• 4 The authors warmly thank Arlo Griffiths for his editing of the English-language.

1After 25 years of archaeological research on old settlement sites of North Sumatra Province, it seems appropriate to review what is now known of such sites and associated ones dated prior to the sixteenth century in this province and in the adjacent areas of West Sumatra and Riau provinces.4

• 5 We thank Véronique Degroot for pointing out other similar motifs on Pasai old tombstones, in partic (...)

10It may not be a mere coincidence that in 2009 we had already noticed the similarities in form and style of PNG2 with two tombs of Pasai documented in 2006, located at Dusun 44 near Kuta Kareueng, an old graveyard identified as Peut Pleuh Peut by C. Guillot and L. Kalus. Carved in granite showing black inclusions, their inscriptions unfortunately do not include names or dates (Guillot and Kalus 2008: 272–273).5

• 6 On these points, see Perret 2014.

13Still on the west coast of North Sumatra, in the southern sector of Sibolga Bay, we looked near Lumut (Kecamatan Badiri, Kabupaten Tapanuli Tengah) for the discovery site of the stone Gaṇeśa observed nearly twenty years ago by Lucas P. Koestoro (2001). The aim was to check whether the surroundings might subsequently have provided indications of an ancient settlement site. The discovery site (01°35’16” N, 98°49’57.5” E) is located on the slopes of a hill overlooking a small tributary of the Lumut River upstream from the village of Jago-Jago. Gold is being mined at the foot of the hill. The statue carved in tuff is still there, but in a fragmentary and very degraded state (height 66 cm, base 45×28 cm) (fig. 7). At the same place, fragments of another image (undetermined) carved in tuff are also visible. The thick vegetation covering the hill at this location prevents effective survey, but the former landowner is not aware of any chance discovery of shards or bricks made recently, either on the slopes or at the top of the hill. The presence here of these two apparently isolated statues remains unexplained. As early as the eighteenth century, explorers who wanted to venture into the hinterland from the west coast left from the Bay of Sibolga (also known as Bay of Tapanuli), particularly from the Lumut region. This was the case of Miller, Willer, van der Tuuk and Junghuhn, the latter then reaching Portibi in Padang Lawas, after following the Batang Toru, Batang Angkola and Sirumambe rivers. It is known also that the Lumut region was at that time already rich in camphor trees.6 In addition, the search for gold veins still active today was certainly practiced in the area in ancient times.

• 7 Muṇṭam means “head” or “forehead” in Tamil (https://0-dsal-uchicago-edu.catalogue.libraries.london.ac.uk/cgi-bin/app/tamil-lex_query.p (...)

14Further south, still on the west coast, the place name Batu Mundam/Mundom (Kecamatan Muara Batang Gadis, Kabupaten Mandailing Natal) caught our attention because the word mundam is still in use in the Tamil language,7 and this raises the question of the existence of a settlement frequented by Tamils in former times, as in Barus further north. The only significant feature in the landscape near the village is a steeply sloping hill called Bukit Peti (01°16’37.7” N, 98°51’42.2” E). This hill, no longer frequented, is at the centre of various beliefs among the locals (presence of a pond with perfumed water, an anchor, a carved house post, also a place of meditation in former times). The present village of Batu Mundam is said to have been founded five or six generations ago. According to the locals, the village’s founders originated from Kampung Sawangan, on the opposite bank of the Batang Toru River. During our brief visit, we didn’t hear of any find indicating the presence of an ancient settlement site in the vicinity.

• 8 Notulen van de Algemeene en Directievergaderingen van het Bataviaasch Genootschap van Kunsten en We (...)

15Further south, Singkuang (Kecamatan Muara Batang Gadis, Kabupaten Mandailing Natal) is located at the mouth (now silted up) of the Batang Gadis River. The toponym Picar Koling was still known near this mouth in the second half of the nineteenth century. Picar/Pijor Koling is a relatively common place name in North Sumatra, recorded from the west to the east coasts. It may refer to a gold smelting workshop run by people from the southeast coast of India. According to this interpretation, Pijor/Picar would be the modern Indonesian pijar-pijar, namely borax, a flux used to lower the melting point of gold, and Koling would be a variant of Keling, a term well known in maritime Southeast Asia to designate people from the Indian subcontinent. Finds of “antique objects” were reported there at the end of the nineteenth century.8 In addition, in his pseudo-historical work entitled Tuanku Rao, Parlindungan (c. 1964: 33–4, 38) probably made use of oral traditions to assert the presence of various foreign populations in Singkuang in ancient times. However, data collected during our survey do not go back beyond the colonial period (two Muslim tombs: Makam Syeh and Makam Raja – 01°03’40.6” N, 98°55’53.2” E), except perhaps the foundation of the Singkuang village itself by people of Tarutung village at about one hour by boat. The memory of the toponym Picar Koling seems definitively lost and no finds indicating presence of an ancient settlement site in the vicinity seem to have been reported recently.

• 9 For references about the points discussed below, see Perret 2014.

17The many gold-washers still in action in the torrential part of the Natal River have swapped sieves (dulang) for pumps that suck sediment from the river bed. The richness in gold of the Natal River basin has been known for a long time.9 It is already mentioned by Marsden at the end of the eighteenth century, gold making the reputation of the port of Natal itself. Moreover, the discovery of old gold ornaments and rings by gold panners was reported at the beginning of the twentieth century. Whereas in the middle of the nineteenth century, the Batang Natal was navigable nearly twenty kilometres upstream of its mouth for boats carrying cargoes up to three tons, this is no longer the case today: navigation is prevented by trees falling into the river and by the erosion of its banks. The hinterland of Natal was also known for its camphor tree forests. Willer thus noticed that in the first half of the nineteenth century, camphor trade in the region was monopolized by traders from Natal. And as for Singkuang, Parlindungan’s Tuanku Rao (c. 1964: 33–4, 38) claims that various foreign populations were present in Natal in former times. All these indications of course encouraged us to visit Natal.

• 10 See Soedewo (2014) for a review of the research on these remains.

21In the interior of this kabupaten of Mandailing-Natal, the Simangambat temple remains (01°02’31.1” N, 99°28’54.8” E, Desa Si Mangambat, Kecamatan Siabu), near Panyabungan, constitute the oldest ‘classical’ monument known to date in the northern half of Sumatra. Dated from the second half of the ninth century ce, a number of indications suggest that it probably served a Javanese community of Śiva worshipers. Spotted by a Westerner in the 1840s, it was cleared and excavated in 2009–2010 (fig. 9).10 Candi Simangambat is strategically located at the crossroads of two river systems, the Batang Angkola to the north and the Batang Gadis flowing from the Muara Sipongi region to the southeast, with a mouth into the Indian Ocean at Singkuang mentioned above. It is also near the Sorik Merapi volcano, at the top of which four small stūpas bearing Old Malay inscriptions have been found, one of which dates back to 1242 ce (Griffiths 2014: 233–235; Perret 2018: 261–263). The Sorik Merapi site also revealed three brick structures containing funeral urns. In addition, the Batang Natal River was formerly connected by a path to this volcano. It has been suggested in an earlier publication that the Javanese community linked to Simangambat could be at the origin of the building of the oldest monuments at Si Pamutung in Padang Lawas (Perret 2014: 327).

• 11 On these sculptures, see Soedewo 2014: 196, 210; Degroot 2014: 45; Perret 2014: 310, 318, 322.
• 12 Makara: height 83 cm, base 40×60 cm (the height of the figure in the mouth is 48 cm); the dvārapāla(...)

24All these elements suggest that Pasaman was home to a major settlement in ancient times, especially since Hindu-Buddhist remains have been found there since the 1960s. At Padang Nunang (Nagari Lubuk Layang, Kecamatan Rao Selatan) a stone dvārapāla and a stone makara (fig. 11) were found in the Sibinail/Simunair river that passes through the village, which explains their poor condition.11 Although their style is very similar to Padang Lawas sculptures, the material is different. These two images are made of sandstone while the dvārapālas and makaras of Padang Lawas are made of tuff. Both statues are now protected under a shelter (00°33’10.5” N, 100°03’11.0” E).12 Two months after our July survey a much better preserved makara was discovered in the same area (fig. 12, 13). Its style is very similar, if not identical, to the Padang Lawas makaras.

• 13 Point 1 (near the northern limit of the hill): 00°32’38.3» N, 100°04’17.4” E (depth 70 cm); point 2 (...)

27The most unusual site in the Rao region is undoubtedly Bukit Koto Rao (Jorong Tanjung Aia, Nagari Lubuk Layang, Kecamatan Rao Selatan), a long hill of several dozen hectares near the confluence of the Batang Sumpur, Sibinail and Batang Asik rivers. A succession of terraces, on which rows of brick and stone mounds can be seen (or at least made out), have been built around its periphery. The Balai Arkeologi Medan and the BPCB Batusangkar have already conducted limited excavations there, the latter having cleared in 2008 the remains of a brick structure measuring 3x3 m (Sri Sugiharta 2008). The hill is currently covered with a rubber plantation and generally thick vegetation, making it impossible to get a precise view of its topography and the density of mounds. A topographical survey would of course be essential, but it would require considerable clearing work, unless aerial survey techniques such as Lidar or drone can be used if not disturbed by the vegetation cover. During this survey, we carried out soil core sampling (between 37 cm and 86 cm deep) at ten points on the northern part of the hill to get an idea of the stratigraphy and density of archaeological finds.13 The more or less sandy humus, 30 cm thick on average, covers a yellowish brown or reddish brown sandy-clayey layer. Apart from a few brick fragments in soil core no. 6, these ten cores did not reveal traces of ancient occupation surfaces or artefacts. As this survey covered only a small part of the hill, it would be premature to rule out the presence of any old settlement site.

• 14 The boulder was found on the bank of the river (Setianingsih 2005).

30Further south, the inscribed tuff block of Ganggo Hilia (Jorong Pasar, Nagari Ganggo Hilia, Kecamatan Bonjol, Kabupaten Pasaman) is now protected under a shelter (00° 00’ 59.2” S, 100° 13’ 29.3” E) along the Batang Bubus (fig. 18). A tentative reading of the inscription was published in 2005,14 but an in-depth study remains to be done. This river does not seem to be a tributary of the Batang Sumpur, but flows towards the Indian Ocean.

• 15 Atmodjo et al. 1997; https://dapobud.kemdikbud.go.id/objek-benda/5bfc133b4abcfb04b4a6d55a/kompleks- (...)

31The most famous Hindu-Buddhist remains of the Kampar River Basin in Riau Province are those of Muara Takus (Desa Muara Takus, Kecamatan XIII Koto Kampar, Kabupaten Kampar), on the right bank of the river (Kampar Kanan). Revealed in 1860 by the mining engineer Cornelis de Groot, clearings of the monuments yielded four Buddhist inscriptions (three on stones and one on gold plate) paleographically dated between the eleventh and thirteenth centuries ce (Griffiths 2014: 236–7). A monumental complex of four structures, including the imposing Candi Mahligai (00° 20’ 09.2” N, 100° 38’ 31.8” E), has been restored between 1977 and 1994.15 The site is surrounded by an earthen wall, now partially submerged, along with several of the formerly surrounding villages, in the Koto Panjang reservoir, upstream from Bangkinang. Researches conducted since the 1970s have focused on the mapping of the earthworks and the excavation of remains of structures made of permanent materials inside and outside the walled area (Kusen et al. 1995a, 1995b). The Balai Arkeologi Medan recently cleared the remains of a brick structure on the periphery of the complex (see Soedewo et al. 2015).

• 16 Point 1 (near a concrete and steel tube fence erected in 2014): 00°20’16.3” N, 100°38’43.8” E (dept (...)
• 17 Point 10 (the nearest to the river): 00°20’22.5” N, 100°38’16.1 ”E (depth 102 cm); point 11: 00°20’ (...)
• 18 Point 13: 00°19’51.4” N, 100°38’34.2” E (depth 100 cm); point 14: 00°19’53.0” N, 100°38’34.2” E (de (...)

32The main purpose of our brief survey was to search for indications of settlement within the enclosed area. Using an auger, a first East-West series of nine cores (covering a distance of about 200 metres) were bored in the lower part of the site. With a depth varying between 22 cm and 78 cm,16 they reveal a topsoil layer of grey-brown humus about twenty centimetres thick above a yellowish clayey sediment. These nine soil cores yielded only a few fragments of charcoal in the humus layer. The only artifact collected was a fragment of brick in the humus layer at point 5, and it should be noted that boring at points 6 and 9 was blocked by stones and tuff blocks respectively, possible indications of remains made of permanent materials. Three soil cores were bored along the river (fig. 19). Their depth varied between 102 cm and 123 cm.17 Point 10 revealed a grey clayey sediment with red spots to a depth of 69 cm, then a grey sediment layer about 12 cm thick, above a grey clayey sediment with red spots down to a depth of 95 cm, covering a reddish brown sandy-clayey sediment down to 102 cm. Point 11 revealed the same compact grey clayey sediment with red spots down to a depth of 35 cm, then a layer of dark grey clayey-sand sediment to a depth of 50 cm, covering a reddish brown sandy-clay sediment down to a depth of 123 cm. Point 12 revealed a reddish brown sandy-clayey sediment to a depth of 33 cm, then a light reddish brown sandy-clayey sediment to a depth of 88 cm covering a light sandy-clayey sediment to a depth vering a distance of about 525 m) was carried out in the highest part of the site (fig. 20). With a depth varying between 52 cm and 100 cm,18 these soil cores show a more or less sandy grey-brown humus topsoil layer about 20 cm thick above a brown sandy-clayey sediment. These eight soil cores yielded only a few fragments of charcoal in the humus layer.

• 19 For a review of archaeological research in Kota Cina, see Perret et al. 2013. The publication of th (...)

36In the Medan region, the current capital of North Sumatra province, it is on the Kota Cina settlement site that almost all archaeological excavations in the region have focused since the 1970s.19 At that time, it was still possible to observe the remains of two earthworks at Deli Tua, on a hill overlooking the Deli River south of Medan. Pedestrian survey carried out at that time revealed earthenware and stoneware shards indicating an occupation that could date back to the fourteenth century (Milner et al. 1978: 29–30; Miksic 1979: 234). Of this site nothing more remains today than a segment of earthwork on the edge of the hill (03° 28’ 57.3” N, 98° 40’ 27.1” E). The hill is now a housing area (fig. 22).

• 20 Different from the Tanjung Anom (Tandam Hilir) site, about 16 km west of Kota Cina, reported by Mil (...)
• 21 See Harkantiningsih & Wibisono 2012; Edwards McKinnon et al. 2012.

37The only place reminiscent of a distant past remains Pancuran Gading, a regularly frequented sacred spring at the foot of the hill (03° 28’ 59.6” N, 98° 40’ 28.5” E) (fig. 23). Tanjung Anom (Kecamatan Pancur Batu, Kabupaten Deli Serdang),20 another old settlement site on the left bank of the Deli River, has recently been levelled, probably to become a housing area as well. Only a few shards of earthenware remain visible. Two soil cores were bored, including one (to a depth of 130 cm) in an area that had escaped the levelling operations (03° 31’ 00.3” N, 98° 36’ 05.8” E). It did not yield any ancient artefact. The old settlement site of Kota Rantang (Kawasan Mojopahit, Dusun 1, Desa Kota Rantang, Kecamatan Hamperan Perak, Kabupaten Deli-Serdang) was discovered in the late 1970s, and yielded Chinese, Thai and Vietnamese wares (Milner et al. 1978: 29). Excavations carried out thirty years later revealed artifacts dating the occupation between the twelfth/thirteenth and seventeenth centuries.21 Earthenware and stoneware shards are still clearly visible on the site today (03° 44’ 19.3” N, 98° 35’ 19.9” E), but the strong disturbances caused by agricultural activities (rice fields and oil palm trees) have certainly eliminated any possibility of stratigraphic excavation (fig. 24). On the last day of our survey in July, a visit to the small site museum at Kota Cina gave us the opportunity to observe a first assemblage of finds (earthenware, stoneware and glassware) coming from the recently rediscovered Bulu Cina site. If this site has not already suffered the fate of all the other old settlement sites around Medan, it should provide a significant contribution to the history of the region, or even the Strait of Malacca. Exploratory fieldwork has just been conducted there by the Balai Arkeologi Medan at the time this report is written.