly repeated down to the present time without any reason at all[^2).

c5.s1.t1.14

On the inscriptions in this character in Java, and on the early Kawi (Old Javanese) character, see Appendix B.

B. Western Cālukya

c5.s1.t2.1

The earliest specimen of the Western Cālukya character was, hitherto, supposed to be a grant by Pulakeçi, dated ç. 411 (or A. D. 489), and of which an abstract is given in the Journal of the R. Asiatic Society[^4). This has, however, been found by Mr. Fleet

c5.s1.t2.2

[^1): Second edition (1820) p. 3. The error is probably of native origin as this mark is called in Telugu—talakaṭṭu. [^2): See the last published Telugu Grammar by the Rev. A. Arden (1873) p. 7 where it is called a ‘*secondary*’ form of a. [^3): The origin of this name (which is also written Calukya, Callukya, Caulukya, Calikya and Calkya) is obscure. A grant of 1086 (E. of p. 21 *n.*) mentions a ‘Calukyagiri’ where Gaurī was worshipped (see next page). But the event to which this passage refers, must—if it ever occurred—be put about the 4th century A. D. or some seven hundred years before the date of the inscription which records it! I am unable to find any other traces of this hill; it may be one of the numerous mountains in Central India where barbarous rites (as described by Colebrooke and, in later times, by Forsyth) still prevail. The legend evidently belongs to a comparatively recent period in the history of the family, and when it had become of great consideration in Central India. The family appears to have been known even in Java; in a document of 841 A. D. an interesting list of names of countries occurs: “kliṅg, . . . gola, tjwalikā, malyalā, karṇake” . . . kliṅg=kaliṅga; gola is correctly copied from the original plate, but it must be for Cola; tjvalikā (cvalikā)=cālukya; malyalā=malayāḷa; karṇake may perhaps be read karṇāṭaka; anyhow, it is plain what is intended [Cohen-Stuart, “Kawi Oorkonden,” p. 8 (tekst) and 5 *b.*, 4 of the facsimile]. A purely mythical explanation of the name has been found (by Mr. Fleet) in an inscription; this derives the name from *culuka*=a water-pot! *I. A.* vi., p. 74. *Cfr.* also Vikramāṅkadevacarita I., 46, for the same derivation. [^4): Vol. v. pp. 343 flg. Mr. Fleet has sent me a photograph of a leaf. The Tamiḻ inscriptions of the 11th century (Tanjore and Seven Pagodas) call this kingdom “iraṭṭa-pāḍi (*i.e.* Reḍḍi-kingdom) seven and a half lakhs.”

c5.s1.t2.3

to be a recent forgery; the character of the writing, I find, makes this perfectly evident. The earliest I can use is, however, a grant on copper plate, of Maṅgaḷa of about 578 A. D. which is a little before the most flourishing period of the Cālukyas in the beginning of the seventh century A. D.¹ It is not the earliest known authentic grant of

c5.s1.t2.4

¹ The defeat of Harshavarddhana, the king of Kanoj, by a Cālukya which is satisfactorily established by Cunningham ("Reports" i., pp. 280-282), shows the rapid growth in power of the Cālukyas of Kalyāṇapura. This defeat was not, however, by Vikramāditya (as Genl. Cunningham states) but by Satyāçraya, his father, as is proved by several inscriptions. One (in possession of a Jain ācārya at Hyderabad) has: Çri-Pulakeçimahārājasya prapautraḥ....Çri-Kīrtivarmapṛi(thi)vīvallabhamahārājasya pautraḥ samarasamsaktasakalottarāpatheçvaraÇri-Harshavarddhanaparāja(yopala)bdhaparameçvarāparanāmadheyasya Satyāçrayaçripṛithivīvallabha..sya priyatanayaḥ etc. Another (photographed in the Mysore collection) has *nearly* the same phrase: ....Çri-Harshavarddhanaparājayopalabdhaparameçvarāparanāmadheyaḥ Satyāçrayaçripṛithivīvallabhamahādhirājaparameçvaras etc. This defeat must be put near the end of the 6th or beginning of the 7th century. The genealogy of the dynasty of these kings was first given by Sir W. Elliot in the *London Asiatic Society's Journal*; and his paper was afterwards reprinted with corrections in the *Madras Journal* (vol. vii., pp. 193-211). With a few additional corrections required by inscriptions since discovered, and some of which were pointed out by Lassen (I., A. K. iv.), also omitting the mythical beginnings of the dynasty, the table is as follows:—
1) A grant of 1086-7 A. D. (referred to as E on p. 21, n.) traces the Cālukya family to the Somavamça through a number of mythological personages and kings of Ayodhyā to a Vijayāditya who: vijigīshayā Daxiṇāpatham gatvā Trilocanapallavam adhixipya daivadurīhaya lokāmttaram agamat. Tasmin samkule purohitena vṛiddhāmātyaiç ca sārddham anta(r)vva(tī pa)tnī tasya Mahājavamuṛilemunāmāgrahāram upagamya tadvāstavyena Vishṇubhaṭṭasomayājinā duhitṛinirvviçesham abhiraxitā satī Vishṇuvarddhanan namdanam asūta; sā ca tasya kumārakasya k(ā)lakramopetāni karmāni k(ā)rayitvā tam avarddhayat; sa ca mātrā viditavṛittānto niggitya (? nirgatya) Calukyagirau Namdābhagavatīm Gaurīm ārādhya Kumāranārāyaṇamātṛigaṇāmç ca samttarpya çvet(a)tapatraikaçamkhapañcamahāçabdādīni kulakramāgatāni nix(i)ptāni ******** samādāya Kadambagaṅgādibhūmipān nirjitya SetuNarmadāmadhyam daxiṇāpathan pālayām āsa. Tasya Siddijayādityo Vishṇuvarddhanabhūpatiḥ Pallavānvayajātāyā Mahādevyāç ca nandanaḥ. | tatsutaḥ Pulakeçivallabhaḥ; tatputraḥ Kīrttivarmmā. Tasya tanayaḥ .... Satyāçrayavallabhendrasya bhrātā Kubjavishṇuvarddhano dvādaça varshāṇi Veṅgīdeçam apālayat etc.
2) This date is due to Profr. Eggeling.
3) See Mr. Fleet's remarks *Indian Antiquary*, vi. p. 73.

c5.s1.t2.6

o It is necessary to remark that this genealogy and dates can only be regarded as provisional, not as definite.

c5.s1.t2.7

this dynasty; for there is an inscription of Maṅgalīçvara *d.* ç. 500 in the Bādāmi cave, but this is not accessible to me. The Aihoḷe inscription is *d.* ç. 556 as finally read by Mr. Fleet¹).

c5.s1.t2.8

¹ ^1) I owe a facsimile of Maṅgala's inscription to Mr. Fleet; he has edited it in the *Indian Antiquary*, vol. vii., p. 161.

c5.s1.t2.9

All these inscriptions are in an upright square hand with the letters very well formed; the forgeries (of which there are several) do not imitate this style, but resemble the inscriptions of their real dates.

c5.s1.t2.10

So far the flourishing older dynasty of the Cālukyas, which after Vikramāditya II. appears to have been for a time almost overthrown by feudatōries such as the Rāshtrakūṭa, Kālabhurya, and Yādava chiefs, and the history of this kingdom is, thus, very obscure for the eighth and ninth centuries. With Tailapa the restorer of the Cālukya power in the later dynasty, all once more becomes tolerably certain, especially as regards the dates of the reigns. A very poetical account of the first sovereigns of this line is given in Bilhaṇa's Vikramāṅkadevacarita; it is often contradicted in details by the Coḷa inscriptions.

c5.s1.t2.12

Mr. Fleet has pointed out (*Indian Antiquary*, vii., p. 20) that the form 'Cālukya' was used by this later dynasty.

c5.s1.t2.13

A feature common to all the later inscriptions of the Western Cālukyas but which does not occur in any others, is a marked slope of the letters to the right. The Eastern Cālukya character is, on the other hand, remarkably square and upright; this distinction is quite sufficient, after 650 A. D., to show the origin of an inscription.

c5.s1.t2.14

Somewhat later, about 700 A. D., is the beginning of the change in writing subscript vowels which afterwards formed the chief difference between the Telugu-Canarese alphabets on the one hand, and the Grantha on the other—a tendency to bring the marks for **ā**, **ē** and **ō** from the side of the consonant to which they are attached to the top, and again to bring the mark for a subscript form from underneath the consonant to its right side. The character in Pl. iv. (690 A. D.) uses almost universally the older form (cfr. **ku**, **tu**, etc. in Pl. iv.).

c5.s1.t2.15

Only the cursive forms of **a** and **ā** occur in the later inscriptions of the Western Cālukyas after about 650 A. D., so far as they are known to me, and this again distinguishes them from those of the Eastern dynasty which preserve most generally the older forms of these letters up to the middle of the tenth century, though we find both forms co-existing in inscriptions of the eighth and ninth centuries.

c5.s1.t2.16

The W. Cālukya method of writing **r** above a following consonant is primitive, like the Veṅgi, and differs from the Cera form (see Pl. ii.); it did not last long, for about 700 A. D. the **r** is made clearer by a slope to the right, and this (between 1200 and 1300 A. D.) developed into the modern Telugu form, and then, at last, was written separate to the right of the consonant it precedes in utterance.

c5.s1.t2.17

**Ch** appears at the time of the oldest South-Indian inscriptions to have had the form of ⋈ (cfr. pl. iv. **ččh**); in the modern alphabets this is quite lost, and this letter has the ordinary form of **č** with the addition of a small stroke underneath, such as marks the aspirate in ఛ, ಛ etc.

c5.s1.t2.18

Interesting as the inscriptions of the Western Cālukyas are historically, owing to the synchronisms with events in the history of Northern India that they exhibit, they are but of little importance in the literary history of the South of India; for it is certain that the kings of Kalyāṇapura always favored the culture of the north¹.

c5.s1.t2.19

¹ Vidyāpati of Someçvaradeva (Āhavamalla) I. (1040-1069) was Bilhaṇa, a native of Cashmere.

c5.s1.t2.20

With the temporary fall of this dynasty the Western Cālukya alphabet appears to have gone entirely out of use².

c5.s1.t2.21

² Several of the inscriptions of this earlier dynasty have already been published in the *Journal of the R. Asiatic Society of London*, and in the *Bombay Journal*. (See for the last: Vol. ii., 1-12, pp. 262-3; Vol. iii., pp. 203-213. The first of the grants described belongs to the reign of Vijayāditya, and is dated, ç. 627 = A. D. 705. The second is dated in the tenth year of Vijayāditya; the third appears to belong to a feudatory). Facsimiles of some from the sixth to about the fourteenth century are given in the "Collection of Photographic Copies of Inscriptions in Dharwar and Mysore" published by the Committee of the Architectural Antiquities of Western India. See also I. A. i., p. 80 (Tribhuvanamalla 1083 A. D.); p. 141 (? Someçvaradeva ii., 875 A. D.); iii., p. 305 (Maṅgalīça of 578 A. D.); iv., p. 278 (Jayasimha ii., A. D. 1028); v., p. 67 (Pulakeçi ii., 585 A. D.); p. 342 (in the reign of Vikramāditya, 1093 A. D.); vi., pp. 72-8 (Satyāçraya, 613 A. D. and a doubtful one of Vikramāditya i., without date); pp. 85-94. (Vinayāditya, 690 A. D.; do: 692 A. D.; do: 695 A. D.); p. 137-142 (Vikramādityatribhuvanamalla, A. D. 1096; Jagadekamalla, ? 1139 A. D.)—nearly all by Mr. Fleet.

c5.s1.t2.22

The earlier Kadamba inscriptions are in a character very near what is here described.

C. Eastern Cālukya

c5.s1.t3.1

In the early history of the Dravidian part of India, this dynasty is of the greatest importance, but as yet no account of it has been published. It succeeded the Vengi kings early in the seventh century, not long after the famous defeat of Harshavardhana by Satyāçraya of Kalyāṇapura, and was founded by his younger brother¹). In the

c5.s1.t3.2

¹ The dynasty is given as follows in a number of inscriptions which I have been able to consult; nearly all of which (an unparalleled circumstance in India) give the number of years that the several kings reigned. A. (from Masulipatam) *d.* 5th year of Vishṇuvardhana II. B. (in the Nellore Sub-Collector's Office on five plates) contains a grant by Yuddhamalla (about 950). C. on five plates (? the Godāvari Collector's Office) *d.* ç. 867 = 945 A. D., being in the reign of Ammarāja. D. a grant of Kulottuṅga-Coḷa-Deva, *d.* 1085 A. D. E. = a grant by Kulottuṅga in the 23rd year of his reign (= 1087) from Pittapur. F. a grant by Kulottuṅga (Vīracoḷadeva) son of Vikramacoḷadeva, *d.* ç. 1056 = 1134 A. D. The number of years each king reigned follows in () his name. Those names which are not of actual sovereigns of Veṅgi are in spaced type.

c5.s1.t3.4

1) B. D. E. F. make Jayasimha reign 33 years.
2) A. "*Çrīkīrttivarmaṇaḥ* pranaptā.....*Çrīvishṇuvarddhanamahārājasya* napt(ā).....*Çrījayasiṃhavallabhamahārājasya* priyabhrātur anekayuddhālaṅkṛtaçarīrasye 'ndrabhaṭṭārakasya priyatanayaḥ çrīmān Vishṇuvarddhanamahārājaḥ" etc. D. F. make Indrabhaṭṭāraka reign for seven days.
3) D. has: "*tasya* (*i. e.* Kokkileḥ) jyeshṭo bhrātā tam ucchāṭya saptatriṃçat.

c5.s1.t3.5

inscriptions of this dynasty their territory is often called Veṅgi, and it forms the second and last Veṅgi dynasty.

c5.s1.t3.7

Vimalāditya was succeeded by Rājarāja Coḷa owing (it is stated) to an intermarriage of the Coḷas and Kaliṅga Cālukyas, which really occurred. His son Kulottuṅga succeeded him in 1064 A. D. (*Madras Journal*, xiii., Pt. 2, p. 40), and as Rājarāja reigned 41 years (D. E. and F.) this makes the date of the end of Vimalāditya's reign to be 1023 A. D. Both D. and E. *explicitly* term Rājarāja son of Vimalāditya. For the passage in E. see note 4 below. The Tamil inscriptions (*e. g.* the inscription at Tanjore) state that Veṅgināḍu was conquered; Indian history shows that such marriages as the one mentioned above, were *always* the result of a conquest.

c5.s1.t3.8

¹ D. F. make his reign last 36 years. E. makes it 26.

c5.s1.t3.9

² C. F. have: dvyardhavarshāṇi; B.—Ashṭādaça māsā(n); D.—dvyardhavarsham. E. has: adhyardhavarsham; is this a mistake in copying an older document?

c5.s1.t3.10

³ E. has eleven months.

c5.s1.t3.11

⁴ D. E.—Tāḍapa. E. omits the length of this reign.

c5.s1.t3.12

⁵ D. tam ucchātya diçād Ammaraj(ā)nujo Rājabhīmaḥ dvādaça varshāṇi.

c5.s1.t3.13

⁶ In D. only the years are clearly legible. E. has after Ammarāja II. ta(j)jyeshṭo Dānā * * as trimçat; tatputra(*h*) Çaktivarmā dvādaça; tadanujaVimalādityas sapta; tatputro Rājarājadeva ekacatvāriṁçat; tatputra(*h*) çrīKulottuṅgacoḍadeva ekonapañcāçat etc. E. has: tatsūnur Ammarājaḥ pañcaviṁçatim; tasya dvaimāturo Dānanṛipaḥ trīṇi; tatas saptaviṁçativarshāṇi devadurdhayā Veṅgimahīr anāyikā 'bhavat; tato Danānnavasutaḥ Çaktivarmanṛipo dvāda(ça) varshāṇi bhūm apālayat; tatas tadanujas sapta vatsarān bhūtavatsalaḥ Vimalāditya (*sic*) bhūpālayāmāsa medinī(m) | tattanayānayaçālī jayalaxmīdhama Rājanarendraḥ catvāriṁçatam abdān ekam ca punar mahīm apālayad akhilām etc. The causes of the time of anarchy at the end of the 10th and beginning of the 11th century are not known, but may be fairly attributed to the Coḷa invasions. At the beginning of the 11th century (see Tennent's "Ceylon" i., p. 402) the Coḷas had conquered that island, and not long after they must have effected the successful invasion of Bengal which is recorded in the great inscription at Tanjore. C. carries the genealogy down to Ammarāja, and it is dated 945 A. D. in his reign. The grants D. and E. would make the beginning of his reign four and six years respectively after this date. The discrepancy is not, however, sufficient to throw doubts on the list given above, and is probably owing to the uncertainty of the Çaka era. It is obvious that the number of entire years of most reigns only being given, the list cannot be *absolutely* correct. The total of the reigns of sovereigns of this dynasty amounts to about 403 years, which brings the first year of Kubja Vishṇuvardhana to about 620 A. D., and as his elder brother Satyāçraya reigned in Kalyāṇapura about 600 A. D., this date is by no means improbable. It is nevertheless impossible to suppose that the Kaliṅga Cālukyas were established in the old Veṅgi kingdom for some years after that date. Thus the grant printed in pl. xxvii. was found far north (in Vizagapatam), and it seems probable that the Cālukyas first seized the northern part of the Telugu sea-coast, and then conquered the south. The eclipse mentioned in the earliest known grant of this dynasty (see App. C. and plate xxvii.) shows that Vishṇuvardhana was reigning in 622 A. D.

c5.s1.t3.14

This dynasty is of the greatest importance so far as S. India is concerned, for it can be traced with some certainty, and affords clues to the dates of important events.

c5.s1.t3.15

The earliest inscription I have seen, is a grant by the first sovereign Vishṇuvardhana I.; it is on copper-plates, and was found in the Vijayanagaram Zamindary in 1867 (Pl. xxvii.). Except in regularity and neatness, the character of the writing of this document differs very little from that already described as the Veṅgi character, and does not exhibit any cursive forms; these first appear in the latter part of the seventh century.

c5.s1.t3.16

The chief distinctions between the characters used for the Western and Eastern Kaliṅga¹ inscriptions have already been given. As the two countries were under branches of the same royal family about the same periods, it is convenient to call the respective characters after the two dynasties of the Cālukyas; but it must be recollected that there is no real connection between them palæographically, except so far as their common origin through the ‘Cave character’ is in question.

c5.s1.t3.17

¹ Kaliṅga, or rather Tri-kaliṅga is a very old name for the greater part of the Telugu Coast on the Bay of Bengal. The latest mention I know, is in the grant of Yuddhamalla (already referred to as B.), which says of this king (about 950 A. D.) “*Veṅgībhuvaḥ* patir abhū(t) Trikaliṅgakoṭṭeḥ” (4 line 3). Hiouen-Thsang also mentions Kaliṅga (7th cent.). Pliny (vi., 67 of the edition published by Teubner) says: “Insula in Gange est magnæ amplitudinis gentem continens unam, nomine Modogalingam.” Dr. Caldwell (Comp. Gr. pp. 64-5) took this to be for the old Telugu, ‘Modoga and liṅga’ and to mean “three-liṅgas”, and, thus, accepted the native etymology of ‘Telugu’. There can be no doubt that it is merely Mūḍu-Kaliṅga or Three Kaliṅgas, and has nothing to do with liṅga. The native etymology of ‘Telugu’ first occurs, I believe, in the Kārikā of Ātharvaṇācārya who copied and *quotes* Hemacandra, and therefore could not have lived before the thirteenth century. In his second edition, however, Dr. Caldwell gives up this explanation (p. 32). ‘Telugu’ is evidently from a common Dravidian root √teḷ or √tel[^1)] which means ‘to be clear or bright’, and the Tri-liṅga theory is certainly not supported (as Dr. Caldwell appears to think) by Ptolemy’s Triglypton or Trilingon (vii., 2, 23), which is most probably a copyist’s error for Trikaliṅgon. At all events a derivative of ‘glypho’ could never mean liṅga. Cunningham (“Ancient Geography of India,” p. 519) recognizes three Kaliṅgas, and rightly doubts the name having anything to do with liṅga.

c5.s1.t3.18

The decided tendency of the Eastern Cālukya character to preserve archaic forms, clearly distinguishes it from the character used under the Western dynasty. This last seems to have been affected by the North-Indian early Nāgarī, as it almost copies the horizontal stroke at the top of letters used in the latter. It also uses cursive forms to a large extent.

c5.s1.t3.19

1) Tamil: ḷ=Tel: l; Dr. Caldwell “Comp. Gr.” p. 194. A. D. Campbell suggested this derivation: Tel. Gr. p. vi.

c5.s1.t3.20

The Plates iii., iv. and v., if compared, will show how correct is the account by Hiouen-Thsang (about 640 A. D.) of the writing used in his time in the Deccan and on the sea-coast. He says¹: "La langue et la prononciation diffèrent beaucoup de celles de l'Inde centrale; mais la forme des caractères est en grande partie la même."

c5.s1.t3.21

¹ "Voyages des Pèlerins Bouddhistes," iii., p. 105. In 1031 A. D. Albīrūnī mentioned this character which he called 'andri'. [^2): *Cfr.* Whitney's remarks: "Atharvaveda Prātiçākhya," p. 140 on the 'characteristic tendency' of the Hindu Grammarians 'to arbitrary and artificial theorizing'.

c5.s1.t3.22

All unquestionable grants by kings of both the Cālukya dynasties that I have met with are in Sanskrit. The later they are, the greater is the neglect of the minute rules for orthography laid down by the Sanskrit grammarians, especially as regards the use of the bindu. I shall give a summary of the results that I have ascertained, further on, in describing the modern alphabets used in the Telugu and Canarese countries. (p. 29.) It is, perhaps, to be regretted that editors of Sanskrit texts, in these days, are in the habit of restoring exactly the orthography of the earliest grammarians; if we rightly do so in the case of Vedic texts where the MSS. justify this course, as well as the fact that we have here to deal with relics of a time when Sanskrit was a living language, the case of most texts is quite different; they were written long after Sanskrit had become a dead language, and represent no real pronunciation; to complicate the orthography of such texts is unmeaning pedantry, and can lead to no good result[^2). It is certain that their authors (often profound grammarians) never followed the primitive orthography—why should foreign editors in the present day? The oldest inscriptions (*e. g.* Girnar) use the bindu for **m** in the cases where it does not represent **m** before **h**, and this is the only exception.

c5.s1.t3.23

I may, however, here properly call attention to the remarkable practical results of the minute studies of the early Indian grammarians as regards the analysis of Sanskrit phonetics. When the Brahmans from the North of India introduced literary culture to the Dravidians (except probably the Tamil race) and to the Polynesians, they came armed with the results of these studies which might seem of not the least practical value, but it would be difficult to find a clearer instance of the ultimate practical utility of the most recondite scientific research; for they were thus able, in what was certainly a very short space of time, to furnish a number of foreign and uncultivated languages with admirably exact *phonetic* systems of writing, and to this alone is to be attributed the rapid growth of indigenous literatures among those peoples. The merit of this work will be more apparent, if one compares it with early attempts of Europeans in the same way.

c5.s1.t3.24

The monstrosities of the English are, unhappily, too well known, but there is a singular example in a book written by an Italian of learning and culture, who was one of the first in modern times to travel in Western India—L. Varthema. He knew Malayāḷam well, and gives several specimens of it; the following sentence will do for my purpose: “Matile matile: gnan ciatu poi”. This he explains by “non piu non piu chio son morto,”¹ and it represents the Mal. മതി മതി ഞാൻ ചത്തു പോയി maδi maδi ńān čattu pōyi. A knowledge of the Italian pronunciation would perhaps enable any one to utter this in a way intelligible to the natives of Malabar, but the value of the native letters being once known, no difficulty could occur. So if we take Father Estevâo’s transcriptions of Koṅkaṇī², and the Dutch way of writing Malay³ both of about 1600 A. D., their inferiority to the Indian adaptations is evident, though all are really very well done. Nor were the men who adapted the Sanskrit alphabet to the Dravidian languages and the Javanese, mere mechanical workmen, they could add the necessary new signs to some extent as will appear in the course of this work. They often, however, used Sanskrit letters to express others which are, phonetically, different, though similar. (App. A.)

c5.s1.t3.25

¹ Fol. 75 *a.* ed. of Venice, 1517. I have correctly divided the words printed: gnancia tu poi.

c5.s1.t3.26

² See my “Specimens of S. Indian Dialects” No. 1. pref. pp. 11-13 (2nd ed.).

c5.s1.t3.27

³ See the Dialogue in De Bry’s smaller voyages (E. Indies) Pt. ix. pp. 33, ffg. (Latin).

D. Transitional

c5.s1.t4.1

What I have termed the transitional period, or from 1000-1300 A. D., marks the rise and most flourishing period of the North Dravidian literatures. During the whole of this time the older kingdoms decayed rapidly, feudatories became more or less independent, and changes in the limits of territory subject to the different sovereigns were perpetual. The encouragement of literature was, however, general, and this period is also marked by the rise of several religious sects. The result, palæographically, was that by 1300 A. D. the old Tĕlugu-Canarese alphabet which was in use from the coast of Canara to Rajahmundry, presented scarcely any varieties or differences of form of the letters sufficient to justify a distinction being made⁴. From 1300 A. D. up to the present time, however, a marked divergence has arisen between the alphabets used by the Telugus of the coast and the Canarese people; and this divergence has been much increased since the introduction of printing in the course of the present century.

c5.s1.t4.2

⁴ Al-Bīrūnī, however, (Reinaud, “Memoire”, p. 298) distinguishes in 1030 between the Karnata and Andri characters.

c5.s1.t4.3

The feudatories which overthrew the Western Cālukya kingdom appear to have been partial to the N. Indian culture, and used the Nāgarī character for their grants¹. The Coḷas (who succeeded the Eastern Cālukyas) preserved the indigenous character and used Sanskrit for the northern part of their territories, but soon gave these up for Tamiḻ. Thus, at the time of the Muhammedan invasions and settlements in the peninsula about the beginning of the fourteenth century, the use of the South-Indian alphabets was confined to the extreme south of the peninsula, and did not extend much beyond the present northern limits of the Madras Presidency. That the Telugu and Canarese alphabets and literatures did not become entirely obsolete, is owing to the considerable power of the Vijayanagara² dynasty in the 14th, 15th and early part of the 16th centuries, and to the steady patronage of South-Indian Hinduism by the kings of this dynasty during that period of time³. It is owing to this influence that many inscriptions from about 1500 to 1650 A. D. in the North-Tamiḻ country and even still further South are in the Telugu character. This is especially noticeable in the old Tŏṇḍaināḍu (or neighbourhood of Madras), and it is to the same influence that must be attributed the numerous settlements of Telugu Brahmans over greater part of the Tamiḻ country, and especially in Tanjore.

c5.s1.t4.4

¹ I shall for this reason notice them when describing the varieties of the Nāgarī character used in the South of India.

c5.s1.t4.5

² Or Vidyānagara. The last (Mr. Kittel tells me) occurs in the C. Basava-Purāṇa, ch. lxiii., 2-3 (1369 A. D.). Both forms seem equally authentic, but the first seems to be the earliest, and occurs in a grant of 1399. Cfr. Colebrooke's *Essays*, ii. p. 263. Couto's explanation (Dec. vi. f. 92 *b*. of orig. ed.) shows that about 1600 ‘Vijayanagara’ was the accepted form. He says that the name signifies: “Cidade de vitoria.”

c5.s1.t4.6

³ The Telugu poet Bhaṭṭamūrti was encouraged by Narasarāya, and Allasanni Peddaṇṇa by Kṛishṇarāya. (“*Madras Journal*,” v. pp. 363, 4.)

c5.s1.t4.7

The transitional type of the Telugu-Canarese alphabet differs from the Kaliṅga-Cālukya by the admission of a number of new forms which eventually became permanent; they are used, however, concurrently with the older forms except in a few instances.

c5.s1.t4.8

The exclusive new forms of letters are: 1) č, in this the top is opened out; 2) dh, in which the old square form is now provided with a ˇ at the top, 3) and bh. This last was evidently written in the alphabet of 945 A. D. by two strokes, the second being made from the first, and prolonged down in a curved form; in the transitional alphabet which began in the next century these two strokes are separated. 4) ç has a more cursive form than in the alphabet of the previous century.

c5.s1.t4.9

As in the alphabet of 945 A. D. there is little distinction between the long and short i superscript. In the older alphabets the long ī is marked by a curl in the left end of the circle which marks this vowel, *e.g.* ∩ (i) and 𑌲 (ī), but from the tenth century this distinction is almost lost.

c5.s1.t4.10

In the eleventh century the modern form of the subscript **u** begins to appear, and is used far oftener than the old form written underneath the preceding consonant; but the reverse is the case with the long **ū** which rather preserves the old form. In the next century the modern form of **ū** (to the right of the preceding consonant) prevails nearly universally, but the old form of the short **u** is by no means entirely disused. The secondary forms of **e** and **ui** and **ai** are very nearly the same as in the alphabet of 945 A. D., *i.e.* written at the top of the preceding consonant, whereas in the earlier forms they are on the left side. **O** and **au** are also very little changed in form.

c5.s1.t4.11

It is necessary also to notice the changes in the way of distinguishing **ph** from **p**. In the earliest form (Pl. i.) this is done by the upper end of the stroke on the right side being curled round to the left; in the later alphabet of the tenth century there is a loop on the middle of the inner side of this stroke. In the alphabet of the next century this loop has become a slanting stroke across the upright stroke, and finally about a century later this is underneath the middle of the letter.

c5.s1.t4.12

It is necessary here to notice the use of a sign for **ç**, and also the signs added to the original Sanskrit alphabet to express the Dravidian letters **ḻ**, **ṟ**, etc.

c5.s1.t4.13

A sign for **ç** does not occur in the S. Açoka and Cave inscriptions, but only **sh** and **s**. In the Sah, Veṅgi, and later characters we have a distinct sign. Now in some of the earliest Ceylon inscriptions M. Rhys Davids detected two sibilants: one the ordinary **s**, the other a 𑀲, and the two are indiscriminately used for **s**1), and this has been assumed to be quite peculiar to Ceylon2). The importance of the discovery in respect of phonetics is very great, but I think it will eventually be admitted that this letter is merely the Sanskrit sign for **ç**. If the sign for this letter in the earlier S. Indian inscriptions be referred to, it will be seen to be almost the same. It is certain, at all events, that this sign was early in general use, for we find it in the earliest Javanese (Kawi) inscriptions3); but the Ceylon form is evidently the earliest and, I think, will help to show the real origin of the sign for the Sanskrit **ç**. In the later Phœnician and in the Aramaic character4) of from the seventh to the fourth century B. C., the letter ‘shin’ has the form ש or 𐡔 or 𐤔, and this is obviously the original of the **sh** of the earlier southern inscriptions, which is ᑌ (*e.g.*) in the Mathura (or E. Cave) character¹. In the northern character this is inverted to express **sh**, and it appears to me that this same letter, but inverted, is used in the S. alphabets to express **ç**. Such divergences between the N. and S. Açoka characters exist, and prove an independent development. The N. and S. (Açoka) signs for **s** are, however, clearly derived from the later form of ‘Samech’.

c5.s1.t4.14

1) *Indian Antiquary*, i. p. 140.

c5.s1.t4.15

2) See the late Dr. P. Goldschmidt's *Report* (of 1876) p. 4 in which he says: "A graphical peculiarity of the most ancient inscriptions is the use of two *s* (one the common *s* of Açoka's inscriptions, the other one resembling a Greek Digamma, a form unkown in India), which it would be difficult to account for without the supposition that the pronunciation of *s* in Ceylon must have struck the Hindu introducers of the art of writing as somewhat different from their own." See also *Academy*, xi. p. 139.

c5.s1.t4.16

3) See especially ‘Çrīman’ in the Tjaroenten inscription (? middle of the 5th century A. D.). It occurs in several inscriptions of the 8th century A. D., *e.g.* in the Çaivite one from Brambānan (Cohen-Stuart, No. xxiii., lines 1-4 etc. repeatedly), also in the Sumatra inscriptions.

c5.s1.t4.17

4) See pl. ix. of vol. i., pt. 1 of Lenormant's "*Essai*".

c5.s1.t4.18

¹ *J. R. As. Soc.* New Series, vol. V. pl. i. Both ᖵ and ᑌ occur in the Kangra inscription. See Prinsep's “Essays” ed. Thomas, i. p. 159. (pl. ix.)

c5.s1.t4.19

The sign for **ḷ** in the Telugu and Canarese alphabets is a development of the sign for **ḍ**; it is very nearly the same up to about 1200 A. D.; a little after it appears with the tail turned to the left, and thus has become a distinct sign².

c5.s1.t4.20

² In the inscriptions from the Telugu country we find Coḷa always written Coḍa, but the Telugu ḍ = ḷ is probably a late degeneration; ḷ seems to be a primitive Dravidian sound.

c5.s1.t4.21

The Telugu and Canarese languages also required an additional sign for **ṛ**, this is represented by ఱ or ಱ. The origin of this sign is not clearly ascertained; it occurs in early inscriptions in Canarese but later in Telugu.

c5.s1.t4.22

There is also the form ೞ which represents the S. Dravidian **ḻ** (ழ) and which does not occur in Telugu. The origin of this is also unknown; it is used often to express the Sanskrit **sh** before **p**, *e. g.* in pushpa, and may be intended for ≍ and to suit the vulgar pronunciation pufpa.

c5.s1.t4.23

It seems likely that these additional signs were the invention of people from N. India—the first Jain or Brahman pioneers in the South—whose attention would at once be attracted by these strange sounds; for though the Canarese-Telugu alphabets are mere adaptations from the Sanskrit, they were not complete and wanted signs for **ĕ** and **ŏ**. Now to this day, the people of N. India cannot distinguish or pronounce properly the Dravidian **ĕ**, **ē**, **ŏ** and **ō**, and it, therefore, follows that the adaptation was by northern people.

c5.s1.t4.24

The transitional stage continued till the end of the thirteenth century A. D., and includes a period of great literary activity not only as regards the Telugu and Canarese languages, but also in Sanskrit. The reforms of the Vedāntist Rāmānuja belonged to the twelfth century, and he obtained great influence in Mysore where he converted the sovereign (a Yādava of the southern dynasty of Dwārasamudra) from the Jain persuasion. This king appears to have encouraged Telugu literature (because, no doubt, it was thoroughly brahmanical and orthodox), as much as his immediate predecessors had encouraged the Canarese³; and Nannaya Bhaṭṭa (a native of the east coast) composed under his patronage (about 1180 A. D.) a Telugu Grammar in Sanskrit, and began a translation of the Rāmāyaṇa which was finished by another Brahman, also a native of the east coast, a little later[^1). These events are nearly contemporaneous with the final ruin of the Western Cālukya dynasty which fell in 1182, and then the Yādavas became independent both in the North (Devagiri) and South, and thus shared the greater part of the old Cera and Cālukya kingdoms.

c5.s1.t4.25

³ See Mr. Kittel's preface to his edition of Keçirāja's Canarese Grammar.

E. The Old and Modern Telugu-Canarese Alphabets

c5.s1.t5.1

*(Plates ix. and x.)*

c5.s1.t5.2

The next stage in the development of the northern Dravidian alphabets is the Haḷakannaḍa and old Telugu, between which it is impossible at present to establish any distinction. This alphabet dates from the end of the thirteenth century, and the distinction between it and the character I have termed transitional consists merely: 1) in the disuse of the few remaining older forms which I have described in the last section as being found in that alphabet, and the exclusive use of the new forms; 2) in the absence of distinction between **ḍ** and **ḍh**, **p** and **ph** and some other aspirates; 3) in the absence of marks to distinguish **i** and **ī**[^2). Between this alphabet and the modern forms the differences are but trivial.

c5.s1.t5.3

As will be easily understood in the case of an alphabet like this which was in use from the Canara coast to the mouths of the Kṛshṇā and Godāvarī, there were several slight varieties or hands, but it would take far too much space to notice here more than a few points, even though such details are of interest as partly subsisting up to the present time.

c5.s1.t5.4

The earliest important variation, I have noticed, is in the form of **t**. About 1300 this letter appears in inscriptions on the west (or Canara) coast with a double loop [char: Telugu/Canarese letter], that to the left is only partly closed, whereas on the east coast and the central territory between the two, the form [char: Telugu/Canarese letter] with a single loop is preferred[^3). In the modern Telugu and Canarese alphabets, this is exactly reversed. In the inscriptions in the Canarese country visarga is represented by a circle large enough to occupy the same space as the other letters, in the eastern country a very small circle only was in general use. Again the

c5.s1.t5.5

[^1): This poet (?) was named Tikkaṇṇa; he died in 1198 A.D. (Brown's “Cyclic Tables,” Madras edition, p. 58). Nāgavarma, the author of the Canarese Prosody, was also a Telugu from Veṅgi; his date is, however, uncertain. (Mr. Kittel's ed. p. xxv.) [^2): See Mr. Fleet's remarks on an inscription of A.D. 1510. *Indian Antiquary*, v. p. 73. [^3): For the Canarese forms I use an inscription on stone at Mangalore, d. ç. 1225 = 1303-4. I have an excellent photograph (by Messrs. Orr & Barton of Bangalore) from an estampage-impression by Mr. Kittel, and have examined the original; Mr. Kittel has also kindly given me a transcript.

c5.s1.t5.6

Canarese form of **k** (ಕ) was originally the most general one¹), whereas the modern Telugu ళ was confined in the fourteenth, fifteenth and sixteenth centuries to the northern part of the present Nellore district, where a very *round* hand has always prevailed. Owing to that part of the Telugu country having been one of the earliest British possessions in Southern India, this hand was adopted as the model, on the introduction of Telugu printing in the beginning of this century at Madras. At present, the Canarese is especially distinguished from the Telugu alphabet by the method of marking the long vowels **ī**, **ē** and **ō**, by the addition of a separate sign (—ೕ) following the consonant with the usual short vowel affixed; this is entirely wanting in Telugu. The earliest instance I have noticed is in a palm-leaf ᴍs. of the first half of the sixteenth century A. D., but it does not occur in any old Sanskrit ᴍss. in the Canarese character at all, nor commonly in Canarese ᴍss. till much later. The Telugu method of marking the short and long **e** and **o** does not appear till the seventeenth century. About this period apparently owing to the revival of Sanskrit studies for a time, the distinction between aspirated and unaspirated letters becomes again usual, and has continued up to the present, though really alien to the Dravidian languages. It began much earlier in Telugu than in Canarese, and even in the Sanskrit ᴍss. on grammar written in the latter character, it is but seldom made; a fact, which, by itself, proves the prevalence of oral teaching²).

c5.s1.t5.7

¹ See pl. xxxi.

c5.s1.t5.8

² Cfr. the alphabet given in pl. ix. I have already given a facsimile of a Canarese Sanskrit MS. of about 1600 A. D. in my edition of the Vamçabrāhmaṇa. The difference between the writing of MSS. of the fifteenth and sixteenth centuries is very slight; the body of the letters in the latter is not so large, or so round and close together.

c5.s1.t5.9

From the earliest inscriptions down to the latest, the gradual extension of the use of the bindu (o) is very remarkable, and appears a tolerably safe test of the age of a document. I shall therefore give briefly the results I have gathered.

c5.s1.t5.10

In the early inscriptions the Cera bindu is *above* the line, the Cālukya *on the line*³): but after the twelfth or thirteenth century it is always, and in all S. Indian characters, written *on* the line. This is even the case in the Nandi-nāgarī, though here, it, by being in this position, renders the writing unsightly.

c5.s1.t5.11

³ Irregularities however are found; see Mr. Fleet's remarks in *Indian Antiquary*, iv. p. 85.

c5.s1.t5.12

As regards the employment of the *bindu*, the broad rule is: the later the inscription, the more incorrect and indiscriminate is its use. In the earlier inscriptions it is seldom used for **ṅ**, **ñ**, **n**, and **m** before a consonant of the same class *in* a word; but it is used for all these nasals except **ṇ**, by the fifteenth century; and from that time to the present one occasionally finds **mḍ**. The common practice of using the bindu to express all the nasals, even including a final **m**, which some editors in Europe have copied from the more modern MSS. from N. India, is, therefore, a very old practice in the South, though it is most certainly erroneous according to the chief grammarians, and, therefore, as Profr. Whitney contends, is to be rejected, though convenient in practice¹. It is hardly necessary to remark that the *bindu* is properly the sign of the unmodified nasal or anusvāra.

c5.s1.t5.13

¹ Profr. Max Müller (Hitopadeça, p. viii. and S. Gr. pp. 6-7) allows it as a convenient way of writing.

c5.s1.t5.14

I have not noticed in any inscription the nasalized semi-vowel; it sometimes occurs in Telugu Vedic MSS. and then has the form of ఁ. Nor have I met with the ardhānusvāra to which some Telugu grammarians allude². The ఱ (ṟ) of the Telugu inscriptions is now disused³.

c5.s1.t5.15

² *v.* App. A.

c5.s1.t5.16

³ This letter is etymologically of significance; and, therefore, cannot be neglected.

c5.s1.t5.17

The use of visarga is generally incorrect in the inscriptions; it is seldom converted according to rule. In S. India the alternative allowed by the grammarians of assimilating visarga to a following sibilant is almost universally accepted, and the reduplication of the sibilant then omitted. This is a common source of error in reading S. Indian inscriptions and MSS. The separation of the superscript **r** from the following consonant (as pronounced) above which it is written, begins about 1300 A.D. After 1350 it is always on the right hand, *e. g.* **rka** is written ౯క (kr). By 1550-1600 A.D. the modern secondary form of **e** is always used, *e. g.* **ve** is written వె. (For the older form see Pl. viii.) In the fifteenth century both forms co-exist; in the fourteenth the modern form begins to appear.

c5.s1.t5.18

Allusions to the current alphabets are almost as rare in the S. Indian mediæval works as in the Sanskrit. Ātharvaṇācārya (who cannot be earlier than the end of the twelfth or beginning of the thirteenth century) describes the transitional alphabet just as it was changing into the earliest modern form ("*Kārikāḥ*" 29-32)⁴:

c5.s1.t5.19

⁴ I follow a transcript of the unique and very incorrect MS. in Mr. Brown's collection at Madras.

c5.s1.t5.20

29. *b.* pañćavargādayo varṇāḥ çaṅkha-(a ఱ)çārṅgā-(g ౯)disamnibhāḥ ||
30. tiryagrekhāyujaç čo 'rdhvam daṇḍarekhānvitā adhaḥ (✓ *and* |) |
ta eva ca dvitīyā(ḥ) syur ūrdhvam rekhādviranvitāḥ (*sic*) ||
31. prathamās tu tṛitīyā(ḥ) syus tṛitīyānte čaturthakāḥ |
rekhādvayādhodanḍena yuktā(ḥ) syur anunāsikāḥ ||
32. miladdaṇḍadvayopetāḥ prathamā paya\*smṛitāḥ |
pūrṇendusadṛiçaḥ pūrņas tv ardhas tv ardhendusannibhaḥ ||

c5.s1.t5.21

There is much here very unintelligible, but the description of some of the letters clearly points to about 1200 A.D. The Canarese "Basavapurāṇa" (of 1369 A.D.)

c5.s1.t5.22

mentions the Telugu, Canarese, Grantha, Tamiḷ (Drāviḷa), Lāḷa (*i. e.* Lāṭa or Gujarat) and Persian alphabets¹, and this would seem to indicate that there was then a greater distinction between the Canarese and Telugu alphabets than we actually find.

c5.s1.t5.23

¹ I owe this reference to Mr. Kittel; it occurs in ch. v.

c5.s1.t5.24

Nor is it quite clear what letters the mediæval grammarians considered to belong to the alphabet. Al-Bīrūnī of Khwārizm (who lived from 970-1039 A. D.)² puts the number of Sanskrit (Nāgarī) letters at fifty³; Nannaya Bhaṭṭa, in his Telugu Grammar (and of the twelfth century), also puts the Sanskrit letters at fifty, the Prakrit at forty, the Telugu at thirty-six⁴. The commentators are, however, not agreed as to whether both **x** (ksh) and **ḻ** are intended to be included among the Sanskrit letters⁵. Lassen (“Indische Alterthumskunde” iv. p. 796) takes the Ṛigveda **ḻ** to be the fiftieth letter of Al-Bīrūnī; it may reasonably be doubted if that was the view held in India. The Canarese Grammar includes both **x** and **ḻ**⁶.

Linked texts (6)

Basavapurāṇa — Sanskrit
Tanjore — inscription (SII 4.1289) — Tamil
Mahābhārata — Vyāsa, Sanskrit
Vālmīkirāmāyaṇa — Vālmīki, Sanskrit
Badami
Vikramāṅkadevacarita — Bilhaṇa, Sanskrit