Located in Vidisha District, Madhya Pradesh, the area of Badoh-Pathari is home to a rock shelter with a sculpted panel depicting seven mother goddesses. A weathered inscription next to the sculptures was reported as early as 1926. The inscription is dateable to the fifth century on the basis of its palaeography and the art-historical dating of the site. Though partly effaced beyond hope of decipherment, roughly half of the text can be read with confidence, while some of the rest may be restored conjecturally, and some speculatively. The epigraph pays homage to Rudra and Skanda in addition to the Mothers themselves, and is thus a key resource concerning mātṛ worship in the Gupta period. It mentions the otherwise unknown local ruler Jayatsena of Avamukta (a region also named in the Allahabad pillar inscription), and may refer to the reign of Kumāragupta (I).
Off the beaten path in Eastern Malwa there is a pair of villages, Badoh and Pathari. The surrounding landscape (Fig. 1) is dotted with monadnocks that tell a tale of geological time, microliths in the soil speak of human prehistory, rock paintings bridge the gap to history, and stone monuments—some now hidden in the soil—bear witness to the Gupta period and later historic times.1 Because the area provides excellent constructional sandstone, intensive opencast mining threatens the remains, only the most prominent of which are under protection. An inscription of the time of Kumāragupta (GE 116, ca. 436 CE) found in Tumain, about 80 kilometres to the northwest, mentions a place called Vaṭodaka, probably as the birthplace of Harideva and his brothers, who commissioned that inscription.2 Vaṭodaka has been identified as present-day Badoh. A later inscription in neighbouring Pathari, engraved in the reign of the Paramāra king Jayasiṃhadeva II (VS 1326, 1269 CE), mentions the locality by the name Vaḍovā.3 The name Pathari comes from the dialectal word paṭhār (meaning a hillock or plateau), and both the modern villages evidently occupy parts of the area of ancient Vaṭodaka. J.D. Cunningham (1848, 305), who visited the site in the first half of the 19th century, recorded that according to local tradition the ancient name of the place was “Barnuggur,” which he believed was derived from varāha-nagara, but which must in fact be the vernacular form of vaṭa-nagara.
In addition to Tumain, nearby centres of Gupta-period culture include Eran and Udayagiri. A mere 18 kilometres to the north-northwest, Eran (ancient Airikiṇa) is best known for an inscription of Samudragupta and an ensemble of Vaiṣṇava temples which includes a pillar inscribed in the time of Budhagupta as well as a colossal theriomorphic Varāha sculpture inscribed in the first year of the reign of Toramāṇa. Udayagiri is a well-known cultic site of the Guptas with a number of rock-cut shrines and rock reliefs, most of which were probably created during the reigns of Candragupta II and Kumāragupta I as indicated by three inscriptions at the site.4
Much like Udayagiri, the Gupta-period remains at Badoh-Pathari (aside from a few scattered finds), consist in rock-cut shrines, rock shelters and cliff reliefs. Indeed, the larger such site, a place called Sātmaṛhiyā near modern Ramgarh, has been described as “a sub-imperial version of Udayagiri.”5 The smaller site is at the southwestern foot of Jñānnāth Hill (23°55′31″N, 78°12′42″E) and is comprised of a large rock shelter with a small, possibly artificial, cave shrine and the remains of some small, later temples. The cave probably used to enshrine a four-faced liṅga.6 Within the rock shelter and to the right of the shrine there is a sculpted panel depicting the seven mother goddesses (Fig. 2). The relief is eroded, but still in much better condition than the stylistically similar Saptamātṛ panels at Udayagiri. It shows the goddesses in bhadrāsana on separate benchlike seats.7 At the far left, a male figure sits in lalitāsana on another bench, identified as Śiva because he is ithyphallic. This row of eight is in a recessed panel carved into the rock, about 50 centimetres high and 280 centimetres wide, positioned 60 to 120 centimetres above the current ground level, which slopes to the right. An inscription occupies a smoothed and slightly recessed area about 100 centimetres wide by 30 tall, immediately to the right of the group. The upper edge of the sculpted panel is level with that of the inscribed one. There are no further associated sculptures, but the rock surface is similarly smoothed in a second panel above the inscription. In 2004 the silhouette of an elephant was still recognisable in the traces of paint in this upper panel,8 and there may have been a different painting here at the time the inscription was engraved. Scattered across the rock shelter are vestiges of other paintings, a number of carved petroglyphs mostly depicting liṅgas and their worship in clumsily stylised lines, and several engraved graffiti including some in Gupta or early post-Gupta characters. At about waist height on the far left of the shelter, there are vestiges of an all but obliterated second inscription consisting of at least four lines, probably also in Gupta-period script, though not a single character is readable with certainty. Further rock shelters on lower levels, some painted, were still in evidence in the vicinity in 2004. Most of these have been destroyed by quarrying or filled up with debris.9
Indologists have been aware of the goddess panel inscription for almost a century thanks to the report of M.B. Garde (1926, 12, 25). The inscription was also reported, without any additional information, by H.N. Dvivedi (1947, 90–91 No. 661) and in the List of Archaeological Monuments in Madhya Bharat (Patil 1952, 109), and the Annual Report on Indian Epigraphy for 1961–62 (ARIE 1961–62, 173 No. 1695). J.D. Cunningham (1848, 311) appears to report the cave shrine, but does not mention the goddesses or the inscription. Remarking that the text was largely illegible due to damage, Garde deciphered enough of it to learn that it mentions a day (śukla trayodaśī) at the beginning of the seventh line, but the preceding text that would have recorded the year and month are lost beyond hope of retrieval. He also read the phrase bhagavatyo mātaraḥ in the same line, deducing that the inscription must concern the installation of the goddess images; and the words viṣayeśvarasya mahārāja-jayatsenasya in line 8, noting that since the subsequent characters are illegible, the reigning ruler at the time the inscription was created may have been a descendant of this Jayatsena. With the recorded date lost, Garde assigned the script to the fifth century on a palaeographic basis, and this has been confirmed by several art historians10 who observed that the style of the sculptures is comparable to and hence roughly contemporaneous with those of Udayagiri, though as Harle cautions, the provincial characteristics of the Badoh-Pathari panel preclude an accurate chronological assessment.
No scholar since Garde has attempted to obtain a fuller reading, presumably due to the bad condition of the inscribed surface coupled with the remoteness of the site. Another factor in this lack of interest may have been the implication of Garde’s report that the epigraph furnishes no historic data beyond that stated above, and that the rest of the ten-line text is not worth attempting to read in terms of “return on investment.” While a substantial part of it is indeed effaced beyond any hope of decipherment, bits and pieces can be read with confidence, and the partially damaged areas offer tantalising clues that invite conjecture and speculation. As antiquarians increasingly recognise information other than the names and dates of prominent personages as a genuine, interesting and research-worthy subject, the Badoh-Pathari inscription may merit attention once again. In particular, the cult of the Saptamātṛ(kā)s has been studied in great depth11 and, though most experts on the subject are primarily historians of art and culture, the pertaining inscriptional evidence has not been neglected.12
However, such evidence is available in a lamentably meagre quantity for the Gupta period (and not at all for earlier times). There is in fact only one known inscription earlier than the sixth century with more than a passing reference to a group of mother goddesses as patrons/protectors or as the deities of a temple. Dated 430–431 CE,13 the Gaṅgdhār inscription of Mayūrākṣaka records the construction of a Viṣṇu temple by a minister of the Early Aulikara king Viśvavarman, along with a subsidiary shrine dedicated to the Mothers. One (slightly lacunose) stanza of the long inscription (verse 23, in lines 35 to 37) is dedicated to this shrine, and half of that stanza describes the goddesses. On account of the scarcity of similar references, this brief and rather obscure passage has fuelled much discussion in the context of goddess worship and Tantrism.14
The Badoh-Pathari inscription is from the same time bracket as the Gaṅgdhār inscription, and even in its fragmentary state it reveals more of the mātṛ cult than any other known epigraph of the period. Moreover, unlike the Gaṅgdhār record (the monuments mentioned in which have not been found and are in all probability not recoverable), it comes to us complete with a carved set of Mothers.
Figures 3 and 4 below show the inscription. Figure 3 is a digital composite of numerous detail photographs, most of which were taken by the author in January 2017, with some being taken at a second visit in February 2018. Each picture was illuminated with grazing light from a handheld source positioned close to the surface. The compositing and enhancement process involved geometric transformation and individual cutting of the best-lit areas from each original photograph, as well as global exposure and sharpness alterations, but no details in any image have been selectively altered by hand. Figure 4 is an eye tracing of the text, where clearly visible lines are shown in black, while damaged and uncertainly identified strokes are drawn in grey; wholly obliterated strokes confidently or tentatively restored are shown in outline. Both these images, along with some additional variations, are available for download in high resolution from the Zenodo archive.15
As indicated above, the inscribed area is about 100 centimetres wide and 30 centimetres high. The inscription consists of nine lines, badly weathered. The condition of the engraving is best at the top and near the left edge, deteriorating to the right and downward. Most of the first four lines are at least tentatively legible, but only the beginnings of the remaining lines can be made out. There was probably a tenth line that is now wholly effaced. The lines are slightly uneven, but the left and right margins are quite straight. Line height (measured as the distance between the tops of character bodies in one line to the tops of bodies in the next line) is about 2.5 centimetres, and the characters themselves have bodies 10 to 13 millimetres tall. Characters with large descending components (such as hmyā in line 3) thus share space with the ascenders of the next line.
The script is of the type generally referred to as the southern class of late Brāhmī. Most strokes are curved, resulting in rounded glyph shapes. Characters as a rule sport prominent headmarks which may be either oblong rectangular (box heads) or triangular (nail heads), without any readily apparent system determining the shape. Thus, ta has a nail head in all clear cases (e.g. prāvṛtya, pretālaye, l1; though some of the unclear specimens, such as tasyā° in l2, may have rectangular heads); ga may have a box head (gaja, mahoraga, l1) or no headmark (yogaiḥ, ugra, l3; bhagavatyo, l7), while ra usually has a nail head but may have a box head (mahoraga, l1). Rectangular and triangular headmarks appear to alternate freely in sa and ma, though erosion interferes with the clarity of the shapes. Characters consistently lacking headmarks include ja, ṇa and, probably, gha. Chhabra (1948, 132–133) points to a similar mix of headmarks in the Bamhanī plates of Bharatabala. The strokes for the vocalic ṛ and for subscript r and the single specimen of subscript c (but not subscript y, as a rule) are enlarged. This probably serves a decorative purpose, but there is no other ornamentation discernible in the text.
The inscription bears a general palaeographic similarity to the above-mentioned Bamhanī plates of Bharatabala (dated 455–456 CE) and Tumain inscription (GE 116, ca. 436 CE), as well as to the Udayagiri Cave 6 inscription of the time of Candragupta II (GE 82, ca. 402 CE). Other inscriptions with similar script styles include a number Valkhā copper plates bearing dates from ca. 50 to ca. 100 (in all probability of the Gupta Era, thus around 370 to 420 CE) and the later Sunao Kala plates of Saṃgamasiṃha (Kalacuri Era 292, ca. 540 CE).
The characters ra and ka always have an elongated stem with a hook at the bottom. There are no extant initial vowels in the text, but presumably initial a would have had a similar stem. Ma has the southern form with a closed loop and a headmark on both of its arms. All clear specimens of ma have a rather triangular body, with the arms starting from the apex, giving the appearance of a glyph executed as a single looped line. La has a vertically elongated stem that curves back to the left, but does not continue downward on the left of the body. This form is used in several Valkhā plates and in the Udayagiri Cave 6 inscription, whereas in the Bamhani and Sunao Kala plates the tail of la encircles the entire body, and the Tumain inscription uses a quite short la that may or may not curve left at all. Bha is of the broad type, with a curved limb that dips in the middle, then joins the vertical stem below the headmark. Valkhā plates offer plenty of similar specimens, but most of the comparable inscriptions noted above have the limb and the stem joining only at the head. The upper and lower halves of ja are symmetrical, with arms that are quite straight, and a left edge that may be a straight vertical or may be notched in the middle. Valkhā plates tend to use a form of ja in which the lower half is wider than the upper one; most other inscriptions noted above also differ in small details of the shape of ja. Ca has a rounded oblong body with a very pronounced dip in the upper outline, but no conspicuous notch in the bottom. The closest parallels of this form occur in Valkhā plates and in the Sunao Kala record, but the specimens in the other inscriptions considered here are also quite similar overall.
As noted above, the text has no surviving independent vowels. Marks for the dependent vowels ā, e, o (and, probably, ai) normally consist of strokes above the character body, slanting to the left and/or right. Most of these vowel strokes follow a rounded curve, but some start out almost vertically and have a sharper bend, while a select few may start horizontally and bend sharply downward. Certain consonants regularly attach such strokes to the body rather than the head, thus ṇā in l1 (and, possibly, ṇai in l2), mā (generally, but not in mātṛvan, l6) and probably brā (l3); a badly weathered lo (l4) is tentatively identified but if correct, it has the vowel strokes attached to the middle of the stem, so that the left-hand stroke is encircled by the curve of the stem. The sign for i is circular and may or may not be fully closed, while ī is represented by a counterclockwise spiral open on the left. Dependent vocalic ṛ is represented in multiple specimens, all of which have the usual shape of a curved subscript stroke curling in on itself at the end. The vowel mark for u is attached to most consonants as a simple vertical stroke (nu, mu, yu, lyu, su), but a hooked u mātrā is also found (gu, tu, ntu, śu). Most comparable inscriptions seem to have a strong preference for the hooked u, but the plain form is attested in the Udayagiri inscription (nu and ṣṇu). Instances of ū are unclear but appear to be of the regular form.
There are two disputable occurrences of the upadhmānīya (ḫpū, l2; ḫpra, l3), both of which may be merely headmarks enlarged by erosion, combined with the omission of a preceding visarga. If they are upadhmānīyas, they probably resemble a small horizontal figure-eight attached to the top of the following consonant. The extant text includes a single specimen of halanta m, a diminutive subscript character which has distinct headmarks like the full form and no discernible horizontal stroke above it. No clear numerals, punctuation characters or other symbols are preserved, though what looks like a short horizontal stroke positioned near baseline height at the end of verse 4 (line 6) may be a punctuation mark employed to separate the verses from the following prose. Elsewhere, hiatus serves the function of punctuation, as in the halanta m of siddham at the beginning and at the ends of some verses and verse quarters.
The orthography of the inscription corresponds to the conventions of the period. Consonants following an r are doubled, with the possible exception of a tentatively reconstructed rbha in line 2. Consonants preceding an r are never geminated in the legible text, nor are there any instances of gemination in conjunction to y. The word tatva may be written with a single t, as commonly found in epigraphs, in line 4. The use of the visarga is close to standard, though the conjunct ss enjoys preference to the combination ḥ s, and as noted above, there may be an upadhmānīya preceding p in lines 2 and 3, though the regular visarga is definitely used before p in line 7. Anusvāra appears in a few places where a nasal consonant would be standard: certainly in nukaṃpī (l1), probably in °āṃjali (l3), and possibly in caraṃty (l3); conversely, a nasal consonant is used once instead of a standard anusvāra if I read/reconstruct the words tatvan na (l4) correctly.
The text opens with the word siddham spelled out in letters, followed by a laudatory stanza dedicated to Rudra. Most characters in the first line are, to a varying extent, blurred because the stone surface has flaked off around the engraved lines. The first two quarters of the stanza are quite confidently legible and say prāvṛtya yaḥ sa-rudhiraṃ gaja-carmma raudraṃ pretālaye niśi mahoraga-koṣṭha-sūtraḥ. The god is thus described as draped in a bloody elephant skin and wearing a snake for a waistband in the funeral grounds at night. The word sūtraḥ is tentatively read and I have not found the compound koṣṭha-sūtra attested anywhere. However, udara-bandha is widely used, where udara is synonymous to koṣṭha, while the interchangeability of bandha with sūtra in such a context is demonstrated by the attestation of the words kaṭi-sūtra and kaṭi-bandha, both meaning a waistband.
The third pāda is badly damaged; only the characters tṛgaṇā near the end are reasonably clear. The largely obliterated character before these had two headmarks and a rounded body like m, so the reading mātṛgaṇā is secure. The characters after this are most likely nuyāto; alternatives may be possible, but the reading is plausible and fits the blurred vestiges very well. Rudra is thus accompanied by the band of Mothers.16
The fourth pāda in all probability reads rudro jayaty amara-vandita-pāda-yugmaḥ. The text from tya, at the beginning of the second line, is only slightly damaged and entirely secure. At the end of the first line, the consonant r is mostly clear, as is the subscript r of the following character. The rest of rudro jaya is conjecturally restored but practically certain. Faint traces of the o mātrā of dro and the left-hand curve of ya can be made out in the stone.
Returning to the lacuna at the beginning of the third quarter, the traces of the first character suggest bha or ca. The next two characters are completely indistinct but both seem to be ligatures. The first of these may well incorporate a repha or a damaged i mātrā attached to the head, which is slightly lowered to accommodate this superscript stroke. The fourth akṣara’s consonant is almost certainly r. This is followed by what looks like t (possibly g or ś), perhaps with part of an i mātrā preserved; then probably ś (possibly t or g), likely also with a small i on top, and finally one more character with two curved sides, possibly t or n. I have no confident solution for the problem of this locus. For the first few characters, bhasmākta or bhasmāṅgarāgi- may be appropriate in the context, but I fail to find a plausible continuation to these. The same applies to strings beginning with bhakta-, bhadra- and bhīma- or bhīṣma-. The word preceding mātṛ may be śata, but the compound śata-mātṛ-gaṇa would strike me as very awkward. Taking the context into consideration, the subordinate sentence with yaḥ lacks a finite verb in the legible sections, though one is implied by the absolutive prāvṛtya and by the adverb niśi, neither of which would fit smoothly, in my perception, into a nominal sentence. For this reason I assume that this lacuna must hide a finite verb, and tentatively reconstruct the end as karoti śiva-mātṛ-gaṇānuyāto, understanding the verse to say that Rudra does something accompanied by the band of benevolent Mothers. The term śiva-mātṛ is used in the Āraṇyakaparvan of the Mahābhārata where Skanda grants a group of goddesses their desire to be worshipped as mothers of all the world, and instructs them to be of two kinds: malevolent and benevolent.17 However, the akṣara I propose to read as va does not seem to have had a bottom stroke, so the word śiva remains doubtful. As to the object of karoti, the word caryyāṃ (perhaps with ry instead of ryy, and possibly with ṅ conjoined to the following k in place of an anusvāra) is the least dissatisfactory solution I have been able to find. This matches both the context and the vestiges fairly well, though ceṣṭaṃ/ceṣṭāṃ and bhikṣāṃ/bhaikṣaṃ appear barely less plausible.18 At any rate, this segment of the text remains a moot point, and my restoration is rather in the nature of a diagnostic conjecture.
The second stanza, more damaged than the first, is another laudatory verse. It is fairly common practice in the Gupta period to begin inscriptions with a verse in praise of a deity, and to follow with a second verse in praise of a king or emperor. A link is usually established between these two stanzas by using the verb jayati in the first and the phrase tad-anu jayati in the second, with some variation of the phrasing.19 In the present case the phrase is tasyānu … jayati,20 and the addressee appears to be another god rather than a king. The stanza is badly damaged and the legible parts do not include a clear identification of this deity. He is male, and is described in the first quarter of the stanza as benevolent to devotees (bhakta-jana-nitya-hito) and compassionate ([a]nukampī). In the second quarter, only the word jayati can be read confidently. This verb was probably preceded by the name (or at least an unambiguous identification) of the god, which definitely begins with vī and almost certainly continues with rā. The remaining two characters are almost wholly obliterated, but the first must be prosodically short, the second long; the vowel of the latter is probably ā or o.
Although compassion and benevolence to devotees are principally associated with some avatāras of Viṣṇu, we expect this deity to be connected to Śiva and the Mothers. The most plausible candidates are Vīrabhadra (one of the leaders of Śiva’s gaṇas or a particular form of Śiva), Skanda (who is traditionally associated with groups of mother goddesses, being the son of the seven Kṛttikās) and Gaṇeśa (who in post-Gupta times becomes the standard companion of the Saptamātṛs, replacing Skanda).21 The characters vīrā could suggest Vīrabhadra, also frequently called Vīreśvara. This latter name would fit the prosodic template, but re can be confidently excluded, since the stone is intact where the e mātrā would be expected. Nonetheless, phrases such as vīraś ca so or vīras tato cannot be ruled out, and Vīra alone may be a name of Vīrabhadra. The damaged part of the quarter stanza following jayati might be read as bhūta-gaṇādhināthaḥ, which too could be an appropriate description of Vīrabhadra. However, this fierce martial deity does not strike one as compassionate and caring. These attributes are much more applicable to Gaṇeśa, who could also be aptly described as a bhūta-gaṇādhinātha, but who is not as a rule very heroic (vīra).
To complicate the matter further, bhūta-gaṇādhināthaḥ is a very tenuous reading. The first character of this string is largely indistinct, but close examination shows that it may have had a double headmark, indicating m. The second character is nearly as damaged, but a t component is very likely. There is, however, a long slanting line underneath it that seems to extend into the spot of damage below the previous character (where the ū of a putative bhū may have been). Taking these two observations into consideration along with the context, mātṛ is perhaps a more likely restoration than bhūta. Any of the three deities mentioned above may be described as lords of the host(s) of Mothers, and Skanda is definitely a heroic personage, although compassion is not a core quality of his.
Before coming to a conclusion about the god addressed here, we need to scrutinise the rest of the stanza for clues. The third quarter does not help with the identification. Its beginning can be confidently read in spite of damage as ya pūjyate satatam. There is definitely no visarga after ya, but one may have been omitted by the engraver or, more likely, there may be a damaged upadhmānīya attached to pū. In the last quarter, engraved at the beginning of the next line, the strings praṇava and, at the end, pahārayogaiḥ are clear. The rest of the text is obscured by damage, but the most likely reading is rijyāṃjali-praṇava-baly-upahāra-yogaiḥ, with the initial r belonging to the word at the end of the third quarter (presumably forming part of -bhir or -air, a plural instrumental ending expressing the agent of pūjyate). In this phrase I take yoga to mean simply “method,” in essence signifying nothing more than an emphatic instrumental (as in English “by means of”) attached to the dvandva elements preceding it in the compound. Ijyā is a rather generic term meaning worship or offering, though it may be used in the more limited sense of a Vedic sacrifice. Añjali is again quite vague and probably means nothing more than salutation or veneration in a general sense; it may refer specifically to the añjali gesture as an act of homage. In the context, praṇava probably indicates chanting or recitation and is not restricted to the syllable oṃ. Instead of ijyāṃjali, the reading ijyā-śruti may be possible. In this case I would construe śruti-praṇava as “recitation of the Veda.” This would be apt in the context, but the reading is definitely less likely than the one first proposed.22 Bali and upahāra both mean offerings; to eliminate redundancy, I assume that the former refers to animal sacrifice and the latter to non-violent oblations such as flowers.
The fragments gathered so far are still vague, but they have much in common with some textual references to the cult of Skanda. Notably, the Āraṇyakaparvan of the Mahābhārata23 mentions appeasing malefic hosts of mothers and male grahas24 called Skandagrahas through a ritual involving bathing them, offering them incense and ointments, bali and upahāra and, in particular, the ijyā of Skanda. Even closer to the mark, there exists a text that describes what this ijyā may have been: the Skandayāga or Dhūrtakalpa. Being the twentieth pariśiṣṭa of the Atharvaveda, this scripture (or something close to it) was presumably familiar to royal priests in fifth-century central India.25 The Skandayāga describes Skanda as a youth surrounded by the hosts of Mothers,26 corresponding to the probable mention of the mātṛ-gaṇa in the verse (to which I shall return below). The text requests the god to accept praise (corresponding to añjali in the inscription?) and offerings (bali and upahāra as in the inscription) presented with devotion (bhakti), in turn expecting him to be well-disposed toward his bhaktas and endow them with material wealth and prestige, of which the inscription’s nitya-hita and anukampin may well be a paraphrase.27 The ritual of the Skandayāga involves the recitation of numerous verses from the Atharvaveda,28 which may be implied by the word praṇava in the inscription. It is even possible that -yogaiḥ, though clear in the stone, is an engraver’s mistake for -yāgaiḥ, which would be a fairly certain indication of the specific text called Skandayāga.
Given this amount of circumstantial evidence, it is safe to conclude that the second verse of the inscription describes Skanda. Returning with this assumption in mind to the problematic parts of the stanza, the locus with vīrā may perhaps be restored as vīrārbhako. I offer this as a desperate conjecture and a better restoration may eventually be found, but at least it does fit the context as I understand it, and nothing visible in the stone contradicts it.29 The reading mātṛgaṇa (replacing the first suggestion, bhūtagaṇa) can now be taken as confirmed by the context. Moreover, mātṛgaṇais sanāthaḥ seems to fit the traces in the stone even better than mātṛgaṇādhināthaḥ, though neither of these readings can be established securely.30 If sanātha is correct, then the verse depicts Skanda as a helpless baby (arbhaka) who, in spite of simultaneously and paradoxically being a hero (vīra), needs guardians (nātha) to look after him and finds them in the persons of all the mother goddesses. But the message of the verse would be equally coherent with adhinātha, which would imply that that Skanda, expressly said to be amenable to the worshipper’s prayers, can exercise control over the mothers, who on their own would be dangerous without such control. In this case, arbhaka would simply be a synonym of kumāra, without the overtone of helplessness. There is also a roughly contemporaneous partial parallel for this thought: the first Bihar Sharif pillar inscription of the time of Skandagupta (ca. third quarter of the fifth century) speaks of the mother goddesses as headed by Skanda.31
The third verse is in even worse condition than the second. Some of its words can be read with a fair degree of certainty, but I have struggled to coax a coherent structure from these fragments and may be wrong about the most likely readings of the badly damaged loci. The problematic metre of the stanza adds another level of uncertainty. It seems to be mandākrāntā, but if so, then the third quarter is hypometrical, lacking one of the sequence of five short syllables in the first colon. The corresponding sections of the other three quarters are all damaged, and though they do seem to have the expected five short syllables, I cannot altogether exclude the possibility that these are likewise one syllable short. There is, however, no metre corresponding to this prosodic pattern in Apte’s classified list of Sanskrit metres (V.S. Apte 1957 Appendix A. II), so a faulty third quarter is more likely than a stanza in an unknown metre.
At the beginning, the akṣaras brā and hmyā are practically certain in spite of the damage obscuring them. The third and fourth characters are indistinct except for an ī (or possibly i) mātrā over the third. The fifth akṣara is badly damaged but seems to be a ligature of three consonants with v as the last, probably s as the second, and possibly another s as the first. The first words are thus probably brāhmy-ādīkās, though they could perhaps be brāhmy-ādīnāṃ or brāhmy-ādīkān. The first of the latter two was my tentative reading of this locus at an earlier time,32 but then the fifth character would form the beginning of a word and would thus need to be read as sva or some other combination of no more than two consonants, which contradicts the traces in the stone. As for brāhmy-ādīkān, it would require a masculine plural object to qualify, and no such object can be found in the context.
The sixth character is again indistinct, but its two headmarks and a slanting line connecting to the right-hand headmark are preserved, so this akṣara must be ma. The seventh is wholly lost except for a vestige of its headmark, but it was quite narrow and the intact surface around it shows that it had no extensions either above the headline or below the baseline. Thus, ta is a likely guess, but other values such as va or na cannot be excluded, and it is even possible that the marks at this locus are only noise or that they form a larger glyph with what I read as ma. The next two characters look much the same, with two legs descending from a headmark. The second is smaller and may have something of a neck between the head and the legs, so I read the first as ga and the second as ta. However, the second could, on its own, be bha or ga or even na; or, combined with what I perceive as the left arm of the next character, could form sa. The last akṣara of this string is again badly damaged, but the edges of the damage on the left, bottom and right correspond to the outline of a ya. The vowel mark above it looks like au, but being unable to find a meaningful reading with yau here, I attribute the straight vertical component to damage and identify the two curved strokes as an o mātrā.
Considering all alternatives including the possibility that there may be only four short syllables here (as in quarter c), the most intelligible restoration obtained is brāhmy-ādīkās sva-mata-gatayo. Here brāhmy-ādīkās (the i being lengthened for the sake of the metre) would be equivalent to brāhmy-ādikās, referring to the mother goddesses. This appears to be the earliest known epigraphic occurrence (and possibly the earliest of all textual attestations dateable with any confidence) of the name Brāhmī as the first of the mātṛs. Correspondingly, the first of the goddesses sculpted here is probably indeed Brāhmī, holding what appears to be a lotus flower. Although Joshi (1986, 86) identifies her as Lakṣmī on the basis of this lotus, Lakṣmī is not normally a member of sets of mātṛs. Harper (1989, 79), Schastok (1985, 67) and Meister (1986, 239) all concur on Brāhmī, though they differ in some of their identifications of the other goddesses in the Badoh-Pathari group. I understand sva-mata-gati to be synonymous with the attested compound mano-gati, meaning that the Mothers go about as they please or act as they will.
The following words appear to be yaś caraty ugra. Since the area above ya is damaged, it is possible that yāś was originally engraved. In order to link this phrase to what has been read so far, I prefer this latter reading. I must note, however, that caranty is impossible: the character tyu is damaged in some places, but is definitely not ntyu. Caraṃty is also unlikely, though it is not impossible that there was an original anusvāra higher above ra than other anusvāras in the inscription. My interpretation thus requires emending caraty to caraṃty (or, more invasively, caranty). This still seems more plausible than a nominal sentence consisting only of brāhmy-ādīkās sva-mata-gatayo and a new sentence starting thereafter with a masculine subject (yaś). Yet another alternative would be to emend yāś to yā, but this also results in a new sentence starting here (with a feminine singular subject), and would be an even more invasive emendation since it requires the removal of a ligature component. The two remaining characters that complete this pāda of the stanza are obliterated by damage; the first appears narrow and the second boxy in shape, with traces of a stroke along the bottom and upward from the bottom right corner. In view of the context and the vestiges, rūpā, roṣā or kopā seem most plausible; tāpā or tāpaṃ is also possible, and something quite different cannot be excluded. The second part of the quarter, provisionally reconstructed as yāś caraṃty ugra-**, is thus a subordinate clause describing the Mothers as going about, fierce perhaps in shape, anger or asceticism.
The second pāda seems to read rudra-kṣetre saha vicarituṃ yā prahṛṣṭa-praṇītā. Although nearly all the akṣaras are damaged to some extent, only a few of them are ambiguous in their context. The characters saha are quite small and awkwardly shaped, but nonetheless likely. Of the following vi only an outline remains; the inside has flaked off, but the character is unlikely to be anything other than vi. Next, ca in fact looks like ci or ce, but as these make no sense here (and the latter is unmetrical), I must assume that ca was engraved and what looks like a vowel mark above it is just damage, or that ci was inscribed erroneously for ca. In yā the putative ā mātrā is lost to damage; to make the text conform to the first pāda, I assume that there was an original ā here. There may have been a small upadhmānīya attached to pra, so the pronoun could be yā or yāḥ, and even yaḥ cannot be excluded. The last two characters of the quarter (which are also the last in the physical line) are very badly weathered and only visible as vague shadows. In the first, a circular vowel mark is certainly visible, and since a prosodically long syllable is expected and the last character is not a ligature, this must be ī rather than i. The consonant seems to have the winged top of ṇ, but as far as the body shape is concerned, a number of other readings are also possible (especially t and v; possibly s with the vertical stroke that I see as belonging to the last character). The form of the final character suggests t (though g or ś may be possible), and there may also have been a visarga (or one may have been intended but omitted) at the very end. No vowel mark is discernible in the last akṣara, so I tentatively restore ā to obtain a feminine ending. As far as the stone is concerned, a, e or o may also be possible, and a circular depression above the body may even suggest i or ī.
The probable meaning of this quarter is that the Mothers are, or can be, summoned to roam together joyfully in a field of Rudra. Rudra-kṣetra is widely attested in Śaiva texts including the Early Skandapurāṇa33 in the sense of a site sacred to Śiva. The reading praṇītā(ḥ) is, as noted above, very uncertain. The verb pra-√nī has a wide gamut of meanings ultimately going back to the basic sense “lead forward;” it seems acceptable to understand it as “summon,” but this is not one of its common meanings. In compound with prahṛṣṭa, the most straightforward meaning would be that the person who summons them is joyful, but with some laxity the intended meaning may have been that the Mothers themselves are happy to be summoned. Possible alternatives to praṇītā(ḥ) include praṇītiḥ (their conduct is joyful?) and pratītiḥ (their appearance is joyful?), but these singular nominatives do not fit into the syntax.34 All in all, prahṛṣṭa-praṇītā(ḥ) seems to be the least problematic reading.
In the third pāda, the phrase na vicaritaṃ jñāyate jñāna-vidbhiḥ is unequivocal in spite of some damage. The first na in this string is part of a ligature which is in all probability nna. The four characters preceding that are in a worse condition. The two that precede nna appear to be tatva; at any rate, the first of these looks like ta or ga, and the second is a ligature with v as the subscript component and probably t or n as the upper component. No discernible vowel mark is attached to either; the first probably had none, while the second may have had ā. At the very beginning of line 4 the stone is badly eroded. There seems to be a ghost of the curled left limb and split bottom of y, with a slanted stroke above it to the right which must have been part of an ā or o mātrā. All that can be said confidently of the second character is that it does not extend below the baseline, as there is a small stretch of pristine stone surface below it. It appears that this character involved vertical or nearly vertical strokes at the left and right and did not have a horizontal bottom stroke; it probably also had an ā mātrā attached to it. The most likely reading of these two characters is yāsāṃ, continuing the series of relative pronouns begun in the first two pādas. However, tāsāṃ cannot be excluded (though this would make the line begin with a slight indent), and some other word may also be possible, such as yoge, yogī or yo sau (though these would only work if I am wrong about parts of the context).
The expression na vicaritaṃ is the locus where the metre is deficient: prosodically long taṃ should be preceded by five, not four short syllables. Since na is part of a ligature, the syllable preceding that is necessarily prosodically long even if it does not include an ā mātrā, so the possibility that there are three characters at the beginning of the physical line where I tentatively read yāsāṃ can be ruled out. An akṣara was therefore omitted from the inscription. Assuming the omission to be the engraver’s fault, the composer’s intent may have been na hi, na ca or na tu.
Ultimately, I propose to restore the third pāda as yāsāṃ tatvan na ca vicaritaṃ jñāyate jñāna-vidbhiḥ, meaning that even sages (jñāna-vid, those who are conversant with [spiritual] knowledge) do not fully know who these goddesses really are and what they might do. This seems to be very appropriate in both the wider setting where the mātṛs are in transition from a vague band of potentially hostile folk deities to a circumscribed cabal of stately royal patronesses, and in the narrower context where they have just been described as capricious and equally capable of ferocity and felicity. The repeated use of (vi-)√car would emphasise this connection to the first two pādas. Nonetheless some caution is advisable in accepting this interpretation, especially because nowhere else in the legible parts of the inscription do we find a nasal conjunct where an anusvāra would be expected.35
The final quarter of the stanza is badly marred throughout. The first four akṣaras seem to be kālo deśo hi, though only śo is reasonably clear. The hook of hi may be nothing more than damage, in which case this character is pi. Each of the remaining characters in this sequence could be read in a number of alternative ways, but I have not found another meaningful combination. The next two characters are obliterated by damage. The subsequent word may be niyata, where i is certain and ta is very likely but everything else is fluid. A prosodically long syllable is expected where I read ta, and there may be an anusvāra above this akṣara. I am, however, more inclined to believe that the pit visible at that spot is damage, and the character was to with a small vowel mark consumed by flaking around the head. Only elusive traces remain of the rest of the quarter, except for a multiply bent stroke a few characters onward that I can only identify as the right-hand side of ha, though with a slightly awkward shape and a much wider hook than the one presumed in hi earlier. Some vestigial strokes are still in evidence before this partial character, but practically nothing survives between it and the beginning of the next stanza. The possibilities are too numerous to review here, so I shall only indicate my best guess for the contents of the entire quarter, which may have run something like kālo deśo hi bhava-niyato yo paro hetur attra.
This syntactically awkward reconstruction is probably wrong in several places, but if it is anything close to the original, then the gist may have been that Śiva (Bhava) is the ultimate cause on whom time and place rest, with “time and place” implying the circumstances in which the Mothers become well-disposed or hostile. Moreover, the Early Skandapurāṇa mentions Śiva by the name Hetukeśvara in association with the seven Mothers in Koṭīvarṣa (in North Bengal).36 I must emphasise that both bhava and hetu are conjectural to begin with, the latter being implied only by the preceding word paro (itself uncertain) and by a vestigial character that may be h. Nonetheless, the possibility of a connection is intriguing enough to deserve further investigation, and the corresponding restoration yo paro hetukeśaḥ cannot be ruled out on the basis of the stone. The area with the scanty vestiges that might belong to ttra could instead be a remnant of ke (in which case the partially preserved curve at the bottom belongs to a headmark in the next line, rather than being part of a subscript r), and the effaced area to the right of this spot may well obscure śa or even śaḥ. It may also be possible to restore yo paro hetur āsāṃ (for āsām), meaning that Śiva is the ultimate cause of the existence of the mothers, which would likewise be in line with the extant Skandapurāṇa text.37
Finally, though I believe that my above suggestions render the fragments reasonably coherent, there remains the possibility that the disparate behaviours described in the first two pādas do not apply to all of the Mothers at different times. In the first quarter the verb appears to be singular though the pronoun does not; in the second, the endings of both the pronoun and the verbal participle are unclear. It may be that a single goddess is described as ugra, in which case probably Cāmuṇḍā is meant. The prahṛṣṭa goddess could likewise be single, perhaps identifiable as Māheśvarī since saha applied to just one Mother would imply rudreṇa saha.
The fourth stanza is even more extensively damaged than the previous one. Its end, however, is clearly legible in line 6, showing that the verse is dedicated to the Mothers. Although its metre, sragdharā, is different from that of the preceding verse, the two may be semantically connected. As surmised about much of verse 3, the present one also includes relative clauses describing the goddesses, and the relative pronouns of these clauses (possibly along with those in verse 3) are picked up by tās in the fourth quarter.
At the beginning of this verse, the characters yāsā are clear. The next two characters are lost in a pit of damage. The consonant of first of these must have been m to complete yāsām (as there is no anusvāra over sā), and the shape of the pit permits this reading. The vowel of the second indistinct character seems to have been ā or o and, as the area below is also pitted, its consonant may have been a conjunct or had descending strokes. The word after yāsām thus began with a vowel, most likely a, ā or u, as intact surfaces around the presumed m exclude i, ī and ū and make e, ai, o, au and ṛ very unlikely. This was followed by two or more consonants (though possibly just one if the initial vowel was long), and probably ended in ā or o. In the next word, the first character is very probably ba, as most of a box shape (except for the bottom left corner) is discernible here. Skipping the indistinct second character for the moment, the consonant of the third is definitely gh, and this is followed by the almost clear conjunct rvva. This implies that the akṣara with gh must be read as ghai, and what we have here is the plural instrumental of a noun ending in gha. Now that we have some phonetic context, the indistinct character can be identified as lau with fair certainty. The triple vowel mark is quite discernible, with a pair of small strokes to each side and a much larger stroke going up and curving to the left (resembling the tail of the regular la); there is also a faint trace of what may be the hook of l at the bottom left. The third word of the pāda thus appears to be balaughair.
This seems to be followed by the sequence valabhivihara, the letters increasingly uncertain as we approach the end of the line. Va (in the ligature rvva described above) and la are quite clear, and bhi is damaged but unambiguous. The next character is a blotch of damage with only the i mātrā discernible; all that can be said of the body is that it had no components below the baseline. The fifth syllable is blurred by erosion but a bent left edge and a straight bottom are discernible, and there is very likely a trace of the inverted U-shaped end of ha. This may, however, be random damage, in which case the character may be ca, va or pa. This akṣara had no vowel other than the inherent a, since the stone below it is intact; what looks at first sight like an o mātrā at the top is in fact the subscript r of pra above. The sixth character of this sequence is narrow and predominantly vertical. Most likely it was ra, though ka may be possible (especially if the fifth akṣara is narrower than ha). There must have been one more character in this physical line, but only the faintest trace of it remains, which defies identification. In view of the characters that probably precede it, ṇ is a likely guess for its consonant component. I propose to read valabhi-viharaṇe here, but this is informed by my understanding of the context and, if I err elsewhere in the quarter, other readings are possible. In particular, valabhi-vicaraṇe seems plausible. It is also possible that the word valabhid (rather than valabhi) is involved, for instance as valabhid iva raṇe. Finally, the completely obliterated last character may be something entirely different than ṇ.
The initial two characters in line 5 are barely discernible. The first of them may be ṣ, possibly with ū or v attached below it, though there may also be an i or ī mātrā above the body. The shadow of the second resembles d or v but may be something else and may have a subscript component. It is thus possible that the last word of the previous line is a plural rather than singular locative,38 for instance if reading viharaṇeṣūḍha or viharaṇeṣv eva. Plodding on, the five characters following the first two indistinct ones seem to read gandharvvavataḥ, generally clearer toward the end. Ga is damaged but recognisable, though without context it could perhaps be śa. Ndha is only a blurred outline with the inside flaked off, yet the identification is quite certain. In rvva, only the subscript v is clear; the repha is obliterated and the upper consonant could be n or d on its own, but given the preceding syllables, rvva can be restored with reasonable confidence. The following va is clear, but a prosodically long syllable is expected here. A very small vowel mark may have been subsumed in the abrasion around the headmark, but this does not seem likely. The last character is quite clear and definitely looks like ta. The end of this string may, therefore, have been vātaḥ or, assuming a malformed character, vegaḥ.
The above speculations, however, do not yield a coherent sentence, and since the next quarter of the verse includes a second (and clearly legible) yāsāṃ, the present quarter must be a complete subordinate sentence on its own. For these reasons I prefer conjectural readings that do not present themselves immediately upon studying the engraving. To wit, for the characters that look like vataḥ, I propose varggaḥ. As noted above, the second character could be a misshapen g. There is a spot of damage below it that may conceal a second, subscript g. Admittedly, there is no repha in evidence; the vertical line above the character is clearly not in contact with the headmark and so it is probably a scratch in the stone, along with two more similar verticals at the same height, to the right of this one. I thus resort to emendation, assuming that the engraver omitted the repha and distorted the shape of the conjunct consonant. To continue with conjectures, I read the first two akṣaras in line 5 as siddha. A slightly distorted s with i is easily permitted by the traces remaining of the first; in the second there is no clear evidence of a subscript dh, but some grooves below the main body may be vestiges of such a component. Finally and by dint of mere guesswork, I propose to restore the illegible bit near the beginning of the quarter as ārtto. The reconstructed pāda would thus be yāsām ārtto balaughair vvalabhi-viharaṇe siddha-gandharvva-varggaḥ. The teeming hosts (balaugha) of the goddesses presumably refer to the throngs of potentially malevolent minor mātṛs and may include male semi-divine beings such as the dangerous grahas and kumārakas.39 The text thus obtained says that these mobs (presumably as they surge across the heavens) disturb siddhas and gandharvas who hang around peacefully in the penthouses of their palaces, which makes for a credible image.40
Of the second quarter of the fourth verse, only the beginning remains. The words vāyur yyāsāṃ javena druta are confidently legible. Vā is damaged, but the triangular bottom makes a strong case for v. There is no vowel mark in evidence, but the metre requires a long syllable, so there must have been either a very small ā mātrā to the right of the headmark, or a fairly small curved one above it. There is also rather heavy damage in dru; outside its context the character could perhaps be śru or jña, but the alternatives do not produce coherent readings. After druta the pitting of the surface becomes more severe, but from the vestiges I provisionally deduce ratha-tarasā. Of these akṣaras, ta is almost distinguishable, but the rest are badly obscured and very uncertain. It may be possible to read druta iva instead of druta-ratha (provided that the vertical form of the sign for independent i is used, which would be unusual in a script with so many southern characteristics), and other alternatives could be conceived of. The second quarter thus refers to a gale whipped up by the mothers’ passing; or, construing the genitive yāsāṃ with vāyur, a gale belonging to the mothers, possibly generated by them magically.41
Throughout the remainder of the fifth line of the inscription, only faint traces of headmarks and a few minuscule fragments of characters remain. The middle of verse 4, including the whole of its third quarter, is thus irrecoverable. Its conclusion at the beginning of line 6 is, however, clearly legible: mātaras tās satata-sumanaso mātṛvan mām avantu. The initial mā is weathered but its traces are clear; the n of the conjunct nmā is drawn like t, and there are a few spots of damage that do not impede reading. The verse thus ends with a prayer asking for the protection of the Mothers described above,42 with tās referring back to the instances of yāsām in this verse and, if I am correct to surmise that the first word of verse 3c is yāsāṃ, the previous one as well. The three long syllables lost at the beginning of this last quarter (the end of line 5) might be speculatively restored as lokānāṃ, prompted by what may be the hooked ascender of l about three characters back from the end. There are, however, no discernible traces here, so alternatives including (but not limited to) sa-skandā or viśveṣāṃ remain possible.
After the versified preamble comes an executive section probably written in prose where, in addition to the distributed flaking of the surface, progressively longer stretches of line ends are completely weathered away. There seems to be a punctuation mark in the shape of a short horizontal line separating the verses from the prose.
In line 6 after the last stanza the text dharmma-vijayino mahārā is quite certain; the ja that must have followed this string is obliterated. In the legible text, the character dha is very narrow; vi is badly damaged but the bottom left corner of v is clear and the traces above the body are suggestive of an i mātrā. Ja is all but effaced, but three horizontal lines can be made out amidst the damage. No is weathered but recognisable. The characters mahārā are faintly visible as a set of deeper grooves within a shallow pit.
Since the name of mahārāja Jayatsena comes quite a while later, this section presumably referred to the overlord of the ruler who issued the inscription. Unfortunately, there is hardly a spot in this area where the surface of the stone is not eroded. Nonetheless, I believe that the characters śrī-kumāragupta-rājye can be made out from faint traces. The reading is highly tentative and partly based on deduction. The site was probably inaugurated in the first half of the fifth century as indicated by the similarity of the sculpture to that of Udayagiri, and the palaeography permits this dating. The sovereign to whom the local ruler owed fealty would most likely have been either a Gupta or a Vākāṭaka king. Kumāragupta (r. ca. 415–447 CE) fits the age bracket, though the title mahārāja seems rather modest for him. The Vākāṭakas did use this unassuming title as a rule, but none of the Vākāṭaka monarchs of the period have names that match the traces in the stone. On the weathered surface the conjunct śr is quite recognisable, though the ī mātrā is not. Next to it, the headmark and arms of k are faintly perceptible and a curving line below the right arm may be the mark for u. Mā is largely based on wishful thinking, but much of a short ra is discernible. The spot where gu should be has no recognisable traces apart from the headmark, but the vestiges after this spot indicate p and a probable a subscript consonant. Next, there are traces of a very narrow character with a headmark, followed by a possible j and a very likely subscript y. With all these considerations in mind, the reading kumāragupta-rājye seems feasible. An interesting corollary of this hypothesis is that if it is correct, then the second verse of the inscription can, after all, be understood in praise of the ruler in a secondary sense. As noted above, several inscriptions commence with homage to a god, followed immediately by the laudation of a local or sovereign ruler and employing terminology similar to that found in the present text. The second stanza’s obeisance to Skanda/Kumāra may thus be bitextual, referring to Kumāragupta’s rule on a secondary level.
As correctly read earlier,43 the prose section includes a partial date and another reference to the Mother Goddesses: śukla-divase trayodaśyāṃ bhagavatyo mātaraḥ. The year would have been written near the end of the sixth line and is irretrievably lost. At the beginning of line 7, however, there is a weathered akṣara that, standing immediately before śukla, must have been the end of the name of the month. I read that character as ḍha. Other readings may be possible, but it is reasonable to assume that this character must have been either the last akṣara of a standard month name, or sa (for māsa), or the genitive ending sya. None of these are likely, though da (for Bhādrapada) and kha (for Vaiśākha) cannot be ruled out entirely. However, for either of these one would need to assume that the lines to the right of the character are damage and that there was a wide space between this character and the next. I thus identify the month as Āṣāḍha, which incidentally is one of the three months when the ritual of the Skandayāga needs to be performed.44 After the word mātaraḥ, the characters pratiṣṭh- can be made out with reasonable certainty. The reading does not seem to be pratiṣṭhāpitāḥ (unless the spelling is pratiṣṭhāvitā, which may be possible), but the purport must nonetheless be that the Saptamātṛs were installed on the recorded day. The next two characters appear to be tosa and may be a remnant of tāsāṃ, tāsu, or perhaps tās followed by an obliterated subscript consonant. No intelligible text can be made out in the remainder of this line.
The beginning of line 8 seems to run as follows: kṛtsnāvamukta-viṣayeśvarasya mahārāja-jayatsenasya satputreṇa. At the beginning of the line, kṛ is definite. The next character is a mess of damage but seems to have been a complex conjunct and the faint traces among the pitting indicate tsn, though it is nonetheless possible that the correct reading is something else, for instance involving kṛṣṇā° as part of the name of the province. The following characters va and mu are visible as unambiguous outlines, and kta is entirely clear. Garde too had read the name of viṣayeśvara mahārāja Jayatsena, who is not known from any other source. It has been suggested45 that the Jayatsena mentioned in this inscription is identical to the dynastic progenitor mentioned in the Nandsa yūpa inscription (ca. 226 CE). But as the author of the suggestion himself notes, this would require that Jayatsena be a very remote ancestor of the fifth-century Badoh-Pathari inscription. Unlikely to begin with, this possibility can be dismissed altogether with my reading satputreṇa. Garde’s observation (1926, 12) that the Badoh-Pathari inscription may belong to a descendant of Jayatsena rather than Jayatsena himself was nonetheless on the mark; on the basis of my reading it is safe to say that Jayatsena was the father of the king who issued the inscription. I have not been able to derive an intelligible reading from the damaged characters following satputreṇa. If the son was named here rather than later on in the obliterated section, then he probably did not have a name ending in sena, as the reading senena does not match any of the traces.
The Allahabad inscription of Samudragupta mentions a land called Avamukta among the southern kingdoms, ruled by a king named Nīlarāja, whom Samudragupta overpowered and released.46 The location of Avamukta has not been identified satisfactorily,47 but since the Allahabad pillar mentions it between Kāñcī and Veṅgī, it ought to be in the deep south. Nonetheless, the name is clear enough in the inscription, so it appears that Avamukta viṣaya at least included some lands north of the Vindhyas. Another conclusion is that the Badoh area was, at least at this time, not thought of as a part of the Daśārṇa region (the centre of which was Vidiśā). The word kṛtsna may imply that an extensive territory bore this name. Jayatsena and his nameless son may have been descendants of Nīlarāja: Samudragupta’s “capture and release” (grahaṇa-mokṣa) has been understood as the practical implementation of the concept of dharma-vijaya,48 and our inscription refers to Kumāragupta as a dharma-vijayin. Jayatsena and his son may have continued to pay lip service to Samudragupta’s descendant, or Kumāragupta may have renewed the nominal conquest of their land of Avamukta.49
The beginning of the ninth line reads śīlena daivata-dvijāti-dharmma-guru-bhaktena.50 Some of the individual characters are damaged to a varying degree (dha is completely gone), but the reading as a whole is beyond doubt.51 This must have been part of a long description of Jayatsena’s son. The word śīlena is obviously the end of a lost compound and would have been preceded by some sort of laudable personality trait.
From the vestiges below line 9 it is evident that there was once a tenth line to the epigraph, probably extending all the way to the right margin. Nothing in this line is legible. Judged from the size of the recessed panel that bears the inscription, there would not have been an eleventh line.
4.1 Editorial Notation
Where permitted by vowel saṃdhi, independent words are separated by spaces, and words in compound by hyphens.
halanta consonant form
verse number to mark beginning of stanzas
damaged or unusually formed characters confidently readable in the context
heavily damaged characters tentatively read
text lost to damage and confidently restored on the basis of context
text lost to damage and tentatively restored
lacuna in metrical verse with indication of the prosody of the lost text
lacuna with indication of the exact number of illegible characters
lacuna with indication of the approximate number of illegible characters
characters added by editor (in the curated text only), presumed to be omitted by the scribe or supplied for the sake of standardisation (avagraha)
conjectural translation of tentatively read or tentatively restored text
4.2 Diplomatic Transcript
(s)iddhaM ⟨1⟩p(r)āvṛtya yaḥ (sa)-rudhiraṃ gaja-carmma r(au)dra[ṃ] pr[e]tālaye niśi mahoraga-k[o](ṣṭha)-(?sūt)(ra)[ḥ] (?car)[?yyāṃ ka](r)[?o](?ti śiva)-(mā)tṛ-gaṇ(?ānuyāt)[?o] (r)[ud](r)[o jaya] (ty ama)ra-vand(i)ta-pāda-yugmaḥ ⟨2⟩tasy(ā)nu bha(kta-ja)na-nitya-hito nukaṃpī (v)ī(r)(?ār)[?bhak](o) ja(yat)i (?mātṛ)-ga(ṇ)(?ais sa)(nā)(?tha)ḥ ya(?ḫ) (pūjyate sata)tam (?o) [⏑⏑–⏑–⏓] (r) (ijyāṃjali)-praṇava-(baly-u)pahāra-yogaiḥ ⟨3⟩(br)[ā](hmyā)(?dīkās s)(va)-(?mata-gatayo) y(ā)ś ca(ra)t(y u)g[r](a)-[–⏓] (rudra)-kṣ(et)r(e) sah(a) (?vi)(ca)ritu(ṃ yā)(?ḫ) (prah)ṛ(ṣṭa-pra)(?ṇītāḥ) (?yāsāṃ ta)(tv)(?a)(n na) vicarit(a)ṃ jñāyate jñāna-(v)i(d)bhi(ḥ) (?k)ā(?lo d)eśo (h/p)i [⏑⏑] (?n)i (?ya)(t)(?o) (?yo paro he)[⏑–⏓] ⟨4⟩yā(s)ā(?m) [?ārtt](?o balau)(ghair vva)la(bhi)(?vihara)[?ṇe] (?siddha)-(gandharvva)-va(g)(?g)aḥ (?vā)yur yy(ā) sāṃ jave(na druta)(?-ratha-tarasā) [–⏑––⏑–⏓ ––––⏑––⏑⏑⏑⏑⏑⏑––⏑––⏑–⏓ –––] mātaras tās satata-sumanaso mātṛva(n) mā(m a)vantu(|) (dharmma-vi)[ja](yino mahārā)[ja]-(śr)[ī]-(?ku)[mā](?ra)[?gupta]-(?rājye) [?20] [?āṣā] (?ḍha)-śukla-divase trayodaśyāṃ bhagavatyo mātaraḥ (pratiṣṭh)(?ā)[?2](?tosa) [?1] (?sā) [?2] (?rava) [?1] (?ṣyasya) [?28] (kṛ)(?tsnā)(vam)ukta-viṣayeśvarasya mahārāja-ja(yatsenas)ya sa(tputreṇa sa)  (?na)s(?sva)(jana)(?pura) [?30] (śīl)(?ena) (daivata-dvijāti)-[dha]r(mma)-guru-bhak[t]e(?na) [?45] […]
4.3 Curated Text
⟨Verse 1. Metre: vasantatilaka⟩
p(r)āvṛtya yaḥ (sa)-rudhiraṃ gaja-carmma r(au)dra[ṃ]pr[e]tālaye niśi mahoraga-k[o](ṣṭha)-(?sūt)(ra)[ḥ](?car)[?yyāṃ ka](r)[?o](?ti śiva)-(mā)tṛ-gaṇ(?ānuyāt)[?o](r)[ud](r)[o jaya](ty ama)ra-vand(i)ta-pāda-yugmaḥ
⟨Verse 2. Metre: vasantatilaka⟩
tasy(ā)nu bha(kta-ja)na-nitya-hito ⟨+’⟩nukaṃpī(v)ī(r)(?ār)[?bhak](o) ja(yat)i (?mātṛ)-ga(ṇ)(?ais sa)(nā)(?tha)ḥya(?ḫ) (pūjyate sata)tam (?o) [⏑⏑–⏑–⏓](r)(ijyāṃjali)-praṇava-(baly-u)pahāra-yogaiḥ
⟨Verse 3. Metre: mandākrāntā⟩
(br)[ā](hmyā)(?dīkās s)(va)-(?mata-gatayo) y(ā)ś ca(ra)⟨+ṃ⟩t(y u)g[r](a)-[–⏓](rudra)-kṣ(et)r(e) sah(a) (?vi)(ca)ritu(ṃ y)[?ā](?ḫ) (prah)ṛ(ṣṭa-pra)(?ṇīt)[?āḥ](?yāsāṃ ta)(tv)(?a)(n na) ⟨+ca⟩ vicarit(a)ṃ jñāyate jñāna-(v)i(d)bhi(ḥ)(?k)ā(?lo d)eśo (h/p)i [⏑⏑] (?n)i (?ya)(t)(?o) (?yo paro he)[⏑–⏓]
⟨Verse 4. Metre: sragdharā⟩
yā(s)ā(?m) [?ārtt](?o balau)(ghair vva)la(bhi)(?-vihara)[?ṇe] (?siddha)-(gandharvva)-va⟨+r⟩(g)(?g)aḥ(?vā)yur yy(ā) sāṃ jave(na druta)(?-ratha-tarasā) [–⏑––⏑–⏓][––––⏑––⏑⏑⏑⏑⏑⏑––⏑––⏑–⏓][–––]mātaras tās satata-sumanaso mātṛva(n) mā(m a)vantu(|)
(dharmma-vi)[ja](yino mahārā)[ja]-(śr)[ī]-(?ku)[mā](?ra)[?gupta]-(?rājye) [?20] [?āṣā](?ḍha)-śukla-divase trayodaśyāṃ bhagavatyo mātaraḥ (pratiṣṭh)(?ā)[?2](?tosa)[?1](?sā)[?2](?rava)[?1](?ṣyasya) [?28] (kṛ)(?tsnā)(vam)ukta-viṣayeśvarasya mahārāja-ja(yatsenas)ya sa(tputreṇa sa)  (?na)s(?sva)(janapura) [?30] (śīl)(?ena) (daivata-dvijāti)-[dha]r(mma)-guru-bhak[t]e(?na) [?45][…]
⟨verse 1⟩Victorious is Rudra, whose pair of feet is praised by the immortals, who wears a horrendous, bloody elephant skin and has a great serpent for a waistband as he roams at night in the funeral grounds, accompanied by the band of benevolent Mothers.⟨verse 2⟩After him, victorious is that valiant infant fostered by the hosts of Mothers, who is compassionate and ever well disposed to his devotees, and who is constantly worshipped … by means of sacrifice, homage, chanting, [animal] offerings and oblations.⟨verse 3⟩[The goddesses] beginning with Brāhmī, who go about in fierce [anger? shape? austerity?] acting as they please, who are thrilled to be summoned to the field of Rudra to roam together: their essential nature and behaviour are not known [even] by the experts of knowledge … time and place are set … that which is the supreme …⟨verse 4⟩The floods of whose hosts bother the company of siddhas andgandharvas in their dallying on rooftops,By whose surge the wind with the vehemence of a rushing chariot …… … …—may those Mothers cherish me ever kind-heartedly like mothers.[line 6]During the reign of His Majesty Kumāragupta, the great king who conquers ethically … … …[line 7]On the bright thirteenth day [of āṣāḍha?] the venerable Mothers were established … … …[line 8]by the true son of the great king Jayatsena who is the lord of the entire province of Avamukta … … …[line 9][by him] whose nature is … [and who is] devoted to gods, Brahmins, lawfulness and his elders …
This article is an outcome of research and fieldwork carried out in the Asia Beyond Boundaries project (
(Fleet and Bhandarkar 1981)
Where available, the Siddham ID s of inscriptions are given below in addition to their primary print editions. Siddham (
Agnipurāṇa: H.N. Apte (1900)
Allahabad pillar inscription of Samudragupta: Siddham IN00001; CII3 No. 1; CII3rev no. 1
Arthaśāstra ascribed to Kauṭilya: Kangle (1960)
Bagh charter of Bhuluṇḍa of the year 55: Siddham IN00509; Ramesh and Tewari (1990, 19–21)
Bihar Sharif first pillar inscription of the time of Skandagupta: Siddham IN00047; CII3 No. 12; CII3rev No. 41
Dhūrtakalpa see Skandayāga
Gangdhar inscription of Mayūrākṣaka (/of Viśvavarman): Siddham IN00076; CII3 No. 17; Balogh (2019)
Junagadh rock inscription of the time of Skandagupta: Siddham IN00032; CII3 No. 14; CII3rev No. 28
Mahābhārata digital edition Mahabharata Online http://gretil.sub.uni-goettingen.de/gretil/1_sanskr/2_epic/mbh/sas/mahabharata.htm ultimately based on the critical edition of the Bhandarkar Oriental Research Institute
Pathari inscription of the time of Jayasiṃhadeva II: Ramsharma (1969) and Trivedi (1978, 208–209)
Skandayāga edited by Goodwin (1893), re-edited by Bolling and Negelein (1909, 128–135)
Sunao Kala plates of Saṃgamasiṃha: Konow (1910)
Tumain inscription of Harideva (/of Kumāragupta): Siddham IN00024; CII3rev No. 20
Udayagiri Cave 6 inscription of the time of Candragupta II: Siddham IN00009; CII3 No. 2; CII3rev No. 7
Valkhā copper plates in general: Ramesh and Tewari (1990); see also Balogh and Kiss (forthcoming) for some problems with the chronology of these records
Apte, Hari Narayan. 1900. Śrīmad-Dvaipāyana-Muni-Praṇītam Agnipurāṇam. Anandashram Sanskrit Series 41. Poona: Anandashram Press.
Apte, Vaman Shivaram. 1957. The Practical Sanskrit-English Dictionary, Revised and Enlarged Edition. Edited by Parshuram Krishna Gode and Cintāmaṇa Gaṇeśa Karve. Poona: Prasad Prakashan.
ARIE 1961–62. Annual Report on Indian Epigraphy for 1961–62. Edited by H.K. Narasimhaswami. Calcutta: Government of India Press.
Bakker, Hans T. 2014. The World of the Skandapurāṇa: Northern India in the Sixth and Seventh Centuries. Groningen Oriental Studies Supplement. Leiden: Brill.
Bakker, Hans T., Peter C. Bisschop, and Yuko Yokochi, eds. 2014. The Skandapurāṇa. Vol. IIB Adhyāyas 31–52: The Vāhana and Naraka Cycles. Groningen Oriental Studies, Supplement 1566–7642. Leiden: Brill.
Balogh, Dániel. 2017. ‘The Abominable Yati—Or, An Intriguing Relic of the Prehistory of Sanskrit Verse’. In Puṣpikā: Tracing Ancient India, through Texts and Traditions. Contributions to Current Research in Indology. Volume 4, Proceedings of the Seventh International Indology Graduate Research Symposium (Leiden, 2015), edited by Lucas den Boer and Daniele Cuneo, 14–32. Oxford, Philadelphia: Oxbow Books.
Balogh, Dániel. 2018. ‘The Company of Men. Early Inscriptional Evidence for the Male Companions of Mother Goddesses’. In Goddess|Woman. Devi Cults and Traditional Female Roles in India, edited by Róbert Válóczi, 77–87. Budapest: Ferenc Hopp Museum of Asiatic Arts.
Balogh, Dániel. 2019 (in press). Inscriptions of the Aulikaras and Their Associates. Beyond Boundaries 5. Berlin, Boston: De Gruyter.
Balogh, Dániel, and Csaba Kiss. forthcoming. ‘Valkhā Chronology Revisited: More Plates and More Questions’. Bulletin of the School of Oriental and African Studies.
Bolling, George Melville, and Julius von Negelein. 1909. The Pariśiṣṭas of the Atharvaveda. Volume I: Text and Critical Apparatus. Leipzig: Otto Harrassowitz.
Casile, Anne. 2007. ‘The Archaeological Remains of Rāmgarh Hill: A Report’. In The Temple in South Asia, 29–48. London: British Association for South Asian Studies.
Casile, Anne. 2014. ‘Changing Religious Landscapes in Gupta Times: Archaeological Evidence from the Area of Baḍoh-Paṭhāri in Central India’. South Asian Studies 30: 245–268.
Cecil, Elizabeth A., and Peter C. Bisschop. 2019. “Venerable Monuments and Landscapes of Memory in Early India.” History of Religions 58 (4): 355–403.
Chhabra, Bahadur Chand. 1948. ‘Bamhani Plates of Pandava King Bharatabala: Year 2’. Epigraphia Indica 27: 132–145.
Cunningham, Alexander. 1880. Report of Tours in Bundelkhand and Malwa in 1874–75 and 1876–77. Archaeological Survey of India Reports 10. Calcutta: Superintendent of Government Printing.
Cunningham, J.D. 1848. ‘On the Ruins at Putharee’. Journal of the Asiatic Society of Bengal 17: 305–312.
Dvivedi, Harihar Niwas. 1947. ग्वालियर राज्य के अभिलेख. Gwalior: Madhya Bharat Puratattva Vibhag.
Fleet, John Faithfull. 1888. Inscriptions of the Early Gupta Kings and Their Successors. Corpus Inscriptionum Indicarum, III. Calcutta: Superintendent of Government Printing.
Fleet, John Faithfull, and Devadatta Ramakrishna Bhandarkar. 1981. Inscriptions of the Early Gupta Kings (Revised Edition). Edited by Bahadur Chand Chhabra and Govind Swamirao Gai. Corpus Inscriptionum Indicarum, III rev. New Delhi: Archaeological Survey of India.
Garde, M.B. 1926. Annual Report of the Archaeological Department, Gwalior State for Samvat 1982, Year 1925–26. Gwalior: Alijah Darbar Press.
Goodwin, Charles J. 1893. ‘The Skandayāga, Text and Translation’. Journal of the American Oriental Society 15: v–xiii.
Harle, James C. 1974. Gupta Sculpture: Indian Sculpture of the Fourth to the Sixth Centuries A.D. Oxford: Clarendon Press.
Harper, Katherine Anne. 1989. The Iconography of the Saptamatrikas: Seven Hindu Goddesses of Spiritual Transformation. Lewiston, NY: E. Mellen Press.
Hatley, Shaman. 2012. ‘From Mātṛ to Yoginī: Continuity and Transformation in the South Asian Cults of the Mother Goddesses’. In Transformations and Transfer of Tantra in Asia and Beyond, edited by István Keul, 99–129. Berlin: De Gruyter.
Joshi, Nilakanth Purushottam. 1986. Mātṛkās, Mothers in Kusạ̄nạ Art. New Delhi: Kanak Publications.
Kangle, R.P., ed. 1960. The Kautịlīya Arthaśāstra. Part I, A Critical Edition with Glossary. University of Bombay Studies, Sanskrit, Prakrit and Pali 1. Bombay: University of Bombay.
Konow, Sten. 1910. ‘Sunao Kala Plates of Samgamasimha; [Kalachuri-] Samvat 292’. Epigraphia Indica 10: 72–76.
Mann, Richard D. 2011. The Rise of Mahāsena: The Transformation of Skanda-Kārttikeya in North India from the Kuṣāṇa to Gupta Empires. Brill’s Indological Library 39. Leiden: Brill.
Meister, Michael W. 1986. ‘Regional Variations in Mātṛkā Conventions’. Artibus Asiae 47 (3/4): 233–262.
Modak, B.R. 1993. The Ancillary Literature of the Atharva-Veda. A Study with Special Reference to The Pariśiṣṭas. New Delhi: Rashtriya Veda Vidya Pratishthan.
Panikkar, Shivaji K. 1997. Saptamātṛkā Worship and Sculptures. Perspectives in Indian Art & Archaeology 3. New Delhi: D.K. Printworld.
Patil, D.R. 1952. Descriptive and Classified List of Archaeological Monuments in Madhya Bharat. Gwalior: Department of Archaeology, Madhya Bharat Government.
Ramesh, K.V., and S.P. Tewari. 1990. A Copper-Plate Hoard of the Gupta Period from Bagh, Madhya Pradesh. New Delhi: Archaeological Survey of India.
Ramsharma. 1969. ‘Two Paramara Inscriptions from Malwa’. Epigraphia Indica 38: 33–36.
Schastok, Sara L. 1985. The Śāmalājī Sculptures and 6th Century Art in Western India. Studies in South Asian Culture 11. Leiden: Brill.
Sharma, Tej Ram. 1978. Personal and Geographical Names in the Gupta Inscriptions. Delhi: Concept Publishing Company.
Tiwari, Jagdish Narain. 1985. Goddess Cults in Ancient India: With Special Reference to the First Seven Centuries A.D. Delhi: Sundeep Prakashan.
Trivedi, Harihar Vitthal. 1978. Inscriptions of the Paramāras, Chandēllas, Kachchapaghātas and Two Minor Dynasties. Corpus Inscriptionum Indicarum, VII/2. New Delhi: Archaeological Survey of India.
Venkataramayya, M. 1953. ‘The Ancient Mālavas: Some New Royal Names’. Indian Historical Quarterly 29: 80–83.
Willis, Michael. 2009. The Archaeology of Hindu Ritual: Temples and the Establishment of the Gods. New York: Cambridge University Press.
Yokochi, Yuko. 2013. “The Development of Śaivism in Koṭīvarsa, North Bengal, with Special Reference to the Koṭīvarsa-Māhātmya in the Skandapurāna.” Indo-Iranian Journal 56: 295–324.