In spite of the doubtful aspect of Vietnam's population data (as shown above), we find quite a number of population estimates and censuses, which, though they canft be considered as reliable, still allow to work out population figures for the three main regions of Vietnam, i.e. Annam, Cochinchina and Tonkin (see table 1 in appendix 2). Over the whole period (1901-1943), Tonkin remained the most inhabited region, partly because of the very high density in Tonkin's delta. Annam remained the second region for the number of inhabitants, while Cochinchina was the less inhabited and the smallest region of Vietnam, still ranging, though, a territory of about 60 000 km2.
In 1901, the population of Vietnam was estimated at about 13 millions of inhabitants : 5.5 millions for Tonkin, 4.5 millions for Annam and 3 millions for Cochinchina. From then on, the evolution was almost regular for all of the regions until 1943. At that time, Tonkin counted 9.8 millions inhabitants, Annam 7.2 millions and Cochinchina 7.6 millions. Yet, the evolution was far more linear for Tonkin than for the other regions: at one hand, Annam lost some population between 1911 and 1921 as between 1926 and 1931, whereas Cochinchina lost some population between 1931 and 1936. However, at least part of these losses could be explained by statistical biases.
For the last series of data (1955), we only have two regions: in July 1954, Vietnam was divided into Southern Vietnam and Northern Vietnam bothering the comparison with previous series. The population of these two regions was calculated to be around 13.6 millions inhabitants each, in 1955.
Series of population by province, for the different regions of Vietnam in 1921, 1931, 1936, 1943, 1952 and 1953 are presented in three tables (2, 3 and 4 in appendix 2). First of all, we have to insist on the problem posed by the different divisions, according to the year. As shown above, it is not obvious to use the series that are at our disposal.
First, data for population by province and by gender only exist for 1921. For the other years, figures do not present any detail but provide a global figure for total population. An other problem comes from the incomplete data in 1952 and 1953. As a matter of fact, the population, in all of the provinces, is classified in two categories, the controlled and the uncontrolled population. Yet, for many provinces there is no suggested figure and no estimate, even incomplete, is put forward. This is particularly true for Annam and Tonkin, while Cochinchina has been granted by more reliable data. Moreover, caution is ordered for the uncontrolled population estimates in the different provinces, in view of the very nature of this population (uncontrolled, and therefore not easy to estimate), while it is already difficult to rely entirely on the available figures for the controlled population.
The changes in provinces' territory constitute a third difficulty, but this problem has been evoked already and it is not necessary to bring it up again. Concerning table 2 on the population of Annam, we can notice that, in most of the provinces, there is an increase from 1921 to 1943 (it is difficult to continue the analysis until 1953 because of the gaps observed for many provinces in 1952 and 1953). The most important provinces, in term of demographic weight, were Binh - dinh, Ha - tinh, Nghè-an, Quang-nam and Tranh-hoa. Their population increased over the whole period except for Quang-nam from 1921 to 1931 and Tranh-hoa between 1931 and 1936. These very densely inhabited provinces present important gender disparities in 1921. Female population was higher than the male population in the province Ha-tinh and Nghè-an whereas Quang-nam and Tranh-hoa present an overrepresentation of the male population. We may dare the hypothesis that it reflects migration consequences. We may assume this, unless it concerns, here again, underregistration of female and male populations, for reasons that we would be left to determine.
The population of Cochinchina (Table 3) presents the same pattern apart from the multitude of figures on the controlled population in 1952 and 1953 which allowed in that case to make totals for 1952 and 1953 (which is impossible for Annam and Tonkin). The population distribution in Cochinchina is far more harmonious than in Annam as most of the provinces of Cochinchina are rather close in population size : from 83 000 through 327 000 inhabitants in 1921 and from 119 000 through 498 000 inhabitants in 1943. Only small provinces as Baria, Ha-tièn, Cap St Jacques and the island of Poulo - Condore, present reduced population figures.
In the Tonkin region, the most populated provinces are all located in the delta with exceptionally high density, even by East Asian standard. In the provinces of Hadong, Hai-duong, Nam-dinh and Thai-bin, population reached over 500 000 inhabitants in 1921 and 800 000 in 1943. It is difficult to use the series of 1952 and 1953 since they are very incomplete ; the population evolution is therefor discernible only until 1943. In these provinces of the Tonkin's delta, Hai-duong is the only one to register a population fall between 1931 and 1936, while the population of the other delta provinces increased regularly over the whole period. This pattern concerns also the other provinces of the region except Ha-giang and Hung-yèn between 1921 and 1931, Son-tày and Hanoi between 1936 and 1943, and Haiphong between 1931 and 1943 ; but in the latter case, we can observe a very important growth in 1952 and 1953 with a tripling of the population. We can wonder if this is not the expression of an underregistration of the population in the first series (1921, 1931, 1936, 1943). The same phenomenon characterises Hanoi.
Data on ethnic minorities (table 5 in appendix 2) are available for the different regions in 1921, 1931 and 1936. The "Vietnamese" population (Annamite for the French administration, also known as Kinh ethnic group) is indisputably the most represented in Annam, Cochinchina or Tonkin. They always represented more than 80% of the total population. Other ethnic groups had a more or less important weight : Thais in Tonkin, Indonesians in Annam, Cambodians and Chinese in Cochinchina. Muongs, Thais, Meos and Mans (or Yaos) were concentrated in Tonkin, but absent of the rest of Vietnam. Indonesians were particularly present in the Central part of the country, both in south Annam and North Cochinchina, as for Malays and Chams.
Cambodians and Minh-Huongs (Sino-Vietnamese), who were mentioned in Cochinchina, refer back to a pre-colonial legacy of previous invasion of this formerly Cambodian region by Vietnamese and massive Chinese immigration from the seventeenth through the mid-nineteenth century ; they were not only traders but also engaged in agriculture and fishing. For Vietnam as a whole, the Chinese minority was only the fifth most numerous ethnic group, but in the cities and towns it was the most important minority, and in Cholon Chinese even held the majority. In comparison, European population was insignificant. It only represented about 0.1% of total population in Tonkin and Annam, and 0.3% in Cochinchina. However, Navy officers and marines seemed not to have been considered as part of the resident population. Yet, even by French colonial standards European population was insignificant. This might be explained by the possibility for the colonial administration to rely on Natives for clerical and technical jobs in the administration, on Chinese for the development of food processing, and on domestics or Asians for trading networks and shipping.
At the present stage of research, we have at our disposal only fragmentary data for urban population in Hanoi, Haiphong, Saigon and Cholon (table 6 and 7, in appendix 2).
We observed, in the figures mentioned by Mitchell (1995), several discrepancies when comparing with the data collected in the official yearbooks. Mitchell figures for Saigon seem overestimated, which could be explained by the status of Cholon, an exception in the framework of colonial administration. However, the total figure for Saigon-Cholon remains rather different in the two series.
Part of the gap could be explained by underregistration of urban population corresponding to migrant workers from the countryside and their family. Village chiefs and village councils considered these persons are living temporarily elsewhere, and so did the migrants themselves. They were usually registered in administrative reports and censuses as resident in their native village.
At the present stage of research, only fragmentary data are available before 1913 and the series mainly concern public and private employment : civil and military administration, education and health services, mining. Furthermore, these series are usually characterised by discontinuities corresponding to changes in classification and estimation methods, and probably also by narrow definitions of employment in the branch. Let us take the case of the military administration : the figures clearly indicate the soldiers and officers, excluding therefore a large part of related services in transportation or construction carried out by the Army or the Navy (excluding sub - contracting to private firms). Nevertheless, it seems possible to comment long run patterns observed in these different branches.
For the employment in civil administration (Table 8), we have to rely for the preliminary report on data concerning employment by the General Government in Indochina as a whole, considering all of Indochina as an acceptable proxy for Vietnam. The strong increase of French employees in 1928, make this figure, and consequently the total, almost impossible to use in a comparison between before and after 1928 sub-periods. For this reason, we may focus on native employees series (but in that case, the figure for 1913 is suspect). Considering 1899 as a benchmark, we observe a limited increase for Natives while the share of French employees jumps from less than 10% in 1899 to over 25% in 1914. Assuming that this jump in the number of French employees in 1928 is caused by a statistical bias, we could conclude that this number remained almost stable over the second phase of development of the local administration, after 1914. More precisely, the number of Natives increased steadily during the 1920s, a period of prosperity, declined during the early 1930s, with the economic recession, and increased strongly in the late 1930s early 1940s. This indicates that, as far as the employment trend in civil administration is concerned, no significant change in the trend occurred as a consequence of WWII and Japanese military occupation.
Although the series of military, education and health services employment for Vietnam only are either discontinuous or shorter, we observe approximately the same patterns. Annamite guard enrolment is considered as a much better indicator than the total figures for Navy and army, because the changing military context in Europe and in the other French colonies during and after WWI, so as the changing perception of risks in East Asia before WWII, and, finally, the Indochina War after 1945, induced considerable changes in the level of French forces in Vietnam. After a decline from 1912 through 1922, justified by the enforcement of the French rule, the total number of guards increased during the 1920s, mostly as a consequence of their introduction in Cochinchina ; it stagnated again from 1931 until 1936, before a spectacular rise at the eve of WWII.
This pattern is also observed for the employment in public education and health services (tables 10 and 11). Once again, we use data of public services funded by the General Government or local administrations, because these series are more reliable than those including part of the activity by private agents. However, the double in 1918 of employment in public primary education indicates an institutional change, probably the incorporation of village council schools in the official system. But during the 1920s, the growing share of Europeans in total employment (1,7% in 1919, 4,2% in 1931) followed a double of total employment. This indicates an improvement both of quality and quantity, since very few Natives had the same level of qualification as the French employees in public education in the 1920s.
The difficulties of public finances during the early 1930s did not lead to a contraction of total employment, but only to a short stagnation. The figure increased rapidly during the late 1930s and this went with a rapid decline of French employees. This shift may be understood as a clear indication of promotion of Natives into higher ranks under strong budget constraint (important wage differential existed between French and local employees; although the basic income was similar, French teachers were considered as expatriates and received a bonus).
For health services, reliable figures are available for Vietnam only after 1929, but with both Native and European employees in each region (Table 11). Because of the successive changes in military context already mentioned above, military medical services are excluded. The most remarkable evolution concerns the double of native employees between 1930 and 1934, and a steady increase at the end of the 1930s, counterbalancing the declining number of European employees. Here again, the features could be considered as a consequence of budget constraint combined with the improvement in Natives qualification. The degradation of health service during WWII is followed by a rebound during the late 1940s, but previous levels are not restored before 1950.
The same pattern in resource allocation is observed in the case of private employment, particularly in the easily available series in coal mining (Table 12). Indochina is an acceptable proxy for Vietnam, since almost all of the production is concentrated in Tonkin. There is also metal ore mines employment, but part of the production is originated in Laos and available series are too discontinuous. Coal mining was both a significant exporting sector, extremely competitive in East Asian market, but also an important sector for French companies investing in Vietnam (about 27% of total employment of these companies in 1939, see Table 13). A significant gap exists however between estimates on employment by French or foreign-owned coal mining companies in Vietnam (Table 13) in 1939 and official sources on this peak year (85000 versus 55000). The transformation of coal mine products by subcontractors preparing briquettes of a mixture of coal dust and clay, to be sold in Hanoi and Haiphong to lower class Natives, constitutes a possible explanation.
The major features in coal mine employment are : take-off during the first decade of the twentieth century, steady growth during the 1910s, up to 300 % during the 1920s, stagnation during the 1930s, decline during WWII and stagnation at the 1910 level from then on. Political unrest in Tonkin was the main factor of the post WWII stagnation, but we may suppose complementary factors.