Indian Palaeography c15.s3.t1: (a) Bengal
Indian Palaeography c15.s3.t1
- Language
- English
(a) Bengal
c15.s3.t1.1
Though the writing style of this region is derived from the Middle Ganges Valley, it has certain characteristic peculiarities. The Kauśāmbī style is seen as early as the Susunia rock inscription of Chandravarman, which is better executed than the Allahabad inscription of Samudra Gupta.⁴ The medial vowel *ā*, applied to the bottom right of *ṇā* in both these inscriptions, survives later only in Bengal, and this practice is further extended to *khā*, *gā*, and *bā*. The twisted slant in the medial vowel *e* and the curled form of the medial *i* appear here from the beginning, suggesting that ornamentation in the medial vowels first appeared in Bengal, and it is not unlikely that this twisting of the medial vowels led subsequently to the bending of the right verticals (i.e. *kuṭila* forms). Another practice already noted in the last chapter is the confusion between *sha* and *sa*, which resulted in the total dropping of the form of *sha* in the fifth-century copper plates of Bengal. Only dental *sa* was used for both. In all subsequent records in Bengal *ksha* is formed by combining *ka* with *sa*. Another confusion arose between *ba* and *va*, both of which are even today pronounced as *ba* in Bengal. This led to the dropping of the form of *ba* from the close of the sixth century onwards. It is probably from this source that similar confusion in these letters spread in the other regions. From the palaeographical point of view it is important to note that the style of Bengal does not have the foot-mark, though influence from the Middle Ganges Valley is clearly seen in the thick heads at the left limb of *ra*, *kha*, *ga*, and *śa*. The letters *bha* and *sa* do not have the solid triangular foot-mark at all. They show directly the open-mouthed triangle at this point, as is seen in *kha* of the Pāla period. The palaeography of this region falls into six chronological groups:
c15.s3.t1.2
⁴ See above, p. 101.
c15.s3.t1.3
(i) Pl. XI. 1 is taken from the Susunia rock inscription of Chandravarman, the palaeography of which has been fully described in the last chapter.¹ It is well executed on stone. The medial vowels, such as *e* and *i*, already show ornamental bending. King Chandravarman is generally identified with his namesake defeated by Samudra Gupta. If this is accepted, the inscription may be taken as the basis for palaeographical study in Bengal. As this inscription has a definite stylistic relation to the next group in Bengal, on palaeographical grounds it may be reasonably placed in the fourth century A.D.
c15.s3.t1.4
¹ See above, p. 101.
c15.s3.t1.5
(ii) Pl. XI. 2 represents the writing in the copper plates of the time of Kumāra Gupta I and Budha Gupta, dated between the (Gupta) years 113 (A.D. 432) and 163 (A.D. 482). The change of the writing material makes for the stylistic difference from (i) above. The Susunia rock style shows firm angular lines but the style as known from these copper plates may be termed ‘fugitive’, implying three things: (i) looseness in drawing the lines on the part of the writer; (ii) a tendency to curve the angles, as in *pa* and *ma*, and (iii) a preference for extremely cursive forms, as in *la* and *ha*. As a result of the first feature the head-mark becomes a mere line. No attempt is made to give the letters their full forms. This defect is obviously due to the metal worker who incised the inscriptions on the plates rather than to the scribe himself. As far as it is possible to get at the latter’s hand, there does not seem to have been any deliberate attempt to deform the letters.
c15.s3.t1.6
The initial vowels *a*, *ā*, *i*, *ī*, and *e* are similar to the forms seen in the Gupta inscriptions of the Middle Ganges Valley. But the base of *u* already has a curve downwards, and an optional form of *e* has an upper stem, as has been noted in the inscriptions of the later Guptas.¹ The difference from the Middle Ganges Valley form is also seen in the medial *o*, as applied to *kho*, *go*, *ṇo*, *bo*, and *yo*, where the right-hand mark is given at the bottom right, as in the case of the medial *ā*. The medial *au*, as applied to *tau*, also shows a variation in so far as the third mid-stroke is actually inclined to the left. The Kushāṇa system of a top slant for the medial *ā* is not found here, but it is seen in the case of the medial *e*. The medial *i* is a curve going to the left, and the medial *ī* has two arms in a curve, the right one longer than the left. There is no change in the other medial vowels. The left limbs of *kha*, *ga*, and *śa* have line heads instead of foot-marks, while the right limbs of *ga*, *ta*, *bha*, and *śa* are longer than the left. *Ka* has the curved mid-bar and *gha* maintains the straight base. *Cha* has pointed as well as rounded beaks. *Ja* has its lower arm slightly bent at the tip. *Ña*, as seen in the conjuncts *ñcha* and *jña*, has already developed the cursive form with three teeth. *Ṭa* flattens its top, and *ḍa* has its back slightly rounded. *Ṇa* is of the open-mouthed type with a loop at the left. *Tha* is oval with a mid-line. *Na* is of the looped variety. *Pa* and *pha* have rounded left angles, while *ba* optionally has a slight bend at the left arm. *Bha* keeps up the angular form. *Ma* has two varieties in the tailed form—in the first the base slopes to the right vertical, thus giving the first example of an acute angle in this letter, while the second is a constricted form in which the base curves upwards and replaces the right vertical. This method is also adopted for drawing the letter *pa*, which gives a shape closer to the later proto-Bengali *pa*. *Ya* is of the tripartite form with a loop at the left arm. *Ra* is a straight vertical. *La* and *ha* are of the hooked type but very cursively drawn, especially in the Baigram plate. In this plate alone we get the letter *ḷa*, of the same form as in the Allahabad inscription. The base of the triangular *va* has already started sloping to the right. As said before, *sha* is replaced by the form of *sa*, and the latter is of the looped variety.
c15.s3.t1.7
¹ See above, p. 127.
c15.s3.t1.8
(iii) Pl. XI. 3 represents the writing known from two late Gupta copper plates and five copper plates of Dharmāditya, Gopachandra, and Samāchāradeva—all belonging to the sixth century A.D. A study of these inscriptions shows a transition from the Gupta style noted in the second group to certain new features which reached their maturity in the later half of the sixth century A.D. To this last period belongs the Ghughrahati copper plate of Samāchāradeva. The two Gupta copper plates link this group with (ii) above, and the inscriptions of Dharmāditya and Gopachandra fall in between. Palaeographically the inscriptions of Dharmāditya show earlier forms, as has been rightly noted by F. E. Pargiter in his edition of the plates. The difference can be marked not only in the form of *ya*, as was maintained by Pargiter, but also in the forms of *a*, *tha*, *dha*, *pa*, *la*, and *ha*. The inscriptions of Dharmāditya use the sign for *ba* but do not have *sha*, as was the practice in (ii) above, but the inscriptions of Gopachandra have only the sign for *va*, and the Mallasarul inscription at least contains the form of *sha*. But this last plate mentions the name of Mahārāja Vijayasena, who is identified with his namesake mentioned in the Gunaighar plate of Vainya Gupta. Hence Gopachandra is generally taken to be a successor of Vainya Gupta, though the palaeography suggests that Dharmāditya should come between them. But this point need hardly be pressed further, when it is realized that by about this time greater changes had already taken place in the forms of the letters in the neighbouring province of Bihar, as attested by the inscriptions of the Magadha Maukharis. As the same change was to affect Bengal, it is hard to be dogmatic as to the exact time when the change started here. We must allow a margin for the transition period in which the new and old forms were used simultaneously, some inscriptions preferring the new and the others the old. The lesson must be learnt that the palaeography is of little help in settling the question of the succession of rulers when only a short duration of rule is involved.
c15.s3.t1.9
In all these inscriptions the head-mark is distinctly given. The Gupta form of *a* appears in the plates of Dharmāditya and used optionally in the plates of Gopachandra, but in the later plates the curved form of *a* is seen, as in the post-Gupta records of the Middle Ganges Valley. The initial *i* maintains the Gupta form, but *u* extends its curve upwards. The triangular *e* still has its base at the bottom. Only in the Ghughrahati plate is the base on the top. Only two medial vowels show change: the medial *ā* optionally uses the top slant of the Kushāṇa style, and the medial *i* prolongs its curve to the bottom of the letter. The medial *u* in *tu* is turned down, but in *gu*, *du* and *bhu* it is turned up. *Kha* has both round and flat-topped forms. *Gha* now divides its base into a sloping right half and a curved left half. *Ṅa* drops the loop at the left of the open-mouthed form. *Tha* maintains the oval form in the inscriptions of Dharmāditya, but the bulge form is seen in the other plates. Both *pa* and *ma* show an acute angle at the right. *Ya* has a new tripartite form with the outward curl at the left arm, as seen in the Amauna plate,¹ while the old looped form also occurs. In the plates of Gopachandra the bipartite *ya* appears. *Ra* optionally has an upward slant, recalling the foot-mark. Both *la* and *ha* have the hooked varieties in the earlier inscriptions, and in the later ones we have the full forms with the firm base line. *Sa* optionally has the triangle instead of the loop at the left. Among the conjuncts may be noted the forms of the double letters *ṭṭa* and *rjju*, in which the lower letters are twisted downwards. Similarly, *tha* in *sthā* has its oval in a horizontal position. On the whole the style suggests the gradual adoption of some of the forms which were already in use in Bihar.
c15.s3.t1.10
¹ See above, p. 126.
c15.s3.t1.11
(iv) Pl. XI. 4 is taken mainly from the Nidhanpur copper plate of Bhāskaravarman, but the Tippera plate of Lokanātha and the Kailan plate of Dhāraṇa Rāta have also been consulted. These inscriptions fall in the first half of the seventh century A.D. Many of the features seen in the Middle Ganges Valley at this time are found here. The head-marks are solid triangles. The right verticals are bent (*kuṭila*), but on the whole the ornamental forms are meagrely used.
c15.s3.t1.12
The form of *a* is of the curved type but develops an eastern variety by sloping the middle bar and lengthening the right vertical. The other initial vowels adopt the Middle Ganges Valley forms. For the medial *ā* the curved mark at the right bottom of the letters is now given up. The medial *i* has its curve going to the left and the medial *ī* to the right. The medial *u* in *ḍu*, *bhu*, and *ru* is curved downwards. *Ka* adopts the looped variety from the Middle Ganges Valley. *Tha* and *dha* show a definitely bulging body. *Da* is of the double-curved type with a tick at the lower curve. *Ṇa* also adopts the head form of the Middle Ganges Valley. *Pa*, *pha*, and *ma* all have the acute angle at the right. *Bha* still maintains the angular form. The *kuṭila* form of the bipartite *ya* is used here. *Ra* has a triangular foot-mark. Both *la* and *ha* have firm bases and develop their hooks in a downward curl. *Va* maintains the triangular form with its base sloping. *Sha* is used with a mid-line in the form of *pa*. *Sa* has the triangular loop. *Jha* is used in combination with *ja* and has an acute angle on the right. The *halanta* (vowelless consonant) is shown by a short slant below the letter as in *kāt*.
c15.s3.t1.13
(v) Pl. XI. 5 is derived from the inscriptions of the Khaḍga dynasty, which ruled in East Bengal in the later half of the seventh century a.d. Here fully developed *kuṭila* forms are used. The initial vowel *a* maintains its eastern variety by extending the left curve upward. Other vowels are similar to those used in the Middle Ganges Valley. The medial *ā* now develops a short vertical line on the right of the letters. The medial *i* and *ī* are extended to the bottom of the letters. The medial *ū* has a new form. It consists of a stroke with a head and is added to the left of the vertical at the bottom of the letters. This form is also seen in the Middle Ganges Valley, and it survives in modern Bengali. But in the cases of *bhū* and *śū* an additional curve is added to the medial *u*. In the consonants some changes are noticeable. In *gha* the acute-angled side is extended slightly downwards, and the curved side is slightly tilted upwards. *Cha* has a sharp pointed beak. *Ja* adopts the curved form from the Middle Ganges Valley, in which the upper limb merges with the head-mark. *Ṭa* develops a tick at the end of the flat top. *Ṇa* has a sloping base and the two outer arms are longer. *Tha* and *dha* are of the bulged form, with the former developing an inner loop. *Da* bends its tick at the end of the lower curve. *Bha* and *sa* open their triangular mouths. The acute angle in *pa*, *ma*, *ya*, and *sha* is very sharp, and their right limbs are prominently bent. *Ra*, as adopted from the Middle Ganges Valley in the earlier period, develops a tail at the foot. The base of *la* curves upwards and its hook shows an inner curl. *Va* is of the rounded type. *Śa* shows a loop in place of the rounded top and a slanting bar to connect it with the right vertical. *Ha* curves its sides in an ornamental bend with its right hook falling in a beautiful curl.
c15.s3.t1.14
(vi) Pl. XI. 6 mainly represents the Khalimpur copper plate of Dharmapāladeva and the two newly discovered Sālban Vihāra plates of Bhavadeva.¹ They have been selected to represent the typically Bengali style of the time. If we base our conclusions on the Mungir copper plate of Devapāla, we can hardly trace any difference from the contemporary writing of the Middle Ganges Valley. It was certainly due to the Pālas that many discrepancies between the styles of Bihar and Bengal were removed. The Bengali style is still preserved in the form of the initial *a*, which shows an extended curve at the left limb. The other form given in the plate is from the Mungir plate. The initial *ā* has now a vertical on the right of *a*, as is also the case in the medial *ā*. The medial *i* has two dots above and a tailed one below, but another form with a curve above two dots is also seen. The latter form is probably borrowed from Central India. *U* extends its curve upwards. The rare *au* is formed by adding an upward curve on the right to the middle of *o*, which has its base bent down and the angles rounded. *Kha* has two forms—one with an inner triangular loop, and the other with an outer triangle with its mouth open and joined to a vertical on the right by a bar. This latter form of *kha* is used in the spurious Gaya plate of Samudra Gupta.² *Ja* bends the middle arm downwards and curves the lowest arm still farther, while the top one is merged in the head-mark. *Ṭa* prolongs the right hand tick downwards, and *ṭha* has a head-mark, from which hangs the circular form. *Ḍa* has a curved back and a bent leg. *Ṇa* develops the cursive form by merging the two middle arms and reducing their length—a form already seen earlier in such conjuncts as *shṇu*. *Ta* has also two forms—one with two arms rounded, as in the older inscriptions, and the other with the left arm forming an angular hook and finally attached to the right vertical—a form seen in the proto-Nāgarī style. *Tha* is of the bulged form with its loop turned either inwards or outwards. Similarly, *pha* has its loop turned outwards. *Ra* develops a cursive variety of the tailed type and has a bent vertical on the right with a mid-bar on the left. Besides the round-topped form of *śa*, we have also the looped form as seen in the last group. In the conjuncts the form of *ñchi* is remarkable, as *ña* shows a cursive form almost identical with that of modern Bengali. The conjuncts *kshmī* and *ksha* also come closer to the modern forms. Other letters and the medial vowels are only one stage further developed than those seen in (v) above.
c15.s3.t1.15
¹ These plates are in the possession of the Director of Archaeology, Karachi, Pakistan.
c15.s3.t1.16
² See above, p. 127.
c15.s3.t1.17
On the whole the style of writing in Bengal starts with certain peculiarities of its own, but later is gradually brought closer to that of the Middle Ganges Valley. Only a few letters point towards the modern Bengali, and as a whole proto-Bengali has not developed. In fact proto-Nāgarī invaded this region and it was long before its influence disappeared and a local style quite distinct from that of the Middle Ganges Valley was evolved.
c15.s3.t1.18
*(b) Nepal*
c15.s3.t1.19
The inscriptions of Nepal were originally published by Bhagwanlal Indraji.¹ To these Bendall added others,² and later Sylvain Lévi³ republished the inscriptions. Finally, in 1956 Gnoli brought out a well-illustrated monograph on *Nepalese Inscriptions in Gupta Characters*. Most of these inscriptions are dated. One of the eras easily recognizable, and generally accepted, is that which is associated with the name of Harsha. There has been great controversy regarding the other era used in them. Fleet in his *Corpus*⁴ argued for the Gupta era, but Lévi proposed a new Nepali era starting in A.D. 110. The root of the controversy lay in Bendall’s inscription no. 1, in which the date was read as 316 or 318. This inscription mentions the name of Śivadeva I, who is known to have been a contemporary of Aṁśuvarman. The latter’s inscriptions are dated in the Harsha era (starting from A.D. 606) and range between 34 and 48. Quite naturally Bendall’s date was referred to the Gupta era (beginning *c.* A.D. 318–19). This reference upset the date of other Nepalese kings such as Mānadeva, whose inscriptions are dated from the year 386 onward. Fleet referred all these dates to the Gupta era. But this is impossible on palaeographical grounds as the details given below will show. Hence Lévi proposed a new era for the Mānadeva group of kings. But such an assumption is not now necessary. Gnoli has corrected Bendall’s reading of the date from 316/18 to 517, thus we have inscriptions regularly following a set era and later passing on to the Harsha era. There is no break in the palaeography, of which the gradual evolution is clear. The palaeography favours the attribution of the earlier years to the Śaka era (starting in A.D. 78). As no known historical events are against this attribution, the following palaeographical study is made by grouping the inscriptions in the sequence as we see them.
c15.s3.t1.20
¹ *Indian Antiquary*, ix. 163 ff.
c15.s3.t1.21
² *A Journey of Literary and Archaeological Research in Nepal and Northern India*, Cambridge, 1886.
c15.s3.t1.22
³ *Le Nepal*, vol. iii.
c15.s3.t1.23
⁴ Fleet, *C.I.I.*, iii. 95–97 and Appendix IV.
c15.s3.t1.24
The inscriptions of Nepal fall into five chronological groups, beginning with the (Śaka) year 386 and ending with the (Harsha) year 159. But for our purposes they are divided into three categories.
c15.s3.t1.25
(i) Pl. XI. 7 includes the inscriptions dated between the (Śaka) years 386 and 427, and Pl. XI. 8 has the inscriptions ranging between the (Śaka) years 428 and 516. Roughly speaking, no. 7 belongs to the later half of the fifth century A.D. and no. 8 to the sixth century. The manner of writing is doubtless derived from the Kauśāmbī style of the Gupta inscriptions, as the analysis of the main body of the letters will show, but the medial vowels agree with the system known from the fifth-century copper plates of Bengal. It seems therefore probable that this Kauśāmbī style reached Nepal through Bengal. But the Nepalese inscriptions have an individuality of their own quite distinct from that of Bengal. This is clearly marked in the manner of using the broad pen. In all the inscriptions in Nepal the head-mark is a well-formed solid triangle, but the Bengal copper plates show a line mark, and only the later writing has triangular heads. Again in Nepal the verticals show a definite foot-mark from the earliest inscriptions to the latest, but in Bengal it is absent. As in the inscriptions of Bengal, the Kauśāmbī style is continued here to the end of the sixth century A.D., and in fact in Nepal it survives even later, though in the Middle Ganges Valley the Western influence came in the beginning of the sixth century A.D. This Western tendency in some letters is seen in Nepal very vaguely in the first half of the seventh century A.D., but more definitely from the later half of that century. Again the ornamentation follows the style of Bengal rather than that of the Middle Ganges Valley.
c15.s3.t1.26
The initial vowels *a*, *ā*, *i*, and *e* are of the Kauśāmbī style. The verticals of *a*, *ka*, and *ra* have the foot-mark. All the medial vowels, except long *ī*, follow the Bengal style of the fifth century A.D. The most important to note is the medial *ā* in *khā*, *gā*, *ṭhā*, *ṇā*, *thā*, *dhā*, and *bā*, in which a hook is added to the lower right of the letters. This practice continued in Nepal in the case of *bā* even in the sixth century A.D. Nepal has the curve of the medial *i* to the left and that of the medial *ī* to the right from the earliest inscriptions, but in Bengal this practice started in the seventh century A.D. From the sixth century A.D. the curve of the medial *i* is extended to the bottom of the letter. Of the consonants the longer right limbs of *ga*, *ta*, *bha*, and *śa* are common to other styles in North India and similarly these letters, except *bha*, have triangular heads at the left limbs. But in Nepal the flat-topped *kha*, *ga*, *ṭa*, and *śa* are seen from the beginning and the round-topped forms lost favour after the fifth century A.D. *Gha* has a straight base in the fifth century, which is divided into a straight right half and a curved left half from the sixth century onwards. *Cha* is of the beaked type, while *chha* in the conjuncts is of the double-looped form. *Ja* has its vertical slightly bent and its lowest right arm curves at the tip. In conjuncts such as *ñja* the letter *ña* is cursively drawn and *ja* is slightly tilted. In the fifth century the open-mouthed type of *ṇa* has a loop on the left, but from the sixth century onwards this loop is given up. *Tha* is oval with a mid-line in the fifth century, but in the sixth it is almost oblong with a flat top. In conjuncts such as *sthu* it lies horizontally. *Da* develops an upward slant at the tip of the curve in the sixth century, while *dha* is oblong in the fifth but in the sixth narrows towards the bottom. *Na* is of the looped type. The left corner of *pa* is rounded, but it has a definite angle on the right. The left arm of *ba* is bent, while *bha* is of the angular form with a foot-mark at the left limb. *Ma* is of the tailed variety. *Ya* is tripartite with a loop at the left arm, but optionally divides the base line into a sloping right half and a curved left. The other characteristic *ya* of Nepal has an outer curl instead of the loop at the left arm—a form also seen in Bengal in the sixth century. In the photographs published by Gnoli this form appears in the inscriptions of Nepal in the fifth century, but in the reproduction of the same inscriptions by Bhagwanlal Indraji the loop is found in place of the curl. As this is a doubtful case, it has not been shown in the plate. If the curl is really present in the fifth-century inscriptions, this is its earliest occurrence in India. Both *la* and *ha* are of the hooked type. The base of the triangular *va* slightly slopes to the right. The hooked type of *sa* is found, but in some cases the loop is triangular. Of the conjuncts, *ṅgha* and *ṇḍa* take the usual form. Cursive forms are seen in *hla* and *tpra*. The cerebral *sha* is not used in Nepal until the middle of the seventh century A.D. Though Gnoli transliterates in the correct grammatical fashion, actually the form of *sa* serves the purpose of both, as is seen in the conjuncts. In the sixth century A.D. the lower *ṭa* in the conjunct *ṭṭa* is turned on one side, with its mouth downwards, as was also the practice in Bengal.
c15.s3.t1.27
(ii) Pl. XI. 9. This group includes the inscriptions, dated between the (Śaka) years 517 and 535 and up to the (Harsha) year 59. Roughly speaking, the inscriptions belong to the first half of the seventh century A.D. During this period of the rule of Harshavardhana in the Middle Ganges Valley new trends from that region reached Nepal. But no sudden changes were made. This group really represents the transitional stage leading on to the next (iii). In this period the only change in the initial vowels is seen in the form of *e*, which from this time onwards has a flat top—a practice in line with that of the Middle Ganges Valley. In the medial vowels the hooked form of *ā*, as seen in some cases before, is now dropped. The system known in Bengal at this time is now followed. *Ja* optionally has the new form, in which the upper right arm is merged with the head-mark and the two lower ones are bent down. *Ṭa* has a tick at the right end of the flat top. *Ḍa* has a rounded back and a bent leg. The base of the open-mouthed *ṇa* now slopes to the right, while the two outer arms are extended farther. In both *tha* and *dha* the left-hand curve is optionally wavy. *Na* is still of the looped type. *Pa* and *ma* have an acute angle, while *bha* has now a triangular foot-mark at the left limb. The tripartite form of *ya* has both a curl and a loop at the left arm. *Ra* has the foot-mark. The hooked types of *la* and *ha* now develop a round-based form, and optionally are also of the new forms from the Middle Ganges Valley, having a firm base and two verticals with a hook. *Sa* has a triangular loop at the left, and in rare cases opens its mouth. *Va* is optionally of the rounded type but also of the triangular form. The cerebral *sha* is still not used. The palatal *śa* is of the old type.
c15.s3.t1.28
(iii) Pl. XI. 10, 11. No. 10 includes the inscriptions dated between the (Harsha) years 60 and 95, and no. 11 those between the years 103 and 159. It is from this period (i.e. the later half of the seventh century A.D.) that the forms associated with the *kuṭila* character in the Middle Ganges Valley influence the style of writing in Nepal. But here the local forms were not altogether given up. They survived to the end of this period. Changes are noticeable in the initial vowels. The form of the curved *a* connects it with Bengal, where the lower curve at the left limb is extended farther. The form of *u*, as seen in no. 11, with a rounded vertical and upturned base, preserves an old type and is peculiar to Nepal. Similarly, the old form of *e* with its base on the right is a very late survival. The medial vowels optionally preserve the old style, but the new tendency is also seen in the addition of a vertical on the right for the medial *ā* and the prolongation of the curves in the medials *i* and *ī* to the bottom of the letters. *Ka* develops an additional right limb attached to its cross-bar, and in no. 11 has the triangular loop at the left. In the eighth century a.d. *kha*, *ga*, *bha*, and *śa* have triangular foot-marks at their left limbs, and their right limbs are slightly bent. At this time the triangle of *bha* and *sa* opens its mouth. *Gha* develops the acute-angled form farther. *Ja* is of the old three-armed form, and also the new form with its two bent arms. *Ṭa* prolongs its right tick farther down and *ḍa* farther rounds its back. The base of *ṇa* slopes and the outer right arm is bent, but no cursive form is used even in a conjunct such as *ṇḍa*. *Da* develops a tail at the end of the lower curve. Both *tha* and *dha* keep up the local forms, with a wavy curve attached to the right vertical, and also are of the new bulged forms. *Na* preserves the looped type, but in the eighth century a.d. we find the form with a head on the left joined by a bar to the right vertical. Both *pa* and *pha* are found in *kuṭila* forms, but in the eighth century a.d. *pha* turns its loop outwards, as was also the case in Bengal. *Ya* is of the tripartite form with a curl at the left arm, but the looped variety comes to an end in the seventh century a.d., and the bipartite form begins to appear at this time. *Ra* has the triangular foot-mark but does not develop a tail. Both *la* and *ha* have firm bases with their hooks falling in an inner curl, as in the Middle Ganges Valley. *Va* is of the rounded type. *Śa* keeps up the old flat-topped form to the end. The cerebral *ṣha* appears in the later half of the seventh century a.d. The dental *sa* has the triangular loop and sometimes opens its mouth. On the whole we find the older forms persisting as optional cases in the medial vowels, and in the letters *a*, *i*, *ja*, *tha*, *dha*, *na*, *bha*, *ya*, *ra*, *śa*, and *sa*. In this survival of the older forms Nepal maintains its individuality.