Abstract
San-fo-ch'i appears in Chinese source materials from the 10th to 15th centuries and has been almost unanimously identified with Srivijaya. It is, however, well known that Chinese records on San-fo-ch'i in the second half of the 11th century contain several explicit mistakes and other parts that seem incomprehensible. The accepted theory that the capital of Srivijaya moved from Palembang to Jambi around 1080 is based on interpretation of these documents.
After reexamination of all the related Chinese records, I reached a fundamentally different understanding on what San-fo-ch'i is in Chinese source materials.
Concerning Ti-hua-chia-lo who sent tributary mission to China in 1077, and who was recorded as ruler of Cola on one hand and as ruler of San-fo-ch'i on the other, my conclusion is, without excluding any information from source materials as mistaken, that he represented the Cola power in San-fo-ch'i and that as such he was the ruler of San-fo-ch'i at the same time. The dominance of Cola in the San-fo-ch'i area seems to have been established through a great expedition to the area mounted in 1025, which is recorded in the so-called Tanjore inscription. His residence seems to have been at Kedah. And the seemingly mysterious descriptions of San-fo-ch'i-Chan-pi (San-fo-ch'i-Jambi) as a tributary state in 1079 and 1082 and of San-fo-ch'i-Chou-lien (San-fo-ch'i-Cola) also as a tributary state in 1082 can be explained if, by analogy with Ta-shih, which is a general name for Arab or East Asian countries, San-fo-ch'i is understood as a general name for countries in a particular area.
Thus I conclude that San-fo-ch'i should be regarded not as a single polity or empire, like Srivijaya of the 7th and 8th century, which can safely be identified with Chinese Shih-li-fo-shih of Tang times, but as a general name for countries on the east coast of Sumatra and in the central and southern parts of the Malay Peninsula. San-fo-ch'i can be identified with Zabaj of Arab records, which includes Kalah (Kedah), Ramni (Aceh), Sribuza (Palembang?) and so on.
This conclusion raises new problems hitherto unconsidered, some of which are mentioned in the last part of this paper.
