Freedmen in a slave economy: Minas Gerais in 1831.

Paiva, Clotilde Andrade

There is little question that Brazil by the early 19th century had the largest free colored population of any slave society in America. By the first national census of 1872 the free colored - all of whom came from slave origins - numbered 4.2 million persons, compared to just 1.5 million Afro-Brazilian slaves. Moreover these free colored were the largest single racial/status group within Brazil itself.(1) Yet this was a time when the slave coffee economy was reaching its maturity and the price of slaves was on a long term rise.(2) There is also little question that Brazilian society, like all other slave regimes, was racist and that the white elite in various ways discriminated against its freedmen, even as it permitted a very active level of manumission.(3) But until now we have little sense of how these freedmen were integrated into the world of the free market. Were they cut off from normal avenues of economic and social mobility, as occurred for example, among the free colored in the United States?(4) Or were they far more integrated than well known cases of racism would seem to suggest? In fact, can Brazil's long resistance to black consciousness and its self-perception as a racially harmonious society be related to the experience of these numerous Afro-Brazilian freedmen long before the abolition of slavery?

We will argue in the following essay, that in fact, the free colored population, except at the elite level, were to be found in all the occupations practiced by their contemporary white neighbors and experienced much of the same social and demographic organization as their non-slave-originated peers. It will also be shown that whether they lived among predominantly Afro-Brazilian populations or among predominantly white ones, there was little difference in the patterns of work and social organization for the free colored from those of their white neighbors. Finally, we will show that freedmen were even significant slave owners in their own right.

Surprisingly, for all the recent studies of African slavery in Brazil, there is almost nothing on the life of the free colored population.(5) It is our aim to analyze this neglected class of colored Brazilians through the same sources that have recently been exploited to study slavery in Brazil. For many years Brazilian economists and historians have been exploring the theme of slavery in early 19th-century Sao Paulo and Minas Gerais by using the previously unpublished and un-analyzed "mappas" or censuses of population and production which were carried out in both regions with some regularity from the 1770s until the early 1840s. These extraordinary censuses have enabled scholars to revise previous visions of rural society and the slave system in these economically dynamic Brazilian regions. Instead of a hegemonic large plantation system as assumed in the seminal work of Gilberto Freyre, it turns out that most of Brazil in the 19th century consisted of small slave estates imbedded in a largely free labor economy. In these two regions, for example, slaves themselves never exceeded a third of the total work force, and slaveowners never were more than a third of all households.(6) In short, the size and relative weight of the slave population and their masters in Brazil differed little from those in the contemporary southern United States.(7)

But where Brazil did differ substantially from the United States is in the race of its free population. Whereas the non-slave households were over 95% white in the United States, they tended to be less than 50% white in Brazil.(8) Freedmen also were to be found as heads of slave-owning households in quite significant numbers, again in sharp contrast to the United States where less than 1% of all slave owners were non-white. But despite their importance, freedmen have been neglected in the recent research.(9)

For this initial study of the freedmen, we have decided to examine the largest province of Brazil in terms of total population, slaves and free colored, that of Minas Gerais. Within this province we have selected two unpublished 1831 censuses from two major municipios,(10) that of Campanha in the southwestern part of the province, and Sabara in the central zone near present day Belo Horizonte. Both regions are roughly similar in the structure and size of their population and in their dedication to agricultural and artisanal activities. Although some mining activity still occurred in Sabara (gold mining had made the region famous in the 18th century but had declined after 1750), it was no longer the predominant sector of the local economy. By the end of the 18th century the province of Minas Gerais bad become a very complex agricultural, proto-industrial and mixed mining economy of which gold production was a minor element. The Minas Gerais economy with its exports of sugar, cane alcohol, food staples and low quality woven cotton cloth, more resembled the economy of neighboring Sao Paulo than it did its former colonial self.

Both Campanha and Sabara were rather typical of the province as a whole in their concentration on agriculture with a minor but important share of activities in crude textile manufactures, some metal and wood working and some mining activity. In the two zones there was a significant sugar refining industry producing cane alcohol (aguardente) for local consumption, and both also had a very active commercial sector. Though there was considerable self-sufficiency, both zones were closely tied into a larger zonal economy that included very active trade with the neighboring coastal provinces.(11)

But there were also important differences between Sabara and Campanha, the most significant of which was their divergence in terms of racial composition. It was this particular factor which justifies our selection of these two differing regions. Using the extremes in racial composition represented by these two municipios, we can control for racial density as a key factor in determining integration or rejection of the freedmen into non-slave society. By the racial standards of Minas Gerais in the first half of the 19th century these two regions incorporate the extremes. Campanha is among the whitest of the Minas Gerais municipalities, and Sabara is one of the blackest in color terms. Sabara had been one of the major mining districts which had possessed a large slave work force, which in turn explains its very high incidence of colored population. The decline of the gold mining economy in the second half of the 18th century had forced Sabara with its many small towns into a greater dedication to agriculture and to artisanal activities. Campanha, which had never experienced the mining boom, was a more recent area of development and had concentrated on agriculture in its richer soils from the very beginning. Because it was located close to the very large Rio de Janeiro market, Campanha's sugar, ranching and agricultural economy was more directed toward exports than were comparable sectors of the more distant Sabara. Its lack of a mining past, its more recent development and consequent in-migration, and its greater export orientation explains why Campanha was a zone with an unusually high concentration of white residents. These two municipios thus represent the most basic patterns evident in the province of Minas Gerais, incorporating both a new major agricultural export economy with low urban concentrations and one with a more urbanized and mixed agricultural and artisanal base including some left-over mining activity. The some 74,000 persons found within the borders of these two municipios represented about 12% of the total provincial population, which in turn was the most populous province of the Brazilian empire in the 19th century.(12)

Despite the fact that Campanha had one of the highest ratios of whites of any municipio in Minas Gerais, it still had a majority of its population defined as Afro-Brazilians (see table 1). In Campanha 54% of the total population, free and slave, were non-white and in Sabara 83% were so defined. But the relative weight of the freedmen differed in both communities. The free colored made up less than half of the total colored population (or 46%) in Campanha. In Sabara, which had roughly the same overall percentage of slaves as did Campanha, the freedmen were the majority and accounted for 66% of all Afro-Brazilians. In Campanha only 34% of the free population were non-white, whereas 77% of the free persons were so defined in Sabara.

Most freedmen did not live in households which held slaves, and this pattern did not differ significantly between either zone (see table 2). Non-slave-owning households made up 70% of all such units in both municipalities, but the total participation of freedmen in these non-slave housing arrangements differed because of the different weight of each group within their respective zones. In Campanha, where freedmen were just 34% of all the free persons, they headed 45% of the households which owned no slaves. In Sahara, were they were 77% of all free persons, they made up 88% of the heads of such non-slave households. In both cases the total number of free colored living within these non-slave households closely approximated their representation among the heads of such households (43% in Campanha and 89% in Sabara). Although most free colored in both communities lived away from the slaves, it is interesting to note that a higher percentage of them in the more white Campanha were found in non-slave households (88%) than was the case in Sabara where only 67% of all freed persons were living within non-slave households.

As is obvious from their distribution, in neither zone were the free colored simply confined to the non-slave holding households. They also appeared as members of slave-owning establishments. This distribution might be expected in any slave society which had large numbers of recently freed slaves who were likely to remain in the households of their former owners. But more astonishing was the significant representation of freedmen among the owners of slaves. [TABULAR DATA FOR TABLE 1 OMITTED] [TABULAR DATA FOR TABLE 2 OMITTED] Though whites dominated the slave-owning households in both municipios, in Sabara a very high 43% of these slave-owning households were headed by nonwhite males or females, while in the more white dominated region of Campanha the figure was still a significant 13%.

This distribution of freedmen between the two sets of slave-owning and non-slave-owning households, though still a biased distribution, could probably be mostly explained by economic factors. To own slaves required a higher level of wealth than was the norm in the society. The legacies of slavery certainly meant that freedmen had much lower levels of initial savings when they reached free status, and they carried over their initial educational disadvantages from slavery. But greater historical and familial poverty can not account for the internal divisions within the free colored of the two municipios. The residue of racism in the society as a whole is quite evident in terms of relative color differences among freedmen. Although pardos (or mulattoes) represented 61% of all Afro-Brazilians slave and free in Sabara, they were 84% of the total free colored. In Campanha 45% of all Afro-Brazilians were pardos, but they were 83% of the free colored class. Here the color bias against pretos (blacks) was clearly evident.(13) The racial prejudices of the white society guaranteed that manumission would favor those of mixed racial background as opposed to their non-mixed brethren.

Table 3 COLOR BY SEX OF FREE PERSONS IN SLAVE-OWNING HOUSHOLDS,
MUNICIPIOS OF SABARA & CAMPANHA IN 1831

 HEADS OF TOTAL HDS-HLD
COLOR HOUSEHOLDS POPULATION TOTAL

 men women men women Sex Ratio

 I - SABARA

White 812 223 2,228 2,301 364 97
Mulatto 534 205 2,039 2,254 260 90
Native Black 14 19 162 230 74 70
African 0 4 39 53 ... 74
Sub Total(*) 1,360 451 4,468 4,838 302 92

 II - CAMPANHA

White 1,152 228 3,586 3,451 505 104
Mulatto 119 35 378 389 340 97
Native Black 19 8 56 52 238 108
African 9 7 45 42 129 107

Sub Total(**) 1,299 278 4,065 3,934 467 103

Source: Same as table 1

* In Sabara there were a total of 1,834 slave-owning households,
of which 23 were headed by persons whose sex or color was unknown.

** In Campanha there were a total of 1,593 households with
slaves, of which 16 were headed by persons whose sex or color
was unknown.

Somewhat surprisingly these color biases do not appear quite so pronounced when we examine the color of the heads of households among the majority non-slave domiciles. In both communities, of course, pardos were the most numerous heads of household among the free colored. But in both communities they were fewer than their total number would suggest. Thus in Sabara where 83% of all free colored persons residing in such households were pardos, only 78% of these heads of non-slave households were of this racial group. In Campanha where 84% of the freedmen in non-slave households were pardos, such mulattoes headed only 81% of such establishments.

In contrast, among the slave-owning households in Sabara at least, the ratios were just the opposite. Pardos were 90% of all freedmen residing in slave-owning households, and 95% of the freedmen heads of such units (and 41% of all slave owners). In Campanha slave households looked more like non-slave ones in terms of the role of such mulattoes. Pardos here were 80% of all freedmen residing in these homes and were 78% of all freedmen owners of slaves (and 10% of all slave owners).

It thus appears that with the exception of pardo slave owners in Sabara, Africans and native-born blacks did better than expected among the free colored in terms of heading households. This would seem to suggest that while entrance into the free colored class was heavily biased by white prejudice, once that class organized itself, such a bias was far less influential in determining the stratification within the autonomous free colored community.

But race was not the only factor which stratified the households in these two regions. The sex of the head of household was more important in determining marital status of the head than was either the race or the ownership of slaves (see table 4). Single-headed households of unmarried or widowed tended to be more the norm for women than for men. In turn these single unmarried or widowed female headed households operated in different economic spheres from those headed by men of whatever color. It has generally been assumed from all studies of household types, that those headed by single or widowed women tended to be less stable in social terms and poorer than households headed by two adults. In this respect color shows little difference, and whites and mulattoes are quite similar in their marital rates in both regions for both sexes and for the two types of slave-owning and non-slave households, with the single exception of non-slave female headed households in Campanha (see table 4).

The influence of sex on marital status of head is present even in the supposedly wealthier slave-owning households. Although women running such slave-owning homes were far more likely to be married and widowed than those heading non-slave units, even these slave-owning domiciles had rates of single persons almost three times higher than among comparable male heads. Moreover this cut across color lines, with mulattoes being little different from whites.(14) Even in Campanha, among the females heading slave-owning households, the best marital rate achieved by any female heads, the ratio of single women was still 21%. The relatively higher wealth of Campanha showed not only among the higher marital and widow rates of women heads of households, but also in the fact that the men in both free and slave-owning households had higher marriage rates than the Sabara male heads of households.

Wealth, of course, also showed its effect, in both the different marital rates for both sexes between slave-owning and non-slave-owning households, and between the richer Campanha and the poorer Sabara. What is interesting to observe (see table 4) is that the ratios among the sexes remain constant even as wealth generated differences between the two types of slave-owning households and between the two communities. It would seem from these data that marital status for heads of households is a quite reasonable proxy for wealth and that the sexual biases were as important even in the wealthiest groups.

Aside from the fact that all slaves were colored (pardos or pretos), in only one other area could we discover color being a more important influence than type of household or relative wealth of the community, and that was in the question of fertility. Because of the lack of life tables, confusion as to color of children, and the tentative quality of using child-women ratios as a proxy for fertility, it is difficult to put great emphasis on the result of such studies in this case. But some tentative findings suggest that living in slave-owning households or those which had no slaves caused little difference in the rates of fertility, but that there were significant differences based on race.(15)

But when it came to the occupation of the heads of households, the factors of slave-ownership and gender were more important than a person's color in influencing the status and economic importance of the occupations listed in [TABULAR DATA FOR TABLE 4 OMITTED] [TABULAR DATA FOR TABLE 5 OMITTED] [TABULAR DATA FOR TABLE 6 OMITTED] these two communities.(16) In rural Minas Gerais in the 19th century, as in fact for the United States in the same period, at least half the rural population engaged in non-agricultural activities.(17) The non-slave households and those headed by women tended to concentrate in these non-agricultural activities while the slave households and those headed by men were far more heavily engaged in agriculture - a pattern not much affected by the color of the household heads. Nor was this a factor in just one of the communities, for we find that in both Sabara and Campanha this same relative balance exists. The fact of holding slaves meant that the domestic unit was far more likely to be producing sugar or other commercial agricultural crops and far less likely to be engaged in artisanal activity, just as households headed by women of whatever color were also far more likely to be engaged in artisanal activities.(18)

This pattern is well exemplified in the case of Sabara. Whereas only 28% of the heads of non-slave-owning households engaged in agriculture or ranching activities, some 51% of the heads of slave-owning families were dedicated to these activities (see table 5). Artisanal activities represented only 21% of slave-owning units, but accounted for 40% among the non-slave homes.

Within this picture, there was an ever sharper dichotomy between males and females. Some 41% of the males were in ranching and agriculture and 21% in artisanal activities among the non-slave households, but 73% of the female heads were listed as artisans and only 4% worked in agriculture and ranching. This same sexual division occurred in the slave-holding homes in Sabara. More women were in agriculture and ranching (30%), but 49% - the dominant group - were artisans. Fifty-eight percent of male owners of slaves were in agriculture and ranching and just 12% of them worked as artisans. Thus while slavery predisposed even female-headed households to work more in primary activities, the sexual differences were still strong.

Although sex and the ownership of slaves directly affected types of economic activity engaged in by households, color was not a significant predictor of economic activity. Non-whites - above all the mulattoes who represented the largest element in the non-white grouping - tended to follow the same trades as their white male or female compatriots and they had roughly the same participation rates in the same industries and activities favored by slave-owning and non-slave owning white heads of household.

The same two factors which influenced economic activities in Sabara did so in Campanha as well, though agriculture here was far more important as an occupation than in Sabara. Just over 70% of the males in slave-owning and free households were in agriculture and ranching in Campanha and half of the non-slave women heads and some 30% among those owning slaves were to be found in artisanal activities (see table 6). Here again the sexual division is more pronounced than the color one with mulatto heads breaking down into the same areas of labor as the whites for both types of domestic units.

In examining the occupations of the non-slave-owning households in greater detail, these same patterns of differences based on relative wealth in terms of slave ownership and the sex of the head of household are even more pronounced, and also cut across municipio boundaries. But there are some subtle differences based on color which do begin to emerge indicating either the continued impact of the poverty due to the slave heritage and/or race prejudice.

Examining the distribution of trades among male heads of non-slave-owning households of Sabara in 1831 (see table 7, panel 1) shows that the whites and mulattoes (pardos) were about even in the agricultural activities, but that mulattoes were more likely to be artisans than were their white compatriots. In general, mulattoes did better than the native-born blacks (crioulo-pretos) in a whole series of higher status occupations though the small number of Africans seem to have fared slightly better than the native-born blacks.

Within these broad patterns there were some modest variations. For example, among the farmers there was a tendency for whites to be more "lavradores" and for mulattoes to be more "agricultores,"(19) but there was otherwise little difference in their weight in the landowning-agricultural occupations. Both also did well as merchants (negociantes) - achieving above average ratios (in terms of their total participation among non-slave-owning male heads of household) in this category. But as the occupations fell in status there was a difference. Both whites and mulattoes were under-represented among unskilled day-laborers (jornaleiros), whereas there is little question that the native-born blacks were over-represented in this category. Native-born blacks did a bit better than their overall ratio in terms of skilled occupations. Whites were under-represented in the leading trades of tailoring, carpentry and shoemaking while pardos did consistently better in these skilled trades than their overall average. The African males, who headed 2% of such households, maintained about this rate in all the major occupations in which they participated.

Among the female heads of non-slave households in Sabara, color was less of a marker of distinction (see table 8, panel 1). In all four color groups spinners (fiadeiras de algadao) were the primary occupation, followed by weavers (tecedeiras) and seamstresses (costureiras). Then came the women lavradores. The whites and pardas were more represented among the weavers than the spinners, and did well among seamstresses, while the reverse held true for native born black and African women. Among those working agricultural lands, the pardas in Sabara were, surprisingly, slightly under-represented and the native-born black and white women moderately over-represented. Equally unexpected was the high proportion of white women who were listed as beggars (mendigos) and invalids (invalidos).

When examining the slave households in Sabara the obvious fact is that color counts simply in terms of numbers. Among males who headed such slave-owning households, 60% were white (see table 9 panel 1) and even among women, whites represented 49% of all such female headed households (see table 10 panel 1). Also, while the relative weight of agriculture and artisanal activity was similar to the free households, whites were more committed to agriculture than were the colored and far less involved with the skilled trades (see table 9 panel 1). Also the high status professional and government jobs tended to be dominated by the white male slave owners, though there was some minor representation among the mulattoes who even counted a priest among their members. The mulatto men, however, did almost as well in the merchant class as did the whites.

Among the slave-owning women in Sabara, agriculture was more important in general than among their fellow women heads who owned no slaves (see table 10 panel 1). The Sabara white women, like their male slave-owning colleagues, [TABULAR DATA FOR TABLE 7 OMITTED] [TABULAR DATA FOR TABLE 8 OMITTED] [TABULAR DATA FOR TABLE 9 OMITTED] [TABULAR DATA FOR TABLE 10 OMITTED] also were involved more in agriculture and less in artisanal activity than were the colored women who owned slaves. But what is equally impressive is that these slave-owning women of whatever color were very committed to spinning cotton cloth, which was the dominant occupation among the colored and the second most important one among the whites. Clearly here gender played a vital part in marking the occupation of the heads of household, with lesser distinctions within these norms due to color.

The impact of gender and slave ownership found in the occupational distribution among Sabara heads of household can also be seen in the wealthier district of Campanha, despite the greater emphasis on agricultural activity. In Campanha, as in Sabara, lavradores were the most important occupational category among the male heads of non-slave households. But here again the whites were moderately over-represented in this category, and the pardos, blacks and Africans somewhat under-represented. In contrast the three colored groups were over-represented in the skilled occupations but also in the unskilled jornaleiro category (see table 7 panel 2 above).

Among women heads of non-slave households in Campanha, the textile trades accounted for over half the households. But as occurred in Sabara, white women were less represented among the lowest category of cotton spinners (see table 8 panel 2). in contrast, such white women did better as seamstresses, though here the pardas also fared well. Again, in the unusual category of beggars (mendigas), white women appear in unusually high numbers, something that cannot easily be explained.

Among Campanha's slave-owning families, the white male heads as usual were over-represented in the "farmer" category and among merchants and under-represented among the crafts. But pardos held their own in these categories, and of course were over-represented in the crafts and even did very well as tavern keepers. In short, here the same ratios we saw for the slave-owning households in Sabara occurred.

Though textile production, from spinning and weaving to lace making, was an important minority occupation among the women slave owners in Campanha, the stress among all racial groups was on agriculture. Because of the predominance of lavradores, there was no sharp variation by race in the various sub-fields of textile activities as there had been in Sabara.

From this detailed survey of individual occupations in these two communities, it is evident that color was influential in defining certain occupations, though less important than gender, the wealth of the individual region or the factor of slave ownership. But what was the impact of color on the actual ownership of slaves themselves? Did white slave owners own more slaves than colored ones?

Before examining this factor, it is worth noting that though these two municipalities differed in wealth, color and emphasis on agriculture or artisanal activity, the actual possession of slaves in their communities differed little. Not only was there a close similarity in the percentage of households that held slaves, but despite the differences in average size of slave holdings between the two communities (see table 11), the actual distribution of slaves among these slave-owning households - as measured by the Gini coefficient of inequality - showed little meaningful difference.(20)

Nevertheless some interesting differences can be noted when we examine the sex and color of the owners in the two communities (see table 11). Evidently men consistently represented a larger share of the slave-owning households than among the non-slave domiciles (75% of the slave households were run by men as opposed to 65% among the non-slave homes in Sabara; and these figures were 83% and 79% respectively in Campanha). Surprisingly, however, women and men held about the same number of slaves on average. In Sabara there was only a small difference of less than 2 slaves between the male and female averages, and in Campanha the females had a slight advantage over the males.

Table 11

AVERAGE SLAVE HOLDINGS OF HEADS OF HOUSEHOLD BY SEX & COLOR, SABARA
& CAMPANHA

 SLAVES OWNED/SLAVE OWNERS

 MALE OWNERS FEMALE OWNERS

 std. no. std. no.
COLOR mean dev. owners mean dev. owners

 I - SABARA

White 13.1 12.2 812 13.3 19.8 223
Mulatto 8.8 6.9 534 6.8 4.1 205
Native-born Black 7.8 8.0 14 4.9 2.4 19
African 0 3.5 1.9 4

Total 11.4 10.6 1360 9.9 14.6 451

 II - CAMPANHA

White 6.5 8.5 1152 7.2 12.1 228
Mulatto 4.3 7.4 119 3.4 4.8 35
Native-born Black 3.1 3.8 19 7.0 7.6 8
African 1.8 1.3 9 8.8 8.0 7

Total 6.2 8.3 1299 6.7 11.3 278

Note: The total for Sabara is 1,811 heads of slave-owning households
whose sex and color are known yields an average of 11 slaves per
household (11.7 std.dev.). There were another 23 owners whose sex or
color is unknown. For Campanha there were 1,577 heads of
slave-owning households whose sex and color are known, and these
held an average of 6.3 slaves (8.9 std.dev.). There were another 16
heads whose sex or color is unknown.

But the color of the slave owners did make a substantial difference. Whites were in general considerably wealthier than colored owners as measured in slaves owned per capita. In the two communities the colored slave owners, as would be expected, held a different share of the total of slaves. In the more colored Sabara, they headed 43% of such slave-owning units, while in Campanha they were a reduced 12%. Although more free colored men were slave owners than women, it is also true that free colored women in both municipios were more important among their same sex owners (both white and free colored) than were their men among all male owners. Moreover, the relative importance of free colored women who owned slaves among all slave-owning free persons was similar in both societies. Nevertheless in all cases white men and women held on average more slaves than their non-white colleagues. This is not surprising given that color indicated a previous condition of servitude and a lesser share of capital and education than was commonly available to the whites. Again, given Brazil's traditional three-part color scheme, it is no accident that the mulattoes (pardos) were the single most important group of free colored among the slave owners, though even in Campanha there were some African-born heads of family who owned slaves. Moreover even in the high end of the slave ownership category the mulattoes did rather well in both regions.(21)

It is evident from the analysis of these two regions in 1831, that the free colored population and the non-slave households in general were the poorer part of each community. Owning slaves was obviously a key indicator of wealth in these communities and probably meant the control over more and better lands as well. Though the majority of even non-slave households were in agriculture, these latter families tended to be far more involved in artisanal activity than households which contained slaves. There were also several demographic indices which suggest that the non-slave households were smaller in size and headed by younger and less married persons than was the case with the slave-owning units.(22)

The picture of the two-thirds of the households which owned no slaves, in both communities, is one of a more artisanal and less agricultural group than the slave owners. These non slave-owning households were made up far more of colored than of whites. Color even in these households has some effect, however, as whites in such households were usually more agriculturally based than their brethren of color and tended to be better represented in the more skilled occupations. Among women, clearly, spinning and weaving are almost as important as land ownership and agriculture, though again it is the white women who tend more to weaving than spinning and more toward seamstresses than the women of color. Though here too, it would appear that the pardas were usually closer to their white peers than they were to the other Afro-Brazilian women.

From this analysis of these two municipios then, what can be said about the free colored in early-19th-century Brazil, and especially of their position in Minas Gerais? They were clearly not a marginalized and isolated group denied access to resources that an open market economy could provide, as was the case in the southern slave states of the United States. A fair amount of economic mobility had already occurred for an important segment of these freemen and in many ways they participated in most of the occupations and household arrangements of their white neighbors. Moreover the comparison between the two municipios shows that the presence or absence of a white majority did not significantly influence this mobility. That prejudice still existed is evident from the color divisions within the freedmen category. Mulattoes were the dominant group among the free colored, and in most cases did better then their fellow freedmen of other color and nativity categories. But for these pardos the social and economic indices show that they were often not that different from the whites of their communities. They were obviously not to be found in the dominant government positions or other occupations representing supreme authority. But they were an integrated mass of workers who shared most of the characteristics of all the freeborn and white population among whom they lived. They were even a significant element among the elite group of slave owners. Brazilian freedmen were thus, a half century before the end of slavery, an important, competitive and integrated element in rural society.

Department of History New York NY 10027

Rua Curitiba, 832 30.170 Belo Horizonte, Minas Gerais Brazil

ENDNOTES

1. See Herbert S. Klein, "The Colored Freedmen in Brazilian Slave Society," Journal of Social History III, no. 1 (Fall, 1969), and African Slavery in Latin America and the Caribbean (New York, 1986), chap.10.

2. See Manuel Moreno Fraginals, Herbert S. Klein and Stanley Engerman, "Nineteenth Century Cuban Slave Prices in Comparative Perspective," American Historical Review 88, no. 4 (Dec., 1983).

3. See e.g. Celia Maria Marinha Azevedo, Onda negra medo branco. O negro no imaginario das elites seculo xix (Rio de Janeiro, 1987); and Lilia Mortiz Schwarz, Retrato em branco e negro. Jornais, escravos e cidadaos em Sao Paulo no final do seculo XIX (Sao Paulo, 1987).

4. Minas Gerais in size and economic activity shares much in common with Virginia. Yet in this most liberal slave state the freedmen were totally marginalized figures, legally denied access to mobility and property. See Luther Porter Jackson, Free Negro Labor & Property Holding in Virginia, 1830-1860 (2nd.ed., New York, 1968).

5. There have been, of course, a number of studies on the free colored. But these have dealt with special aspects of their lives, have discussed very small and unrepresentative samples, or have included them in the larger context of an undifferentiated poor free population. The religious life of the urban free colored in the northeast has been studied by A. J. R. Russell-Wood, The Black Man in Slavery & Freedom in Colonial Brazil (London, 1982); while the wealth of a sample of first-generation ex-slaves, also in an urban context, has been analyzed in Maria Ines Cortes de Oliveira, O liberto: o seu mundo e os outros, Salvador, 1790/1890 (Sao Paulo, 1988). The hardships suffered by both white and free colored poor under slavery have been treated in the classic studies of Maria Sylvia de Carvalho Franco, Homes libres na ordem escravocrata (Sao Paulo, 1969) and Laura de Mello e Souza, Desclassificados do Ouro. A pobreza mineira no seculo xviii (Sao Paulo, 1982). An interesting comparative analysis of their treatment in the urban criminal system is found in Leila Mezan Algrant, O feitor ausente. Estudos sobre a escravidao urbana no Rio de Janeiro - 1880-1822 (Petropolis, 1988). But in general there is still no systematic study of their role in the economy and society either in the urban, or more importantly, in the rural areas where they were predominantly located.

6. An overall view of the recent findings will be found in Stuart B. Schwartz, "Patterns of Slaveholding in the Americas: New Evidence from Brazil," American Historical Review 87, no. 1 (Feb. 1982). Good summaries of the materials for Minas Gerais will be found in Clotilde Paiva and Herbert S. Klein, "Slave & Free in 19th century Minas Gerais: Campanha in 1831," Slavery & Abolition (London), 15, no. 1 (April, 1994): 1-21; and for Sao Paulo in Francisco Vidal Luna and Herbert S. Klein, "Slaves and Masters in early Nineteenth-Century Brazil: Sao Paulo," Journal of Interdisciplinary History XXI, no. 4 (Spring, 1991): 549-573. Among the detailed studies of Minas Gerais are those by Clotilde Andrade Paiva, "Minas Gerais no seculo xix: aspectos demograficos de alguns nucleos populacionais," Iraci Nero da Costa, ed., Brasil: Historia Economica e Demografia (Sao Paulo, 1988) and "A natalidade de Minas Gerais no seculo xix: algumas hipoteses," Paper presented at the Conference on the Population History of Latin America (Ouro Preto, 2-6 July, 1989); Clotilde A. Paiva, et. al., "Estructura e dinamica da populacao de Minas Gerias no seculo xix," Unpublished CNPQ Research Report of 1990; Douglas Cole Libby and Mircia Grimaldi, "Equilibrio e estabilidade: Economia e comportamento demografico num regime escravista, Minas Gerais no seculo xix," Anais: VI Encontro Nacional de Estudos Populacionais (ABEP) (Olinda, 1988), III, 413-442; Clotilde Andrade Paiva and Douglas Cole Libby, "The Middle Path: Alternative Patterns of Slave Demographics in Nineteenth Century Minas Gerais," Paper presented at the World Demographic History Conference, Vera Cruz, Mexico, 1992. Older studies based on these unpublished manuscript censuses include among others, the works of Francisco Vidal Luna, Minas Gerais: Escravos e senhores (Sao Paulo, 1981; IPE/USP, "Ensaios economicos", no.8); Francisco Vidal Luna and Iraci del Nero da Costa, Minas colonial: economia e sociedade (Sao Paulo, 1982); Iraci del Nero da Costa, Minas Gerais: Estruturas populacionais tipicas (Sao Paulo, 1982); Iraci del Nero da Costa, Vila Rica: Populacao (1719-1826) (Sao Paulo, 1979; IPE/USP, "Ensaios economicos", no.1); Donald Ramos, "City and Country: The Family in Minas Gerais, 1804-1838," Journal of Family History 3, no. 4 (Winter, 1978) and his earlier "Vila Rica: Profile of a Colonial Brazilian Urban Center," The Americas XXXV, no.4 (April, 1979). The Sao Paulo province "mappas" have been used in the following studies: Iraci del Nero da Costa and Horacio Gutierrez, Parana: Mapas de habitantes, 1798-1830 (Sao Paulo, 1985); Horacio Gutierrez, "Demografia escrava numa economia nao-exportadora: Parana," Estudos Economicos (Sao Paulo) 17, no. 2 (1987), and in his "Crioulos e Africanos no Parana, 1798-1830," Revista Brasileira de Historia (Sao Paulo), 8, no. 16 (1988); in the work of Jose Flavio Motta, "A familia escrava e a penetracao do cafe em Bananal (1801-1829)," Revista Brasileira de Estudos Populacionais (Sao Paulo, ABEP) 5, no. 1 (1988): 71-101; and Armenio de Souza Rangel, "A economia do municipio de Taubate: 1798 a 1835," Estudos Economicos (Sao Paulo) 23, no. 1 (Janeiro/Abril 1993), 149-179.

7. It should be stressed that Minas Gerais and Sao Paulo were slave systems (as defined by Finley and other scholars) because of the concentration of slaves in the most capitalist parts of the local economy. But slaves themselves were not the most numerous workers in these regions. In the United States South, slaveholders in the Southern states represented only 31% of the total free population in 1850. Lewis Cecil Gray, History of Agriculture in the Southern United States to 1860, 2 vols. (Washington, 1932), I:482.

8. On the roughly 5% of the free households which were colored in 1830, see Carter G. Woodson, Free Negro Heads of Families in the United States in 1830 (Washington, 1925); as well as his study Free Negro Owners of the Slaves in the United States in 1830 (New York, 1924). Gray estimates that in 1860 in the U.S. South, only 3% of the non-slave owning population was made up of free colored. Gray, Agriculture, I:481-482.

9. The few exceptions to this neglect are the works of Eni de Mesquita Samara, "O papel do agregado na regiao de Itu," Anais de Museu Paulista (1977): 13-121, and Iraci del Nero da Costa, Arraia-miuda: um estudo sobre os nao-proprietarios de escravos no Brasil (Sao Paulo, 1992).

10. In the political geography of 19th-century Brazil a municipio was the equivalent of a North American county and defined both an urban district and its surrounding rural areas until the borders of the next municipio.

11. According to an 1836 tax list, Campanha had, 93 engenhos (sugar mills) and Sabara some 157, thus accounting for 14% of the province's total. They were also listed with 472 and 275 country stores (vendas) respectively, which together accounted for 16% of the total in Minas Gerais. Clotilde A. Paiva and Marcelo M. Godoy, "Engenhos e casas de negocios na minas oitocentista," VI Seminario sobre a economia mineira (Belo Horizonte, 1992), 38, tabela 1.

12. Because of changing definition of districts, it is difficult to compare Campanha and Sabara across censuses. In 1821 these two municipios had a combined population of 104,000 persons and accounted for 20% of the provincial total (Raimundo Jose da Cunha Matos, Corografia historica da provincia de Minas Gerias [1837] [2 vols.; Belo Horizonte, 1981], II: 45-50), while the census of 1835 showed the two zones with a total of 87,000 persons and just 12% of the total mineira population. In this latter census Campanha had almost 3,000 fewer persons than in 1831 and Sabara had some 15,000 more. Paiva and Godoy, "Engenhos e casa," loc.cit.

13. Our definition of color used here differs somewhat from the traditional three color designations (pardo, preto and africano) used in the censuses of the period, in that we break down "preto" into two groupings: those born in Brazil, or preto crioulo, and those simply listed as only preto which we consider as African-born, along with those actually given the "africano" definition. We and other mineiro researchers (conversation with Douglas Libby, 17 December 1991), have consistently found that in the manuscript mappas of Minas in 1831 and 1832 the terms preto and africano were often used interchangeably. Where no African identity was given the term preto crioulo was used for distinguishing blacks who were native-born. It was quite typical to have a slave listed as preto whose name was Jose Minas or Antonio Angola. To further test the validity of our usage, we compared demographic structures of all four colored groups, and discovered that the sex ratios, the small number of children and the ratio of elderly and the age heaping between 10-14 and 30-34 found for Africans was quite similar to that for pretos, and both differed sharply from pretos crioulos and pardos.

14. Though the numbers are often insufficient to produce meaningful ratios, it would seem that Native-born blacks and Africans did differ more from the mulattoes than mulattoes differed from whites.

15. Thus taking just one possible index of children 0-9 to women 15-49 shows the following results:

 Non-slave Household Slave Households

 Whites Total Colored Whites Total Colored

Sabara 830 970 1056 923
Campanha 1403 1126 1533 1004

Here, of course, the problem of defining the color of children and the lack of life tables for the various racial groups make it difficult to fully interpret these data. Thus, for example, while children of one white and one black or mulatto parent would be listed as mulatto and lost to the white group, so to would all children of Africans be listed as native-born blacks. For this reason we have decided to include all colored in one undifferentiated category (though of course the dominant racial group among free colored were the mulattoes). This would also mean that white parents are slightly under-represented since some of their births will go to the free colored category. Also without knowing the life expectancy of white and free colored women it is difficult to fully interpret the meaning of the differences in the child/women ratios.

The primary reason that ownership of slaves made little difference overall in rates of fertility has to do with the differing rates between whites and free colored. The free colored in non-slave-owning households had higher rates than those living in the slave-owning units. In turn, the opposite was the case for the whites, so that the two differences balanced each other out. But there is little question that there was a very sharp difference in total child/women ratios based on race.

16. In all the 19th-century "mapas" which have been examined, the occupation of slaves is not given. It has been assumed by almost all scholars working with these censuses that in the majority of the cases the slaves were employed in the same occupation as their masters, with the obvious exception of the liberal professions.

17. Jeremy Atack and Fred Bateman, To Their Own Soil. Agriculture in the Antebellum North (Ames, Iowa, 1987), 26. They estimate that of their almost 21,000-rural-household sample in the Middle West and Northeast, 44% were involved in non-farm activities.

18. Textiles were always the dominant activity among female headed-households in both communities, and among males in Sabara. In Campanha, however, male artisans were much more involved in non-textile activities, particularly woodworking, leather making and metals.

19. These generic titles of lavradores or agricultores give no indication of land ownership or socio-economic status, but most probably are related to agricultural production.

20. The GINI for Sabara of slaves held by masters and their relative shares of totals in each category was 0.599; and for Campanha 0.554.

21. In the actual distribution of owners by size of holding in slaves, the pardos in both communities did surprisingly well. In fact they did better in Campanha than in Sabara. In the former municipio some 3.3% of the pardo males held 21 or more slaves, as compared to 5.5% of the white male owners. In turn 79% of the pardo owners held fewer than 6 slaves, compared to 63% of the white owners. In contrast in Sabara, the relative ratios were 1.3% for the pardo males and 8% for the white males; while those holding fewer than 6 slaves were respectively 84% for the pardo males and 71% for the white male owners of slaves.

22. Only in terms of fertility (as measured in the child-women ratio) do we see a very interesting counter-result. For the free colored living in non-slave households, fertility was higher than among those living in slave-owning households. The cause for this difference is not immediately evident.