The precise area and extent of the battlefields on the Bạch Ðằng River is still in debate. At the time of the 1288 invasion, the Bạch Ðằng River formed a lower part of the large Hong River (Red River) and was connected directly to the capital at Thang Long (Hanoi). Even today the Bạch Ðằng River is still relatively large and its estuary extends 5–6 km and is 2 km wide (Fig. 5). Among a number of questions regarding the battles are: how did Dai Viet troops create battle formations using stakes, and precisely where the battle took place. These questions have been carefully explored, bearing in mind that at least three major naval battles took place in the general area of the Bạch Ðằng River, as mentioned above (Taylor, 1983). Historical records indicate that the use of wooden stakes to trap ships had also been used in the same general area on two previous occasions (Ðại Việt Sử Ký Toàn Thư, 1993: 220–221). In AD 938, for example, Ngo Quyen defeated the Southern Han army in the same river. According to Ðại Việt Sử Ký Toàn Thư and Khâm định Việt sử Thông giám Cương mục (Imperially ordered annotated text reflecting the complete history of Viet) the AD 938 battle took place in the Nam Trieu area. This is also supported by stories related to shrines in honour of Vietnamese General Ngo Quyen located at Nam Hai, Dang Hai commune, in the An Hai district of Hai Phong city (Phan and Diệp, 1970: 66). Another battle occurred in AD 981, when Le Hoan, the founder of the Early Le Dynasty (AD 980–1009), fought with the Chinese Song Dynasty's army and fleet, but the precise location of this battle is not known. With regard to the battle in 1288, the battlefield may have extended to the upper part of the Bạch Ðằng River where many tributaries, including the Da Bac, Gia Duoc, Thai and Gia Rivers, lead into it. There is little information available about whether stake yards still exist or not at these places as no archaeological survey has been done. The first researchers to focus on the AD 1288 battle attempted to understand the landscape at the time of the naval battle (Trần, 1963; Dao, 1969; Phan and Diệp, 1970). Their studies included investigating changes to the topography, tracing ancient waterways and the history of dike construction, and resulted in the opinion that the Chanh River, which is a branch of the Bạch Ðằng River today, probably did not exist as a separate ‘river’ in 1288. Instead it was probably only visible as a waterway or channel when the tide was low. The description ‘Bạch Ðằng river mouth’ in historical records seems to indicate the vast area that includes the present Chanh River that divides Ha Nam Island from the mainland of Yen Hung district, several other channels including the Rut and Kenh Rivers, as well as the Bạch Ðằng River itself. Around the time of the battle, the area of current Ha Nam Island appears to have been a littoral zone, which probably consisted of large areas of tidal mudflats and only a few small dry areas of higher ground or islands. It had several waterways in its lower reaches, which meandered between underwater stone or sand mounds. These waterways later became the Chanh, Kenh, Rut and other rivers, probably after the settlement of Ha Nam Island began in the 15th century, and traces of these ancient channels can be seen in aerial photographs, old maps and charts.