The biggest story in premodern Southeast Asian history is the story of the centuries-long rivalry between Singora and Angkor.
Are you familiar with that story?
No, you are not, because you’ve never heard about it before. And the reason why you have never heard about it before is that for more than a century scholars have been misreading certain historical sources that tell this story, and have thus not been able to tell it.
In particular, there are a couple of place names that scholars have long misunderstood – Sanfoqi 三佛齊 and Java 闍婆 – and these misunderstandings have made it impossible for historians to clearly view the past and to see the story of the great rivalry between Singora and Angkor.
As I have explained in detail in previous posts, Sanfoqi was pronounced centuries ago something like “Kamfyutshai,” and referred to “Kambuja,” which from the ninth through the fourteenth centuries was represented by the empire of Angkor.
However, in an 1876 work entitled Notes on the Malay Archipelago and Malacca: Compiled from Chinese Sources, Dutch scholar W. P. Groeneveldt declared, without evidence or explanation, that Sanfoqi referred to the area of Palembang on the island of Sumatra, and from that point onward, scholars have (erroneously) assumed that this was the case.
This idea became further entrenched in 1918 when French scholar George Coedes examined an inscription from the area of Sumatra that contained the term “Srivijaya” and declared that this was the actual name of the supposed Palembang kingdom that Groeneveldt had claimed Chinese sources referred to as “Sanfoqi.”
That scholars have misunderstood this place name is a serious matter because when historians read historical sources, they use the information that they know in order to figure out other information. So for over a century, scholars have been thinking that they know what Sanfoqi refers to, and then they have been using that knowledge to interpret other information.
This has created a real mess because Sanfoqi does not refer to Palembang/Srivijaya, and so that means that a great deal of information that has been produced over the years based on this idea is wrong.
However, Sanfoqi is not the only term that scholars have misunderstood. There are others as well, such as the term “Java,” and that is a term that we will examine and decipher below.
When one realizes that Sanfoqi is Angkor, then it becomes possible to read Chinese historical sources and see the past from a new perspective. We can see that new perspective clearly, for instance, when we look at a work like Zhao Rukuo’s thirteenth-century Zhu fan zhi (Treatise on the Various Barbarians 諸蕃志).
The Zhu fan zhi has information about various foreign kingdoms. This work was translated into English in 1911 by scholars Friedrich Hirth and W. W. Rockhill and published under the title Chau Ju-kua: his work on the Chinese and Arab trade in the twelfth and thirteenth centuries, entitled Chu-fan-chï (St. Petersburg: Imperial Academy of Sciences).
Hirth and Rockhill followed Groeneveldt’s idea that Sanfoqi was Palembang. Without citing any evidence, they state in a footnote that “All Chinese writers have identified [Sanfoqi] with Palembang” (63). However, I have only come across one Chinese writer who equated Sanfoqi with Palembang, and that was Ming Dynasty voyager Ma Huan in his 1451 Yingya Shenglan 瀛涯勝覽 (The Overall Survey of the Ocean’s Shores), and I’ll explain in the next post why he did that.
In any case, Hirth and Rockhill used that information to help decipher other information in this text. They struggled at times to understand certain place names, and as the above image demonstrates they did not see a logic to the way that Zhao Rukuo presented the information in his text.
In Hirth and Rockhill’s understanding of the Zhu fan zhi, Zhao Rukuo jumped randomly from one geographic location to another in talking about various foreign kingdoms.
In the past, I likewise never saw a logic to the way the kingdoms in this work were presented, but in realizing that Sanfoqi = Angkor, I can now see that Zhao Rukuo organized the information in his text in an extremely logical manner.
Zhao Rukuo begins by mentioning kingdoms in mainland Southeast Asia, the area closest to China, and does so by moving from north to south, and then east to west.
So from north to south, we have Jiaozhi 交趾國 (Đại Việt), Champa 占城國, and Panduranga 賓瞳龍國.
And then moving westward from there we have Zhenla 真臘國, Tonle 登流眉國 (This is what I am calling this kingdom, as it is clearly based in the Tonle Sap – Phnom Penh area, and the description mentions Wat Phnom), and Bagan 蒲甘國.
Zhao Rukuo then talks about “Sanfoqi.” Previously, when I thought that Sanfoqi = Srivijaya, I couldn’t understand why Zhao Rukuo would jump from talking about Bagan to talking about a place on the island of Sumatra.
Now that I know that Sanfoqi = Angkor, its placement right after the above places from mainland Southeast Asia makes sense. Although it was on the mainland, and between Tonle and Bagan, it was diffferent from these other kingdoms in that it controlled a network of port-polities that Zhao Rukuo accordingly discusses after talking about Sanfoqi/Angkor.
After talking about Sanfoqi/Angkor, Zhao Rukuo then talks about various kingdoms that follow a route down the Malay Peninsula and up through the Straits of Melaka to the northern tip of Sumatra.
These are: Tambralinga 單馬令國 (Nakhon Si Thammarat), Langkasuka 凌牙斯國 (Pattani), Foluo’an 佛囉安國 (Marang?), Sunda 新拖國, Jambi 監篦國, and Lambri 藍無里國.
Further, all of these kingdoms are said to either pay tribute to Sanfoqi/Angkor or to have previously paid tribute (Jambi). In addition, under the account of Sanfoqi/Angkor, there are additional places along this same route that are said to be vassals of Sanfoqi/Angkor, such as Grahi 加囉希 (Chaiya)、Kelantan 吉蘭丹, and Palembang 巴林馮 (and note that Palembang is not written here as “Sanfoqi” but as “Balinfeng”).
These kingdoms are then followed by two kingdoms that are closely related to each other, but which are not connected to Sanfoqi/Angkor: “Java” 闍婆國 (Patthalung) and “Sujidan” 蘇吉丹 (Sating Phra).
Before explaining how these two kingdoms were related, I need to take a detour to explain how I determined that these two names refer to Patthalung and Sating Phra. First, we need to talk about “Java.”
There are two “Javas” in Chinese historical sources, and one appears earlier than the other. The first “Java” is the one just mentioned above, one which is referred to by two characters that are today pronounced “Shepo” 闍婆.
It may be difficult to see how “Shepo” could be “Java,” however many of the first place names from Southeast Asia to appear in Chinese sources were recorded by Chinese Buddhist monks who rendered these names using the same techniques that they used to transliterate Sanskrit terms into Chinese. When one follows the logic of how Sanskrit terms were transliterated into Chinese, it becomes obvious that this “Shepo” can be read as “Java.”
This term first appears in the Songshu (History of the [Liu] Song 宋書; completed 492-493 AD) and the Nanshi (History of the Southern Dynasties 南史; completed 659 AD) where there are records of a polity by that name presenting tribute in 410 and 412 AD.
Actually, in these works, this term does not always appear simply as “Java.” Instead, it appears variously as follows: Java islet 闍婆州/洲, Javasa islet 闍婆娑州, Javada kingdom 闍婆達國, Javasada 闍婆娑達, and Javasada kingdom 闍婆娑達國,
There are also statements that indicate that Javasa islet 闍婆洲 was the capital of a certain Haratan kingdom 呵羅單國 (呵羅單國都闍婆洲/呵羅單國治闍婆洲), as well as a record that indicates that Java islet and the Haratan kingdom presented tribute together (闍婆州訶羅單國遣使獻方物).
As such, this “Java” entered Chinese sources through a variety of names. However, starting in Tang Dynasty sources, these various names were homogenized into a standard term: the kingdom of Java 闍婆國.
Much later, in the Yuanshi (History of the Yuan 元史; completed 1370), a different “Java” appears – 爪哇.
These characters represent a phonetic transcription of the sound “Java” and these are the characters that are used today to refer to the island of Java.
Was this “Java” the same as the earlier “Java”? No (and we’ll see below where the first “Java” was located), but the Qing Dynasty-era scholars who compiled the Mingshi (History of the Ming 明史; completed 1739) assumed that it was, because, in a section on the new “Java” 爪哇, they added at the end some historical information about the old “Java” 闍婆, thereby creating a link that the historical sources do not actually support.
Indeed, in the Yuanshi, the Mingshi, and the Ming Shilu (the Veritable Records of the Ming 明實錄) there are accounts of interactions with the new Java at the same time that there are records of tribute being sent from the old Java. Further, that a new term that refers to island Java would appear in the sources for the Yuan period makes sense because that period marked a new era in Chinese interactions with island Southeast Asia as the Mongols launched an attack on the island of Java.
So if the old “Java” was not the same as the island of Java, then where was it? Zhao Rukuo notes in his thirteenth-century Zhu fan zhi that this kingdom was also called Pujialong 莆家龍. To me, this resembles Patthalung, the name of what is now a province in southern Thailand.
Let’s check, however, to see if it is possible that this place that Zhao Rukuo mentioned could indeed refer to a place that we can associate with the area of Patthalung.
In fact, as we read further into Zhao Rukuo’s account, we can see that the place he describes is situated in a location exactly like that of Patthalung. In particular, he notes that one can reach the sea by traveling in all of the four directions. To the east, the sea is very close. To the south, it is a three-day journey, while it is a four-day journey to the north, and it takes 45 days to reach the sea by going westward.
東至海，水勢漸低，女人國在焉；愈東則尾閭之所泄，非復人世；泛海半月，至崑崙國。南至海三日程，泛海五日至大食國。西至海四十五日程，北至海四日程。西北泛海十五日，至渤泥國；又十日，至三佛齊國；又七日至古邏國；又七日至柴歷亭，抵交趾，達廣州
While I think that we always have to be skeptical about numbers in Chinese historical texts (as they are easy to miswrite
), let’s just look at how these numbers relate to each other and consider to what degree they could be describing a place like Patthalung.
It’s quite easy to see that these numbers match the geography of the Patthalung region very well. It borders the sea to the east. The distance to Nakhon Si Thammarat to the north and Perlis/Kedah to the south are relatively close. Meanwhile, there are mountains to the west, and therefore it would take longer to reach the sea by going in that direction.
As for where the capital of this “Java” was, the best candidate for that region would be the place that is now called “Songkhla,” a place that early European travelers referred to as “Singora” and which supposedly derives from the word “Singapura,” meaning “Lion City.”
For the sake of clarity, I am going to refer to this “Java” as “Java/Singora.”
What else does Zhao Rukuo tell us about Java/Singora? He says that there was another kingdom called “Sujidan” that was Java/Singora’s “branch kingdom” (蘇吉丹，即闍婆之支國).
Zhao Rukuo explains the situation as follows:
“There is a vast store of pepper in this foreign country (此 番) [meaning, Java/Singora] and the merchant ships, in view of the profit they derive from that trade, are in the habit of smuggling (out of China) copper cash for bartering purposes. Our Court has repeatedly forbidden all trade (with this country), but the foreign traders, for the purpose of deceiving (the government), changed its name and referred to it as [Sujidan].” (Hirth and Rockhill, 78).
(此番胡椒萃聚，商舶利倍蓰之獲，往往冒禁潛載銅錢博換。朝廷屢行禁止興販，番商詭計，易其名曰蘇吉丹。)
While this passage suggests that Sujidan was simply Java/Singora by another name, there is other information in Zhao Rukuo’s text that indicates that Sujidan was a separate place, but one that was very close to Java/Singora.
I think a perfect candidate for such a place would be the Sating Phra Peninsula, a thin strip of land between the sea and two inland lakes in the Patthalung region. Archaeologists have found abundant evidence of habitation on this peninsula, as well as evidence of canals that were constructed across the peninsula.
Further, Zhao Rukuo states that the defining feature of this location was a mountain that could be seen from afar and that had other peaks nearby (有山峻極，名保老岸。番舶未到，先見此山。頂聳五峰，時有雲覆其上。)
Today the symbol of Patthalung Province is just such a mountain, a peak that is today known as Mount Ok Thalu. This mountain has a hollow opening through its peak, and can be seen from far away.
Zhao Rukuo said that the mountain that could be seen when approaching Sujidan was called “Baolao’an” 保老岸. This could be a transliteration. It could also be the case that the first two characters transliterate a local term where as the third has meaning in Chinese as “shore” and thus could be understood as as “Baolao shore.” Further, the character that can be translated as “shore” (an 岸) is very close to another that means “cliff” (yai 崖). “Baolao” in his historic pronunciation could also be “Thalu.” So it is conceivable that this could have originally been recorded to indicate “Thalu Cliff.”
Regardless of how we interpret the name that Zhao Rukuo provides, the geographic information about Java/Singora and Sujidan make the Sating Phra Peninsula or a place that one reached after passing through a canal on the Sating Phra Peninsula, a likely location for Java/Singora’s “branch kingdom.”
I am therefore going to refer to Sujidan as “Sating.”
So Java/Singora and Sating were closely connected and were the center of so much trade that the Song Dynasty government sought to restrict merchants from trading there for fear that too much copper cash was being lost.
Another interesting point regarding Java/Singora is something Zhao Rukuo recorded about the sea to the south. As noted above, Zhao Rukuo said that to the south of Java/Singora it was only a three-day journey to the sea.
In the early nineteenth century, Europeans mapped out a clear overland trail that went southward from Java/Singora to Kedah on the western coast of the Malay Peninsula. Colonial administrator and scholar John Crawfurd recorded that by elephant the trip took five days. People can move faster than elephants, so perhaps Zhao Rukuo’s statement about it taking three days was correct.
Kedah was one of the main kingdoms in the region and there is clear evidence from Arabic sources that this was a place that traders coming from the Middle East and India frequented.
Interestingly though, Zhao Rukuo did not discuss Kedah. Nonetheless, this has to be the place that was a three-day journey to the south from Java/Singora. What is more, Zhao Rukuo then said that in crossing the sea from that point, one reaches “Dashi” in five days. (南至海三日程，泛海五日至大食國。)
“Dashi” 大食 is a term that is used to refer to the Arab world, but in this time period, it could refer more generally to traders from a variety of places to the west of Southeast Asia, from Arabia and Persia to parts of India.
Regardless of where we see “Dashi,” it was not five days away from Kedah. Seeing information like this, there have long been scholars who have wondered if there was an Arab/Persian settlement somewhere in Southeast Asia.
Again, I think we always have to be wary of numbers. What we can glean from this information, is that Java/Singora was connected to Kedah, and Kedah had connections with the world of Arab, Persian, and Indian traders.
In other words, Kedah was the terminus of an overland trade route that crossed the Malay Peninsula from Java/Singora and that Java/Singora was closely connected to (or perhaps even at times controlled) Kedah. What is more, these three closely-interrelated places – Java/Singora, Sating, and Kedah – were not mentioned as part of Sanfoqi/Angkor’s network of vassals.
When we look at the long list of kingdoms that are connected to Sanfoqi/Angkor, we can see that the final one, Lambri, was also the terminus of a trade route, a route that went down the Malay Peninsula, but bypassed Java/Singora, and up through the Straits of Melaka to Lambri at the end of the island of Sumatra.
While Zhao Rukuo did not have in his mind the concept of “Southeast Asia,” he nonetheless organized information in a way that separated information about trade routes in Southeast Asia from information about places beyond those trade routes on the other side of the Indian Ocean and Bay of Bengal.
The first kingdom that Zhao Rukuo mentions in that distant region is a place that scholars have not been able to clearly identify, and that is a kingdom called “Nanpi” 南毗 which Zhao Rukuo claimed ruled over Sri Lanka. He also claims that this was a vassal of Sanfoqi/Angkor.
These comments have perplexed scholars, but personally, I think we can see these statements as referring to trade rather than political relations.
Zhao Rukuo’s text then mentions a place called “Gulin” 故臨 which scholars have associated with Kollam in what is now the state of Kerala on the Malabar Coast of India.
Zhao Rukuo says that ships arrived here from places to the east such as Sanfoqi/Angkor, Jambi and Kedah 吉陀, and that people from “Dashi” congregated there as well.
All of this discussion leads up to the point that I wish to make here. When one understands that Sanfoqi refers to “Kambuja/Angkor” rather than an imagined “Srivijaya” on the island of Sumatra, and when one understands that the earlier “Java” 闍婆 refers to Singora, rather than the island of Java 爪哇, then it becomes possible to see premodern Southeast Asian history from a very different perspective, but one that fits much more logically with the historical sources.
What Zhao Rukuo clearly documented in his Zhu fan zhi was a world that was divided between two powerful Southeast Asian trading kingdoms and two trade routes.
Java/Singora controlled the trans-peninsular trade route from Kedah to Patthalung. Angkor then at some point tried to bypass Singora by linking together port-polities down the eastern coast of the Malay Peninsula and up the eastern coast of the island of Sumatra.
This was a major trade rivalry, and Zhao Rukuo says exactly that. He notes that Singora has a rivalry with Sanfoqi/Angkor and that the two attack each other. (與三佛齊有讎，互相攻擊。)
When we realize this point – that there was a major rivalry between Java/Singora and Sanfoqi/Angkor – then there are many episodes in the history of the region that start to make sense.
[The stuff I’m going to write below here is coming from my head, so forgive me if I get some details wrong, but I’m sure the gist is right.]
The founding of Angkor supposedly took place when one Khmer ruler escaped from “domination by Java.” This happened in the late 700s/early 800s. However, and if my memory serves me well, the inscriptions that record this information date from the 11th century. This has led scholars to wonder if they are fabrications.
However, Cham inscriptions and Vietnamese historical records indicate that “Java” attacked in the 700s (There is a great article by Arlo Griffiths that talks about this). Further, Chinese sources say that “Java” 闍婆 attacked “Sanfoqi” in the late tenth century. Scholars have seen this as an account of “island Java” 爪哇 attacking “Srivijaya,” but it is a record of Java/Singora 闍婆 attacking Sanfoqi/Angkor.
In which case, we can see a reason why inscriptions about overcoming domination by “Java” would be carved in stone (they may have been recorded before that time on palm leaves) in the eleventh century after another attack from the same “Java.”
In addition, we can also see why Melaka became important when it did. Ayutthaya’s roughly simultaneous conquest of both of these major trading centers in the fourteenth century surely must have disrupted these two trade networks, and Melaka was a perfect “middle ground” between the two routes for all of the traders involved to converge on to establish a new network, or networks.
And last but not least, the Chola invasion of Southeast Asia can also be seen as a part of this trade rivalry. However, I will leave that topic for others to detail.
But finally, why was the name “Java” (Shepo 闍婆), used for so long when that clearly was not the name that the kingdom in question called itself? I think that can be explained simply by pointing out that there are names that foreigners use for places that don’t change even though the polity the name refers to does. “Siam,” for instance, is a perfect example of this.
There is more information that we could point to, but I think this is enough for now. What’s important is that the story that we have been telling about premodern Southeast Asian history for the past century is not the real story of premodern Southeast Asian history.
The real story of premodern Southeast Asian history is the story of the great rivalry between Singora and Angkor. It was a rivalry between one kingdom that controlled the best trans-peninsular land trade route and another kingdom that controlled the best trans-regional sea trade route.
