How Did Racial Categories Form in Malaya?

In 1901, the Federated Malay States conducted a census.

Afterwards, the official in charge wrote a report and submitted it. The report contained many tables, many numbers, and also some recommendations for his superiors. One of them read like this:

"before printing the next set of census forms, the word 'nationality' should be changed for that of 'race' whenever it occurs. It is a wider and more exhaustive expression than 'nationality' and gives rise to no such ambiguous questions in classifying people"

The same official explained this even more clearly in a document that same year. He wrote: wherever the word "Nationality" is used in census forms or report tables, it means the same as "that wider and better expression: Race"—the example he gave was: "After all, it's quite impossible that there can be a Eurasian nationality."^1,2

This recommendation was adopted. By 1911, "race" had become the official term.^1,3

This article is about that word.

Before That, Another Word Was Used#

In the 1881 and 1891 censuses, the heading for that column on the form was not "race", but "nationalities".

The word "race" first appeared in a Malayan census document in 1891—but not in the table heading; it was in an appendix: it was in an operational guideline for the enumerators. In other words, it began as an internal working term, was proposed for official use ten years later, and put into the table heading ten years after that.

Between "nationalities" and "race", thirty years passed, mediated by several administrative reports.

At no point did anyone announce they had discovered something.

And It Was Changed Several Times Later#

If "race" were a stable, discovered entity, it should have remained fixed once established. It did not.

• In 1947, the census superintendent deliberately changed the word, using "community" instead of "race", explaining that the term was meant to convey "a group of persons bound together by common interests, i.e. common language, religion, customs, and allegiances."
• In 1957, the year of independence, it was changed back to "race".
• In 1970, it reverted to "community".
• In 1980, the question on the questionnaire became: "To which ethnic group, community, or dialect group do you belong?"

Scholar Charles Hirschman, who studies these classifications, calls this process a circular movement—from the relatively neutral "nationality", moving to "race" as a pseudobiological concept, and then returning to the relatively neutral "community" and "ethnicity".

In a hundred years, the box that categorises people changed its name four times. The people inside the box did not change.

The Person Managing This Classification Told the Truth#

The 1931 census covered all of British Malaya, superintended by C.A. Vlieland. He dedicated a section in his report to explaining what the word "race" meant in this census.

That section reads as follows:

"The term 'Race' is used, for the purposes of a Malayan census, in a peculiar sense, which requires explanation. … the word 'Race' is used, for lack of a more appropriate term, to cover a complex set of ideas of which race, in the strict or scientific sense, is only one small element.

>

It would be of little use to the administrator or the merchant to attempt a classification of the population by race in the ethnographic sense… It is, in fact, impossible to define the sense in which the term 'Race' is used for census purposes; it is, in reality, a judicious blend, for practical ends, of the ideas of geographic and ethnographic origin, political allegiance, and racial and social affinities and sympathies."

Please read these lines slowly.

The person who wrote this was the official in charge of counting the entire population of Malaya at the time. He wrote down in the census report itself: it is impossible to define the usage of this word; it is in reality a judicious blend for practical ends—mixing geographic and ethnographic origins, political allegiance, and racial and social affinities.^1,2

(This site has not read the original 1932 report for this segment. But two independent scholars have each read it and each cited these words from different sections of it—Hirschman quoted from the introduction, Manickam from pages 73 to 74. Two paths independently converge on the same document.)

He also wrote another sentence:

"most Oriental peoples have themselves no clear conception of race, and commonly regard religion as the most important, if not the determinant, element."

This sentence can be read in many ways. It carries the condescension of European officials looking at the Orient in that era, which needs no covering up. But it also states a very tangible fact: the thing he was trying to measure was something the people being measured did not see in themselves.

The Same Word Meant Different Things to Different People#

Vlieland said he could not define the word. Open the actual operational documents, and you'll see this was not modesty—for the same "race" column, what was measured in different groups was not the same thing at all.

For the Chinese, it measured language. In the 1901 report, Hare specifically noted: "The column 'Race' in dealing with the Chinese should have been made clearer, to show more specifically that the race or tribe of a Chinese was determined by language, and not by birthplace."^1,2

For Sikhs, it measured place names. Indians of the Sikh faith were typically classified in the census as "Bengali by race"—religion, race, and place name thus became synonymous with each other.^1,2

For Europeans, it measured very finely. Although their numbers were very small in the total population, the census provided them with highly detailed racial sub-classifications.^1,2

For the Aborigines, the method of measurement was to go in and find them. A circular for the 1901 FMS census read:

"it is highly desirable that special efforts be made to ascertain the number of Sakai and other indigenous peoples living in each state. If necessary, a few trifling presents might be offered to them in various districts, so that they may be induced to come to the nearest villages for the purpose of the census"

By 1911, Census Superintendent Pountney noted in his report that the census costs were very low—considering they had to send out "special enumerators" to visit Sakai camps.^1,2

This method later acquired a name, the "Sakai census", distinguishing it from the regular door-to-door enumeration.

Scholar Sandra Manickam, who studies this period, uses one word to summarise it: incommensurability. The same "race" measured different things for each group; there was simply no common yardstick between them.

A yardstick that measured everyone differently produced numbers that were placed side by side into the same table. That table is the starting point for all later demographic proportion debates.

So How Were These Boxes Drawn?#

In the 1871 and 1881 census reports, there was no explanation of the basis for classification at all. It was not until 1891 that an explanation of the classification logic appeared for the first time.

The 1891 Straits Settlements census marked the first systematic ethnic classification. The main headings were: Europeans · Eurasians · Chinese · Malays and Other Natives of the Archipelago · Tamils and Other Natives of India · Other Nationalities.

The direct predecessor to today's "three major races" framework lies here.

And Hirschman's judgement on this classification system is: these categories had to be "invented"—from experience, from the needs of colonial administration. He uses this word, and puts it in quotation marks.

The Boxes Kept Moving, That Is the Key#

If these boxes corresponded to some real, innate boundaries, they wouldn't move. Look at how much they moved:

The "Malay" box was widest during the colonial era#

• The main heading in 1891 was "Malays and Other Natives of the Archipelago", which contained: Aborigines · Achinese · Boyanese · Bugis · Dyaks · Javanese · Jawi Pekans · Malays · Manilamen
• By 1931, the main heading had changed to "Malaysians by Race"—broader than "Malays"
• In 1957 and beyond (i.e., after independence), a single term "Indonesian" was used to encompass all peoples originating from the Indonesian archipelago

This box was at its widest when the British were in charge. It only narrowed after independence.

⚠️ This point must be highlighted: for the year-by-year comparison of classifications above, this site has only one source—the classification table in the appendix of Hirschman's 1987 paper. This site searched for comparative materials from a second independent scholar (checking Shamsul, Milner, Manickam) but found none. This "widest during colonial era, narrowed after independence" trend is currently compiled by him alone.

Aborigines were consistently counted within the large "Malay" category#

"Aborigines" was used from 1881 to 1911; the 1911 FMS census introduced the derogatory term "Sakai", which was used until 1931; 1947 and 1957 reverted to "Aborigine" and listed specific ethnic groups; 1970 and 1980 began frequently using "Orang Asli".

And throughout this entire period, the indigenous peoples were generally subsumed under the larger Malay category.

For the significance of this, see this site's article Those Who Were Already Here Before.

The Chinese were never merged into a single box#

This point overturns a widely circulated notion—which this site also believed prior to verification.

In the 1871 census, indeed only the single word "Chinese" appeared under the Chinese column, with no sub-categories. But the biggest change in 1881 was precisely the addition of dialect groups (Hokkiens, Hylams, Kehs, Macaos, Straits-Born, Teochews). Thereafter, every census retained dialect group sub-categories right up to 1980. After 1921, the main heading was standardised to "Chinese", but the sub-categories beneath it did not disappear.

The trajectory is contrary to intuition: it was not a process of merging from specific to general, but rather starting general, then being specified, and never being erased.

Meaning: in the eyes of the enumerators, there was always differentiation within the Chinese. It was not the census form that flattened these differences into a single "Chinese" category.

The Indian box was never consistent#

In 1871, the word used was "Klings"—only for that year. Five hundred years ago in Malacca, that word was a neutral geographical derivative, as this site explained in From Kalinga to Keling; by 1871, it had appeared in the classification column of a government document.

The 1911 FMS census had begun separating Tamil, Telugu, Punjabi, Bengali, Malayali, etc.; only in 1947 were Sikhs listed separately for the first time. As for the classification of Ceylonese and Indian Muslims, to use Hirschman's own words, it was historically "varied and inconsistent".

In 1907, Someone Put It Plainly#

Classification is always written about technically in official documents. But elsewhere, the same system was written about without any disguise.

In 1907, a European miner in Malaya published a book containing this passage (quoted verbatim, without alteration by this site):

"From a labour point of view, there are practically three races, the Malays (including Javanese), the Chinese, and the Tamils (who are generally known as Klings). By nature the Malay is an idler, the Chinaman is a thief, and the Kling is a drunkard, yet each in his own class of work is both cheap and efficient, when properly supervised."

This passage is cited in Hirschman's 1986 paper, The Making of Race in Colonial Malaya (Sociological Forum Vol. 1 No. 2, p. 356).

This site quotes it as evidence, not as a description. The three adjectives that follow are vicious and baseless; this site will not paraphrase, transcribe, or translate them into any Chinese phrasing—they remain in quotation marks, presented exactly as they are, because they themselves are artifacts of that era.

But the true weight of this passage lies in its first sentence:

"From a labour point of view".

"There are practically three races".

The starting point for classification was stated by the man himself: not anthropology, but employment. The "three major races" schema, here, is the administrative convenience of a mine manager managing workers.

Note two details in passing. He wrote "the Malays (including Javanese)"—Javanese were classified as Malays, consistent with the census at the time; that boundary was movable. He wrote "the Tamils (who are generally known as Klings)"—four hundred years prior in Malacca, that word was still neutral.

The Evidence for This Article, Explained Clearly by This Site#

This article stands on the shoulders of two scholars, not on original documents. This distinction must be made clear.

Charles Hirschman's 1987 paper in the Journal of Asian Studies, 〈The Meaning and Measurement of Ethnicity in Malaysia: An Analysis of Census Classifications〉, is the source for the evolution of terminology, the classification comparison tables, and Vlieland's confession. This site obtained the PDF made public by the author himself on his University of Washington personal page—it is a scanned image, which was OCR'd and then visually verified sentence by sentence against the original images before being written here. His 1986 paper 〈The Making of Race in Colonial Malaya〉 (Sociological Forum 1(2)) is the source for the 1907 miner's quote (p. 356), and this site also read the full text. But two papers by the same author do not count as two independent pieces of evidence.

Sandra Khor Manickam's 2014 paper in Asian Studies Review, 〈Bridging the Race Barrier: Between "Sakai" and "Malays" in the Census Categorisations of British Malaya〉, is the truly independent second leg—she directly cites Hare's 1901 and Pountney's 1911 original reports and provides page numbers, rather than citing Hirschman. This site obtained the self-archived version she deposited in the Erasmus University Rotterdam repository, with its text layer intact. The determination of Chinese race by language, Sikhs being classified as "Bengali by race", and the circular about offering "trifling presents if necessary" all come from this paper.

However, this site has not yet laid eyes on the original documents cited by these two scholars. Vlieland's 1932 confession, Hare's 1901 proposal, Del Tufo's 1947 explanation for changing the word—what this site has read are the scholars' citations, not the reports themselves.

This site has not found digitised versions of the original census reports. Vlieland's 1931 report, Nathan's 1921 report, Pountney's 1911 report, and the 1891 report—all were attempted: the PDF certificate on the Malaysian Department of Statistics website has expired, HathiTrust and the National University of Singapore's digital collections deny access, and the Internet Archive does not have them. The only direct primary corroboration this site obtained is the bibliographic record for the 1911 report—its title indeed says "by sex, age, race, birthplace, religion and occupation".

So the position of this article is this: the argument is supported by complete research from two independent scholars, both of whom have directly read the originals, which this site has not yet done. This is much stronger than a single scholar, but still not the highest tier of evidence. If and when those census reports are digitised, this article will be rewritten.

---

Next article: Where did the people who were put into those three boxes come from? Tin, rubber, and the migration of three million people.