Le Minh Khai's SEAsian History Blog (+ More)

The scholarship on early Southeast Asian history that relies on Chinese sources is a complete mess.

From the very beginning, scholars got a lot of things wrong, and then they just kept going and kept producing ever more flawed scholarship.

By this point, the extant scholarship is simply unusable. If you want to learn about early Southeast Asia, the only option at this point is to go to the primary sources, and in the case of Chinese sources, you can’t rely on translations for doing that. You have to go to the Chinese language texts.

As an example, let’s take a look at a 13th-century work called the Zhu fan zhi 諸蕃志 (Treatise on the Various Barbarians). This is an important text because it provides somewhat detailed information about places in Southeast Asia.

It was translated into English in 1911 by Friedrich Hirth and W. W. Rockhill (Chau Ju-Kua: His Work on the Chinese and Arab Trade in the Twelfth and Thirteenth Centuries, Entitled Chu-fan-chi), and there is a recent online translation by Shao-yun Yang (A Chinese Gazetteer of Foreign Lands).

In the image above, I have the table of contents from the above two translations, and then on the right, I have indicated where I think these translations are wrong. As you can see, there are quite a few places, but a couple of them are very significant: Sanfoqi 三佛齊 and Shepo 闍婆.

Sanfoqi has long been understood by scholars to refer to a place on the island of Sumatra, while Shepo has long been believed by many to be an early reference to the island of Java. Both of those claims are wrong. Sanfoqi referred to “Kambuja,” that is, Angkor, while Shepo referred to a maritime “empire” based at what is now Songkhla in southern Thailand.

In what follows, we’ll take Sanfoqi as an example, and will look at the problems one encounters when one reads the Zhu fan zhi while thinking that Sanfoqi indicates a polity on the island of Sumatra.

In 1876, Dutch Sinologist Willem Pieter Groeneveldt published some translations of passages about Southeast Asia in Chinese texts, and in this work, he argued that Sanfoqi was the Chinese name of a kingdom on Sumatra based at Palembang.

In so doing, Groeneveldt decided not to translate information that contradicted his ideas, such as an opening line in the passage on Sanfoqi in the History of the Song that states that it was “neighboring Champa.”

Hirth and Rockhill faithfully followed Groeneveldt’s idea and labeled the section in the Zhu fan zhi on Sanfoqi as “Palembang (Eastern Sumatra) San-fo-ts’i.” They also stated in a footnote that “All Chinese writers have identified [Sanfoqi] with Palembang, the north-eastern [sic] coast of Sumatra.” (63).

Ummm, well, the compilers of the History of the Song did not identify Sanfoqi with Palembang. . . but let’s stick to the topic here.

In the passage in the Zhu fan zhi on Sanfoqi, there is a list of places that were “vassals” of Sanfoqi. One of them is Balinfeng 巴林馮, a name which Hirth and Rockhill state in a footnote is Palembang.

Wait. Let’s stop and think about this. According to Hirth and Rockhill, Sanfoqi was Palembang. But Palembang is also in the list of Sanfoqi’s vassals.

So that means that Palembang was a vassal of itself? How can that possibly make any sense?

Of course, it doesn’t make sense, and that should have led Hirth and Rockhill (and many others) to stop and question how it could be that Sanfoqi was a kingdom on Sumatra at Palembang, but that’s not what happened.

So, by 1911, there were already some serious problems with the way that Western scholars were looking at Chinese sources for Southeast Asia. Groeneveldt had ignored and omitted information that contradicted his claim that Sanfoqi was a kingdom on Sumatra, and Firth and Rockhill had ignored contradictory information that emerged when they tried to view the Zhufan zhi from Groeneveldt’s flawed perspective.

Then in 1918, George Cœdès claimed that Sanfoqi was not only a Chinese name for a polity on the island of Sumatra based at Palembang, but that its real name was “Srivijaya,” and with that, it was full steam ahead, and a century later we are now the heirs of an enormous quantity of fatally-flawed scholarship.

The recent translation by Shao-yun Yang reflects a sincere effort to try to make sense of that century of flawed scholarship. Indeed, Yang makes an admirable effort to try to get the Zhu fan zhi to make sense based on what scholars have written over the past 100+ years, however, that task is ultimately impossible/futile because from the beginning scholars have been wrong.

So, for instance, Yang can see that Palembang was a vassal of Sanfoqi, or what he calls, following Cœdès, “Srivijaya,” and he understands that this is a contradiction. He resolves this contradiction by following an idea that was developed in the scholarship on Sanfoqi/Srivijaya. In particular, over the previous century, other scholars ran into problems trying to get sources to fit the idea that Sanfoqi was based at Palembang, and to resolve these problems they put forth the theory that the capital or center of power of Sanfoqi moved over time.

There is no evidence of the capital or center of power of Sanfoqi, regardless of where we think it was, ever moving. There is no text or inscription anywhere which says “the capital/center was moved” or anything even remotely close to that. Instead, scholars have simply resorted to this explanation when they have tried to view Sanfoqi as a kingdom on the island of Sumatra at Palembang and the information in texts then doesn’t make sense (and that happens a lot).

In other words, rather than recognizing that the sources do not support their view of the past, scholars have simply come up with an undocumented theory to force the sources to fit their preconceived view of the past.

This section in the Zhu fan zhi is one of those places where the information doesn’t make sense if you think that Sanfoqi was a kingdom based at Palembang, as Palembang is listed as one of its vassals. Yang then follows the approach of earlier scholars and states in a note that “By this time, the center of the Srivijaya mandala had most likely shifted from Palembang to Muaro Jambi.”

Ok, so that solves the problem of Palembang being a vassal of Sanfoqi because at this point, according to Yang and others, Sanfoqi was based at Jambi.

But wait, the Zhu fan zhi also seems to list Jambi as one of Sanfoqi’s vassals as well, as one of the vassals is a place called “Jianbi” 監篦, a name that certainly could be a transcription of Jambi. Hirth and Rockhill, however, claimed that this was a reference to “the modern Kampar on the E. coast of Sumatra” (72) while Yang has it as “possibly Kampar” but notes that the sailing times that are mentioned in reference to this place don’t make sense.

Here is what the Zhu fan zhi says about this Jianbi:

“The polity of Jianbi – Its kingdom is at a route opening. Many ships anchor here. From Sanfoqi one can sail here in half a month. It was formerly a vassal of Sanfoqi, but later, because of a war, it established its own king. . . It is a five-day journey by water to the polity of Lanwuli [Lamuri/Lambri].”

監篦國，其國當路口，舶船多泊此。從三佛齊國，風帆半月可到。舊屬三佛齊，後因爭戰，遂自立為王。土產白錫、象牙、真珠。國人好弓箭，殺人多者帶符標榜，互相誇詫。五日水路到藍無里國。

So, for anyone who thinks that Sanfoqi was “Srivijaya” and was located at either Palembang or Jambi, this entry about “Jianbi” makes no sense, because it is said to be 15 days from Sanfoqi/Srivijaya (i.e., Palembang/Jambi) and 5 days from Lamuri, at the northern end of Sumatra.

In other words, assuming that Jianbi was between Palembang/Jambi and Lamuri (and that’s what the text implies), this would mean that it would take around 20 days to go from Palembang or Jambi to the northern end of Sumatra. However, in the following century, we know that Ibn Battuta crossed the much longer distance from northern Sumatra to eastern Java in 21 days.

So, as Yang noted, something is wrong here.

What’s wrong here is that Sanfoqi was not a place on the island of Sumatra, and the contradictions that Hirth and Rockhill encountered in 1911 should have led people to realize that.

If we follow the information that Groeneveldt deleted from his 1876 translation, that Sanfoqi was “neighboring Champa,” and realize that this was a reference to “Kambuja,” then everything makes sense (see the image at the top of this page).

It took 15 days to go from Kambuja to Jambi; 5 days from Jambi to Lamuri; and the place where Jambi is located is like a place where a route opens, as the narrow passage through the Straits of Melaka widens and opens to the sea in the area off the coast of where Jambi is located.

Finally, in understanding this, we also don’t encounter any contradictions here when we see that Palembang and Jambi were both vassals of Kambuja. The information fits. There is no need to come up with an undocumented theory about a moving capital or center of power.

Having said that, this understanding of the past is one that is impossible to reach by reading the extant scholarship, because the extant scholarship (including translations) is a complete mess, and it has been a mess since day one. Yang does his best to try to make sense of that body of scholarship and he ends up with a lot of unresolved questions.

That is the inevitable outcome because the extant scholarship is flawed. One can’t resolve it by consulting it. One has to simply put it aside and start again by looking at the primary sources, understanding them, and building a new argument that those sources support. That is what I recently started to do in this article.