Slavery in a Nonexport Economy: Nineteenth-Century Minas Gerais Revisited Open Access

I

The history of slavery in nineteenth-century Brazil is usually associated with the coffee economy of the centralsouthern region of the country, particularly in the Provinces of Rio de Janeiro, São Paulo, and Minas Gerais. In the three decades following 1850 and the closing of the African slave trade, slaves were transferred from the southern and northeastern provinces to the coffee region on such a scale that the three major coffee provinces alone had almost two-thirds of the empire’s slave population in the year before abolition. Inside the coffee region itself, a similar movement took place. In the last years of slavery, about 90 percent of the slaves of São Paulo lived in the coffee districts and, of those, almost all were employed in the coffee fields. A similar, if less spectacular, trend occurred in Rio de Janeiro Province.

For Minas Gerais—which had by far the largest slave population of the empire throughout the century—very little research has been done. While most Mineiro historians seem to have been charmed by the splendor of the golden age, the majority of the historians of slavery in Brazil—Brazilians and Brazilianists—have been content with projecting onto nineteenth-century Minas findings about São Paulo and Rio de Janeiro.

The main lines of the existing interpretation can be summarized as follows: in the eighteenth century, owing to the gold and diamond rushes, a large contingent of slaves was gathered in Minas Gerais. As the mining boom faded, the slaves scattered throughout the countryside, retreating with their masters into the subsistence sector, where they became economically “redundant” or “underutilized.” In the nineteenth century, after a few decades of economic soul-searching, coffee became the dominant industry in the provincial economy, and the bondsmen were transferred en masse from the subsistence sector to the mushrooming coffee plantations. It has also been argued that slaves from the mining region of Minas Gerais were exported to the coffee fields of São Paulo and Rio de Janeiro and were an important source of labor in the formative years of the coffee industry in those provinces.¹

The findings presented in this article are at odds with most of the preceding contentions. We argue that in nineteenth-century Minas Gerais the association between slave and coffee culture, or indeed, between slavery and any form of export-oriented activity was not as pronounced as has been asserted.

Nonetheless, in spite of its almost absolute dissociation with the export sector of the province, as will be shown, the slave population of Minas Gerais was not just large, but expanding vigorously: from approximately 170, 000 in 1819 it grew to more than 380, 000 in 1873. By this latter date, Minas Gerais alone had, outside the grande lavoura and mining, far more slaves than were to be found in any slave society south of the Río Grande, with the exceptions of Cuba and Saint Domingue in their heydays. This contingent of nonplantation and nonmining slaves also outnumbered the total slave population of any other Brazilian province and even of all the ten provinces north of Bahia, plus Goiás, Mato Grosso, and Paraná combined.

The occurrence of a major slave system not associated with export-oriented activities, like the one described here, has been so far rejected by most historians of slavery in the New World. Likewise, the existence of “a crop, the cultivation of which demands combination and organization of labor on an extensive scale,” and close ties with international markets, or, in a phrase, the existence of the export-oriented plantation, has been claimed to be a sine qua non for the feasibility and survival of slavery as a labor system.² The purpose of this article is to attempt to explain, from a study of nineteenth-century slavery in Minas Gerais, this “apparent” paradox: the survival and continuous growth of a major slave system in a largely nonexport economy.

In section II we show that, in the nineteenth century, the mining industry was a minor sector in the provincial economy and, in particular, that its importance as an employer of slaves was very limited. The growth of the Mineiro slave population in the nineteenth century and the participation of the province in the slave trade are studied in sections III and V. Minas, the largest of the Brazilian provinces, was not a supplier of slave labor for other areas, but, rather, a heavy net importer during most of the century. In section IV the coffee sector of Minas Gerais is analyzed. We argue that the Mineiro coffee plantations’ being essentially an export enclave, with a very limited, if any, impact on the provincial economy, cannot explain the survival and growth of the slave system of the province. In section VI we show that the bulk of the Mineiro economy, where the vast majority of the slaves were employed, did not consist of plantations and was not export-oriented. Isolation from external markets, diversification, and self-sufficiency were its chief characteristics. Finally, in section VII we suggest an explanation for the prolonged attachment of the province to the slave system. We argue that there is nothing paradoxical about it, as soon as one realizes that the crucial condition for the survival of slavery is not the existence of the plantation, but, as E. G. Wakefield and others advanced a long time ago, the availability of free land.

II

By the beginning of the nineteenth century, Minas Gerais could no longer be characterized as a mining economy. The output of gold, which had reached a maximum annual average of 10, 036 kilograms in 1736-51, declined steadily throughout the second half of the century, to only 1, 883 kilograms per year in the period 1801-20.³ The mining towns were partially deserted and decaying; most gold diggings had been abandoned. The population was thinly scattered throughout the countryside, where it engaged mainly in cattle raising and subsistence farming. It has been estimated that, as early as 1776, not more than 20 percent of the labor force, free and slave, was employed in mining.⁴ Thirty years later the dissociation between gold mining and slavery was almost complete. Only 517 lavras remained in operation in 1814 and they were only pale shadows of what they had been a few decades before. Two-thirds of the lavras had ten slaves or fewer, and in most of them the slaves were only part-time miners, spending several months of the year in the cultivation of subsistence crops. W. L. Eschwege reported that between 1814 and 1820 the number of lavras in operation declined “extraordinarily” and estimated that in the early 1820s no more than 6, 000 workers, slave and free, were engaged in gold mining.⁵

The picture of the diamond district was as gloomy as that of the gold areas. Output had been declining steadily for decades, and in the last years of the colonial period, was only slightly over one-tenth of what it had been by the mid-eighteenth century. The number of slaves employed, which may have been as high as 5, 000 when the Real Extração began its operations, was, in the first two decades of the nineteenth century, well below 2, 000.⁶ It thus can be seen that the total number of bondsmen employed in mining at the close of the colonial period would hardly reach 10, 000, representing at most 6 percent of the slave population of Minas Gerais in the period (see Table II). Shortly after independence, the British started to take an interest in the mining industry. The first English company, Imperial Brazilian Mining Company, began operations in 1826 at its Gongo Soco mine in Santa Bárbara. During the empire, at least thirteen other British and a few Brazilian companies were formed to seek out and develop gold mines in Minas.⁷ Their efforts slowed the decline of the industry, but the overall trend was not reversed. The average annual output of gold of Minas Gerais declined to 1, 635 kilograms in the years 1821-60, and increased slightly, to 2, 174 kilograms in the period 1860-84.⁸ Above all, the presence of these companies did not reverse the ever-widening divorce between slavery and mining. Individually considered, each of these companies was a large employer of slaves, and the Saint John del Rey maywell have been the all-time largest slave-based enterprise in Minas Gerais. Taken together, and viewed against the background of the provincial slave population, however, the slave labor force of the English companies was utterly unimportant. The reason for these companies’ trifling impact on the employment of slaves is that they were few, used much more capital-intensive technology than had the old lavras, and were, with very few exceptions, short-lived, unmitigated fiascos. In the early 1840s, when four companies were active, the total number of slaves employed by them did not exceed 1, 750 (500 in Morro Velho, 300 in Cocais, 450 in Cata Branca, and 500 in Gongo Soco). By mid-century, with Cata Branca out of business, the number of slaves employed in mining probably did not exceed 2, 000. From then on, with the closing of Gongo Soco and the slow death of Cocais, the number certainly declined, in spite of the growth of Morro Velho and the formation of some new companies.⁹

We have no reliable data for the provincial slave population at midcentury. Our estimate, based on the annual rate of growth of the slave population between 1819 and 1873, gives 269, 550 slaves in 1850; this is probably low since the slave population in all likelihood increased faster before the close of the African slave trade than after it. The upper estimate of 2, 000 slaves employed by the English mining companies represents 0.7 percent of this probable underestimate of the slave population.

The 1872 census of the empire (which in Minas was taken in 1873) shows that, at most, 1, 625 slaves were employed in all kinds of mining in Minas Gerais. Of the 72 municípios of the province, only twelve had slaves in this occupation and, with few exceptions, they represented an insignificant percentage of slaves in a given município. In the entire province, the mining sector employed at most 0.44 percent of the slave population.¹⁰ In 1873 a provincial tax was levied on slaves directly or indirectly employed in gold mining. The tax returns show that the number of slaves in the sector decreased steadily, from 972 in 1872-73 to 644 in 1882-83, when they represented only 0.21 percent of the province’s slave population.¹¹ In 1887 the occupational analysis of slave registration did not even include the mining sector.

III

The playing out of the mines and the transfer of the population into other economic activities were natural results of the gold rush. In fact, the process would hardly have been remarkable had it not been accompanied by a strong growth of the slave population. It was at the depths of the crisis of the mining sector, in a period described by Celso Furtado as a “process of involution” unparalleled among Western economic systems, that the slave population of Minas Gerais reached an inflection point and started an expansion lasting until the final years of slavery.¹²

The expansion of the slave population closely followed the gold cycle in increasing at a very high rate until the late 1730s, then showing a slight decline until 1749, as the gold rush lost momentum. At some point during the third quarter of the century, gold mining and slavery parted company for good. By 1786, the number of slaves had more than doubled with respect to the 1749 population. The next two decades witnessed a decline in the servile population, but by 1808 the upward trend had resumed. By the beginning of the nineteenth century, Minas Gerais already had the largest slave population of the empire, and its fast growth in the following decades reinforced that position: between 1819 and 1872 the province’s share of the slave population of Brazil increased from 15.2 to 24.7 percent. In the same period, the Mineiro slave population grew at a rate almost two and a half times greater than the national average, and below only those of Rio de Janeiro and Rio Grande do Sul.

It is important to realize that the large and rapidly growing slave population of Minas in the early nineteenth century was not a heritage of the gold rush. These slaves were neither the remnants of those imported for the mining sector nor their descendants; they were the result of fresh imports. Direct evidence of the magnitude of slave imports in this early period is scanty and unreliable, but indirect methods permit us to gauge the flow.

It is clear from Table III that the import estimates are very sensitive to the rate of natural growth, a variable for which we have only sketchy information. It is safe to assume, however, that it was strongly negative. Mortality was certainly very high: with the African trade going at full speed, the proportion of African-born—that is, the portion of the slave population subjected to an entirely new disease environment—was kept very large. This alone would have sufficed to maintain a high death rate.¹³ There was, in addition, general agreement among contemporaries that working and living conditions for slaves were considerably harsher in mining areas than in agricultural ones. Fertility was bound to be low, if for no other reasons than the high morbidity rate and the tremendous imbalance between the sexes. Available estimates of the rate of natural growth of the Mineiro slave population in this period suggest that it was substantially below −.02. Goulart places it at −.03 for the years 1717-23, decreasing to −.04 in 1723-35. Eschwege estimated that around 1820 the rate was −.022 for the mulatto and −.039 for the Black slaves.¹⁴ Since the manumission rate was positive at all times, actual imports must have been above the estimated figures in all periods.

These estimates show that slavery did not follow mining in its decline. Imports actually increased after mid-century despite the crisis in the mining sector. Even in the period 1786-1808, when the slave population fell, they were maintained at a fairly high level.

IV

Most historians, while agreeing that the large and growing slave population of Minas Gerais cannot be accounted for in the nineteenth century by the collapse of the mining industry, have not hesitated to assign that leading role to coffee cultivation. The coffee industry has been commonly described as the activity that rescued the province, and indeed the empire, from an economic impasse, inaugurating a new era and giving a new lease on life to the institution of slavery. The exports of Mineiro coffee increased from an annual average of 472 tons in 1819-25 to only 7, 212 tons in 1846-51. It has been estimated that in the early 1860s there were no more than 300 coffee growers in the province.¹⁵ In the period between the end of the African slave trade and the final abolition of slavery, the expansion accelerated. From 1852 to 1870, the Mineiro sector was growing faster than any other coffee area in Brazil. In the next two decades it was surpassed, both in quantities produced and rates of growth, by the Paulista West, but continued to expand strongly, while the old coffee areas of the Paraiba Valley in Rio de Janeiro and São Paulo were stagnant, as they had been since mid-century. During the empire, commercial coffee production in Minas Gerais was limited to a very small portion of the province’s area, located in a narrow strip of land in the so-called Zona da Mata. In the last years of the imperial period, the coffee zone of Minas Gerais did not comprise more than 4 percent of the province’s territory.¹⁶ Using the returns from the toll stations scattered along the borders of the province, we can reconstitute with reasonable accuracy the regional origin of the Mineiro coffee exports for virtually the entire period under study.

In light of the data in Table V, it is surprising to find how widespread is the notion that the Zona Sul was an important coffee producer during the empire. This error has been repeated often in recent literature; and it is frequently compounded by saying that the region attracted large numbers of slaves to work its coffee fields.¹⁷

There are no direct statistics on the employment of slaves in the coffee sector. We have calculated this employment using the following procedure: we estimated a labor coefficient per unit of coffee produced and then obtained the total manpower requirements from the output data at different points in time, under the realistic assumption that changes in technology were negligible. Using data from several contemporary sources, we have estimated that in the early 1880s the average yield in the Zona da Mata was 36 arrobas per one thousand coffee trees. Data from 1883 for 153 plantations, located in eleven different municípios of Minas Gerais, show that the average number of coffee trees per slave was 3, 706. The two estimates above suggest that on average, one slave produced 134 arrobas of coffee per year. These estimates make clear that the role of the coffee industry was a relatively minor one, even in the final decades of slavery.¹⁸ There is no way in which the expansion of coffee can explain the survival, let alone the growth, of the institution. At the beginning of the period analyzed, the employment of slaves in coffee represented no more than .01 percent of the provincial slave population. Around mid-century the contingent employed in coffee could not have exceeded 2 percent of that population.

The availability of much more detailed data on the slave population in the 1870s and 1880s permits a better appreciation of the role of the coffee sector. The results presented in Table VII seriously undermine any attempt to equate slavery with the coffee industry in Minas Gerais, even in those final years. The comparison with Rio de Janeiro and São Paulo is striking. Let it serve as a plea against the inclusion, so often found in the literature, of Minas Gerais in “the coffee provinces.”

V

There has been no study to date of the role of Minas Gerais in the slave trade during the nineteenth century. Our survey of the contemporary literature and government documents failed to produce a single allusion to exports of bondsmen, or to the existence of surplus slaves in Minas in this period. The opposite—complaints about the high prices of slaves, their scarcity, and references to imports—was frequently observed. Such evidence leaves little doubt about the direction of the flow. Slaves were being sent from Rio de Janeiro (or rather, from Africa via Rio de Janeiro) to Minas Gerais, and not from Minas Gerais to the other provinces. In particular, the coffee plantations of the Paraíba Valley, whether in Rio de Janeiro, São Paulo, or the Mineiro Zona da Mata, were not being manned with slaves from central Minas. Between 1801 and 1851, some 1.5 million Africans were imported into Brazil, of whom more than 80 percent went to the southeastern provinces.¹⁹ This, and not the allegedly underutilized slaves of Minas Gerais, was the source of labor for the take-off of the coffee industry.

We have already shown the strong growth of the Mineiro slave population from the turn of the century to 1872. Using hypothetical rates of natural growth and assuming again that no manumissions took place, the minimum estimates of the net imports of slaves are those shown in Table VIII.

The number of slaves imported is, of course, very sensitive to the rate of natural growth, but the direction of the trade flow is not, over a wide range of rates. The province would have been a large net importer of bondsmen in both periods, even with a stable or naturally increasing slave population. In fact, in order to have been even a minor exporter of slaves, Minas Gerais would have to have been a true breeder, duplicating or even surpassing the unique demographic performance of the servile population of the United States.²⁰ The demographic ordeal of the Brazilian slave population in the nineteenth century does not seem to have been as drastic as that found in the Caribbean societies, but was no exception to it. There, too, the slave population was not capable of maintaining itself without continuous replenishments from the Atlantic trade. We have estimated that between 1819 and 1872 the annual rate of internal growth of the slave population was −9.32 per thousand for Brazil as a whole. This figure is consistent with the results presented by Robert Slenes, who, using an entirely different approach, arrived at the range of 0 to −15 per thousand per year from 1850 to 1870.²¹ There is no reason to suppose that the Mineiro rate of decline was significantly lower than that of the rest of Brazil. Actually, because Minas was a far heavier importer than most provinces, its overall slave death rate may have been higher than elsewhere, other things being equal. It is not unreasonable, therefore, to suppose that Minas Gerais imported more than 400,000 slaves in the first seven decades of the nineteenth century. If, as seems to be the case, almost all these imports occurred while the African trade was still open, then Minas is very likely to have been the largest provincial importer of Africans in the nineteenth century, with a share of at least 30 percent of the total Brazilian imports. After the closing of the African trade, in the early 1850s, the interprovincial trade gained momentum. The western Paulista coffee region was by far the leading importer: between 1854 and 1874 São Paulo imported almost as many bondsmen as during the previous thirty-five years. The Fluminense coffee sector remained stagnant throughout the remainder of the empire and was, owing to its participation in the African trade, well stocked with slaves. The slave population of Rio de Janeiro remained practically unchanged between 1850 and 1872.

Because of the lack of data for the 1850s, it is difficult to assess the role of Minas Gerais in the interprovincial slave transfers from the end of the African trade to the census of 1872–74. On the one hand, we find in this period for the first time references to slave exports from the province. Sir Richard Burton, for example, was told in São João del Rei in 1867 that the slave population was rapidly decreasing, the hands having “mostly been sold off to the agricultural districts of Rio de Janeiro.’’ In Diamantina he heard that the slaves had been “traded off to the coffee growing provinces,” and from Januária he reported that “for some time past the serviles have been traded off to Rio de Janeiro.” In 1865, Dr. J. McFadden Gaston, a South Carolinian visiting São Paulo as a scout for Confederate emigration, learned from one planter that gangs of slaves from Minas could be contracted to lay out new coffee groves, and, from another, that Mineiro slaves could be hired at rates ranging from 40 to 60 dollars a year.²² Other data, however, suggest that the involvement of Minas Gerais in the interprovincial trade before 1872 was very slight. The census registered the place of birth of the slaves, and from these data we can sketch the pattern of the trade up to 1872.

The place-of-birth data confirm the minor involvement of Minas Gerais. In 1873 there were 11,560 Mineiro-born slaves living outside the province (3.3 percent of all Mineiro-born slaves), while 8,578 slaves born in other provinces were living in Minas Gerais (2.3 percent of all slaves living in Minas Gerais).²³ There were about as many Mineiro-born slaves in Rio de Janeiro as Fluminense-born slaves in Minas. The contingent of Mineiro bondsmen living outside Minas was well distributed among the provinces with which Minas had common borders. The impression these data give is that far from there being a unilateral flow toward the coffee areas, what was taking place within the Center-South was a rather sleepy border trade, through which a limited number of bondsmen was filtering across the provincial borders. In the case of Minas Gerais, a considerable part of the transfers seems to have been the result of migrations of small slaveowners who moved to unreclaimed districts beyond the coffee zone in São Paulo or to Goiás, Mato Grosso, and Espírito Santo, and settled into subsistence farming and stock raising.

The interprovincial slave trade had a strong sex-and-age bias, being heavily concentrated in young to prime-age males. Using a relatively large sample of data, Sienes has shown that 85 percent of the slaves sold in the largest slave market of Brazil in the 1860s and 1870s (Campinas) were between 10 and 39 years old.²⁴ It would be reasonable, therefore, to expect to find this bias reflected in the sex-age structure of any geographic unit heavily involved in the traffic of bondsmen. It turns out that, according to the census, the proportion of males between 11 and 40 years was, in Minas, higher than in any other province. Considering both sexes, Minas ranked second from the top in this ratio. From data relating to the province itself, we learn that the age distribution was remarkably uniform across the regions: the percentage of the 11-to-40 age group varied only between 57.7 and 63.1, with the exception of the Triângulo zone, where it was 49.5. In the Zona da Mata, the proportion was 59.9, ranking it sixth among the nine regions. These results strongly suggest that Minas could not be losing significant numbers of slaves to its neighbors and that there was no major redistribution of the slave population within the province in the years preceding the census. It is also very unlikely that Mineiro slaves were being hired out to other provinces in significant numbers: in 1873, according to the census, only 2,876 bondsmen were absent from their parishes of residence. Of these, 1,776 were males, 90 percent of whom were in an age bracket compatible with sustained field work. Keeping in mind that many of these slaves were probably hired out to other parishes within the same município, or to other Mineiro municípios, and, also, that their absence might not have been related to hiring at all, it is evident that the potential number of out-of-the-province hirings was insignificant.

For the 1870s and 1880s, slave population data are available município by município, which allows us to look more closely at distribution of that population within the province. In 1873, the distribution of the slave population by regions in Minas closely followed that of the free population. The concentration in the plantation area, the Zona da Mata, was slight. Only in the three regions north of the nineteenth parallel was the percentage of the slaves significantly lower than that of the free population. These areas, with the exception of some portions of the Jequitinhonha River Valley, were frontier zones or sparsely settled districts, many of which remained so until the present century. The data show an increasing concentration of slaves in the Mata toward the end of the period. As a procedure to gauge the interregional flows of slaves, however, the simple comparison of the slave populations of the regions at the different dates is clearly insufficient. Indeed, because it does not allow for the facts that slaves died and were manumitted, this comparison is downright misleading, with a built-in bias toward the exaggeration of the losses of the exporting zones and the understatement of the gains of the importing areas. We have estimated the net transfers of slaves of each município in the periods 1873–80, 1880–84, and 1884–86, using a methodology similar to that employed in the computations in Tables III and VIII. For the years 1880 to 1886, when the provincial slave population was closed for all practical purposes, the procedure is equivalent to the demographers’ census survival method: the rate of internal growth (changes due to death and manumission) observed for the whole province was assumed uniform across all municípios. For the period 1873–80, we used a death rate (identical to the rate of natural decline, because of the “Free Womb” law of 1871) of 23 per 1,000 per year, as estimated by Sienes for some estates in central São Paulo during the same period. Manumissions, estimated as 7,700 in the entire province during the seven-year period, were assumed to be proportional to the initial slave population of each município.²⁵ These estimates reveal a pattern of trade much more complex than that suggested in the literature. It is true that the Zona da Mata was the leading importer, but the pattern cannot be described as one in which the coffee zone was depleting the rest of the province of its slaves. Most regions were net importers of bondsmen. Of the thirty-four importing municípios, only six were coffee districts; in a few others along the São Paulo border, coffee cultivation was just beginning; and the remainder were scattered outside the coffee region.

Contingency analysis shows that neither mining nor coffee culture can explain the pattern of the slave trade in this period. The municípios were classified as importers or exporters of slaves, on the one hand, and as mining or nonmining, on the other. Three different definitions of a “mining” município were tried: those with gold lavras in operation in 1814, those with slaves employed in mining in 1873, and those with anyone employed in mining in 1873. The resulting contingency coefficients were, respectively, 0.25, 0.14, and 0.18. Classifying the municípios as coffee or noncoffee producers versus importers or exporters of slaves, yields a contingency coefficient of 0.19. The aggregation of the net transfers of the municípios shows that the province as a whole was a very slight net importer of bondsmen in this period. This result is consistent with another body of evidence available for those years: the município entry-exit records. After the slave matrícula of 1873, it was required by law that alterations in the slave population, due to death, manumission, or change of residence, be reported to the authorities. According to these records, between September 1873 and June 1881, 64,718 entries and 58,782 exits were reported in the municípios of Minas Gerais, leaving a positive balance of 5,936 slaves entering the province in the period. Both entries and exits were known to be underreported, and it is likely that the underreporting of exits was somewhat greater than that of entries. The actual net number imported, therefore, may have been smaller than the one recorded.²⁶ The procedure employed cannot be used to determine the number of slaves imported and exported. It can only yield the net balance of these flows, which is, of course, consistent with an infinity of pairs of quantities imported and exported.

It is thus possible, and indeed highly probable, that the pattern of trade was more roundabout, and the gross involvement of Minas in the interprovincial traffic somewhat heavier, than the net figures suggest. The regions of the province were not economically integrated and the communications between them notoriously poor. In fact, it would have been much more natural for some regions to do most of their trade with other provinces than with other regions of Minas. The São Francisco Valley had had closer commercial ties with Bahia and Pernambuco than with the rest of Minas since colonial times, as did the Paracatu zone with Goiás, and the Sul with São Paulo and Rio de Janeiro. The Mata, in particular, was strongly connected with Rio de Janeiro. It was linked by rail to the imperial capital from the late 1860s and exported all its coffee and a variety of other commodities to that market.

The claim, made in 1885 by future President Prudente José de Morais, “that half or even more of the slaves in Minas Gerais and São Paulo have been acquired from the North since 1871,” is certainly an enormous exaggeration, but the inflow was large enough to worry the provincial lawmakers. In December 1880, the Mineiro legislature, fearing that the excessive concentration of slaves in the Center-South might undermine the commitment of the rest of the empire to the institution, decided to impose severe restrictions on further imports of bondsmen from other provinces.²⁷ Rio de Janeiro and São Paulo enacted similar measures in December 1880 and January 1881. The 1880 law effectively ended the importation of slaves into Minas; no revenue was ever generated from the import duty. The entry-exit records suggest that the province became a slight net exporter of bondsmen; the number exported, however, seems to have been very small. For all practical purposes, the Mineiro slave population became “closed” after 1880.²⁸

Between 1880 and 1884 there was a slight intensification of the intraprovincial slave trade. The coffee zone had been deprived of its external sources of supply, whereas the Paulista and Fluminense markets were closed to the exporting areas. Most of the legal trade now had to take place within the provinces. The entry-exit data indicate that the number of slaves transferred among the Mineiro municípios was on average a little higher than in the previous seven years. The pattern of this trade was not, however, significantly different. The Mata continued to be the leading importer, but, with the exception of the Triângulo, the losses of the exporting regions were small relative to their slave populations. At the level of aggregation used in Table XI, the estimates seem to show a very skewed pattern of trade, with the coffee area siphoning off slaves from nearly all other regions. The analysis at the município level shows, again, that it was more complex than this. The coffee districts resorted to other parts of the province more intensely than before, but so did many noncoffee municípios. Of the twenty-nine net importers of slaves, only nine were coffee municípios ; seven were established producers in the Mata, while in two others in the Sul, the industry was incipient. Again, there is no association between being or not being a coffee producer and the sign of the net trade. The contingency coefficient is equal to 0.18.

VI

We have argued that nineteenth-century Minas Gerais cannot be described either as a mining or as a plantation economy. We have shown that both sectors played relatively minor roles in the economic life of the province and that they cannot “explain” the large slave population of Minas. What, then, constituted the provincial economy? What occupied its vast contingent of bondsmen? What were the economic activities of its large free population?

The provincial economy consisted mainly of the cultivation of basic staples and stock raising for local consumption or for sale in the neighboring towns. Small manufacturing, also for local markets, is often mentioned. In very few instances the production was directed toward markets located outside the province. Among these, besides coffee, the most important was the short-lived boom in the exports of raw cotton in the first two decades of the century. Cattle, hogs, tobacco, and cotton textiles were exported in fairly large quantities during most of the period, but all these industries were primarily home-based, and exports were clearly secondary to internal consumption. Some areas of the province supplied urban markets outside Minas—especially the city of Rio de Janeiro—throughout the century with basic foodstuffs, notably dairy products, pork, and beef; but these exports were of a vent-for-surplus nature and the flows were quite small in comparison with the provincial output of these commodities. Furthermore, the units in which they were produced were what contemporaries called pequena lavoura. Their technology was extremely backward and their output highly diversified. It often included “colonial” articles traditionally associated with the export plantation, such as sugar and cotton, but in Minas these items were grown almost exclusively for local consumption. The Mineiro fazendas, although frequently large landholdings, sometimes with large numbers of slaves, were not grande lavoura, or plantations. They lacked most or all of the defining features of a plantation: the concentration on one crop, the factory-like discipline and management, and, above all, the export orientation and mentality.

To contemporary observers, it was clear that Minas was neither an export nor a plantation economy. Louis François de Tollenare, who lived in Brazil from 1816 to 1818, went straight to the point as he observed that “the most interesting province is Minas, which has one million inhabitants, supplies few commodities for the trade, but produces a lot for its internal consumption .... It is reckoned that it is not just the extraction of gold that keeps all that people busy, but rather the pequena lavoura, which we Europeans despise, with eyes only for sugar, cotton, and coffee among the products of America.”²⁹ For the same period, one Augustin Francois César Prouvençal de Saint-Hilaire, a French botanist, made the same remarks and explained that it was “not profitable to cultivate corn [as well as sugar and coffee] beyond the quantity required for the consumption of their own households ... as long as they have to be transported on mule back.”³⁰ As late as 1875, the provincial deputy Batista Pinto agreed with the explanation advanced by Saint-Hilaire at the beginning of the century. After saying that “Minas is prodigiously fertile ... in all municípios, coffee, cotton, and sugarcane produce bounteously .... Nonetheless, we don’t export one single arroba of sugar, one single liter of brandy. In fact, we don’t export anything but some coffee from the rich municípios of the Mata, which are in contact with the province of Rio de Janeiro …” He added that the reason was that “the cost of transportation absorbs the whole price.”³¹ Also in the 1870s, in his description of São José, at the heart of the province, James Wells observed that “there is really no exportation of surplus production, as nearly the whole of the produce is locally consumed …”³²

We do not mean to suggest that Minas Gerais did not have a plantation sector—obviously it did, and a large one, for that matter. After all, the province was one of the largest coffee producers of the empire, and this coffee was grown on plantations. The Mineiro coffee plantation was very similar to its Paulista and Fluminense counterparts. They had about the same size slave labor force, the same number of coffee trees, and used the same technology.³³ Viewed against the background of the entire province, however, the plantation sector of Minas Gerais shows some important differences: unlike those of Rio de Janeiro and São Paulo, it was small in comparison with the provincial economy; it was concentrated in a small area; and, most important of all, it had but a limited impact on the economic life of the rest of the province. Even when the coffee boom reached its apogee, the fazenda, the sítio, the subsistence roça, and the cattle ranch, not the plantation, formed the core and the bulk of the economic life of Minas. Reing essentially inconspicuous and leaving few records for posterity, however, this agricultural complex has been ignored by most historians.

Because of the relatively small size of the Mineiro export sector, some commentators of earlier versions of this paper have manifested a skeptical view of the province’s capacity to pay for the large imports of slaves that result from our estimates. One must bear in mind, in the first place, that a particular province or region does not have to produce by itself the foreign currency necessary to finance its imports. In fact, Minas Gerais might not have had an export sector at all and still have bought large numbers of slaves in the Rio market, paying for them in mil-réis, as was the case during the entire period in question. To finance these investments, all that was needed was a surplus in the sectors that were importing the slaves—a surplus not necessarily generated by exporting the output.

Besides, it is important to note that, contrary to what could be argued by those who associate the occurrence of slavery with an export-oriented economy, the self-contained nature of the Mineiro economy did not threaten the existence of the large slave system of the province. Quite the contrary; it was exactly its almost absolute self-sufficiency that permitted the survival of the slave institution in Minas Gerais. If, on the one hand, the province consumed internally most of its production, exporting only a small surplus, on the other hand, it did not depend on the exterior for much of anything, having to import only a few items, such as gunpowder, some hardware, wine, salt, and slaves. With such a restrictive composition of imports, it is not surprising that even with the lowest per capita export among all Brazilian regions (as is shown in Table XII), the capacity of the province to import was sufficient to maintain and even to increase its slave population.

In his description of some fazendas in Itacambira and Montes Claros, Saint-Hilaire observed that they had to buy only salt from the outside.³⁴ When describing the economy of the Upper Rio Grande region in the southern part of the province, the same author observed that although more market-oriented than in the north, it was still very much self-contained, with only a small part of the output exported, the proceeds being used to pay for the things they could not produce, mainly slaves, salt, and a few manufactures.³⁵ In 1867, Sir Richard Burton described the “typical Mineiro estate,” saying that “these fazendas are isolated villages on a small scale. They supply the neighborhood with its simple wants… they import only iron… , salt, wine and beer, cigars, butter, porcelain, drugs and other ‘notions’,”³⁶ and, in the 1870s, James Wells observed about the small fazendas of Minas that “the little surplus of production of the farm, or the occasional sale of a bullock, affords sufficient means to purchase the few homely necessities that the farm does not produce… .”³⁷ “The fazenda is something between a family and a kingdom,” wrote Pierre Denis in the early twentieth century, adding that “It is in Minas that once existed, and perhaps one can still find, those fazendas where a family lives in plenty but without affluence, exporting little, demanding few things from the rest of the world… ,”³⁸ But perhaps the best observation on the province’s remarkable capacity for isolated life was made by Saint-Hilaire when he wrote, “If there is a region that could ever do without the rest of the world, this would certainly be the province of Minas.’’³⁹

From the very onset of settlement, Minas Gerais “presented features which contrasted markedly with the plantation and latifundia economy of the littoral. The gold strikes created for the first time hundreds of miles from the coast new markets for products such as brandy and sugar, which until then had only been exported. Very soon the miners themselves had begun to produce locally certain basic commodities essential to the provisioning of the expanding mining camps.” The diversification of the regional economy was thus concomitant with the expansion of mining and was, at first, a result of the demand generated in the urban and mining sectors. In some places the establishment of farms and cattle ranches even preceded the mineral discoveries.⁴⁰ The crisis of the mining industry, in the third and fourth quarters of the eighteenth century, strongly spurred the diversification process and steered the economy toward an increasing isolation from external markets, as the decline in gold production steadily decreased the capacity to import. The increasingly restricted access to imported goods set in motion an import-substitution process of far-reaching consequences for the Mineiro economy. Activities that had been ancillary to the mining industry became the mainstream of provincial economic life. The process was accompanied by the dispersal of the population throughout the countryside: the main focus of economic activity shifted from the towns and villages to the farms, cattle ranches, and subsistence roças, which dotted the vast territory of the province. Toward the end of the eighteenth century, Minas Gerais had become self-sufficient in food production and it began to export a small surplus, mostly to the urban market of Rio de Janeiro. The import bill of 1818–19 shows that the captaincy did not import any food staple, only a few luxury food items and salt. In the manufacturing sector, import substitution was much more limited, yet the production of local substitutes, especially of cotton textiles and iron products, was large enough to worry the metropolitan authorities “that soon the inhabitants of the captaincy would become independent from those [manufactures] of the kingdom.”⁴¹

It is fundamental to realize that the transformation through which the Mineiro economy passed after the apogee of mining was not an interim arrangement, not an intermission for economic soul-searching, not a temporary stupor that was ended by the rise of the coffee sector. The main lines of the economic structure that was taking shape by the early 1800s remained essentially unchanged, or were reinforced, in most parts of the province as the century unfolded. In many of its regions, this structure outlived the empire and persisted well into the present century.⁴² The desperate and often unrealistic attempts of the administration to reintegrate the province in the export trade reflected the “colonial” mentality of the time, and, without exception, resulted in complete failures. Minas Gerais remained an overwhelmingly nonexport economy, and had the lowest level of per capita exports among all regions of Brazil.⁴³ The coffee sector provided 69 percent of the Mineiro exports in the period 1850–89, but what this figure actually underscores is the lack of importance of the export sector in the economic life of the province as a whole. The noncoffee portion of Minas, which comprised, in the last decades of the empire, about 96 percent of the territory, 79 percent of the slaves, and more than 80 percent of the free population, generated substantially less than 30 percent of the exports (in nominal values).⁴⁴ The coffee area was from the beginning, and remained throughout the empire, an export-plantation enclave in the province, which had little, if any, impact on the economic structure of the Mineiro heartland.

The behavior of the noncoffee exports is, for this reason, much more representative of the province’s economic reality. The per capita value of those exports was extremely low and declined sharply in real terms throughout the century. In the cases of several commodities, like cheese, hogs, cotton, sugar, pork and bacon, and cotton textiles, the total volume exported decreased, especially after mid-century.⁴⁵

Even though agriculture was the major field of employment of slaves in Minas, they were also used in a variety of other occupations. In some of them, such as cotton textile and iron manufacture, a considerable number of slaves was employed, many times more than in the mining industry. The large provincial output of cotton was consumed internally. Everywhere in the province cotton cloth and yarn were manufactured at home for domestic and local consumption, and a considerable surplus was exported to the neighboring provinces, where the coarse homespun, known as “Minas cloth,” was appreciated for its weight and durability. In 1827–28, the cotton cloth produced and consumed in Minas was estimated at 5.3 million meters, while another 2.1 million were exported to Rio de Janeiro alone. This gives a provincial output of at least 7.4 million meters, almost twice the total production of all Brazilian textile mills forty years later, in 1866.⁴⁷ Slaves were generally employed in the Mineiro textile industry. In 1873, the census recorded 62,873 free persons and 8,031 slaves working in the textile industry in Minas, scattered throughout a large number of parishes. The Mineiro textile labor force represented 49.8 percent of the free and 60.9 percent of the slave textile workers recorded in all Brazil.⁴⁸

The iron industry was an activity almost entirely based on slave labor. When Eschwege arrived in Minas in 1811, he noted that a large number of estates had their own shops, where, using the skills of African slaves, iron was obtained and worked into mining and agricultural tools, horseshoes, and a variety of other items. In the 1810s, owing to the efforts of Eschwege and of the Intendente Câmara (Manoel Ferreira da Câmara Bethencourt e Sá), a period of rapid development was inaugurated in this industry. These men established the first large-scale iron works, which served as models for several smaller ones. In 1815 it was estimated that upward of 70 percent of the iron previously imported was being locally produced.⁴⁹ Another large mill was established in 1831 by Jean de Monlevade in São Miguel de Piracicaba. It was soon producing hoes, axes, knives, domestic utensils, as well as parts for sugarmills, sawmills, and stamp-heads for the Morro Velho mine.⁵⁰ There were 84 ironworks in Central Minas Gerais in 1853, “not counting the numerous small shops,” employing about 2,000 workers, slave and free, and producing more than 2,000 tons of iron per year. The reports of the provincial presidents mention 120 works in 1864, and between 100 and 150 in 1873.⁵¹ Slaves were extensively employed in all the works, large and small. Eschwege initially attempted to employ only free men, but soon turned to slaves, whom he trained and employed in all kinds of jobs, menial as well as skilled. Câmara and Monlevade also used large numbers of them: upon the death of the latter in 1872, his estate included no fewer than 200 bondsmen, many of whom were skilled iron makers. As late as 1883, a professor of the newly founded school of mining engineering reported that most iron makers still relied on slaves.⁵²

The 1872 census provides the only systematic and reliable body of data on slave occupations in Minas Gerais, or, indeed, in any Brazilian province. A comparison between Minas and the plantation provinces of Rio de Janeiro and São Paulo shows remarkable differences in the occupational distributions of the slave labor force. The Mineiro distribution reflects clearly the more diversified economic structure of the province. The slaves were not nearly as concentrated in agriculture; and other occupations, like crafts and manufacturing, had a substantially larger share in Minas than in the coffee provinces. The high concentration in domestic service suggests that economic life was more centered around the household and less oriented toward production for the market.

VII

Ulrich B. Phillips once wrote that the plantation system “was less dependent upon slavery than slavery was upon it; and the plantation regime has persisted on a considerable scale ... in spite of the destruction of slavery.’’⁵³ He was referring specifically to the South of the United States, but identical views have always been widely held with respect to the rest of the New World as well. Most historians have grown so used to the close relationship between forced labor and the plantation system that the simple mention of New World slavery brings to mind the Caribbean factory-island, the cotton plantation of the Old South, and the coffee fazenda or the sugar engenho of Brazil. On the other hand, the cultivation of cereal crops, stock raising, and manufacturing activities, oriented toward the domestic market, such as those found to prevail in Minas, have frequently been considered incompatible with forced labor. The peculiar slave economy of nineteenth-century Minas Gerais is not a curiosum, a perverse or short-lived case in some out-of-the-way place. It was the major slave system of Brazil, and one that survived and expanded in areas where the plantation system never existed. The slaves were employed in a variety of nonplantation jobs, and the nonplantation areas were net importers of bondsmen up to the very eve of abolition. Far from being “eager to see its end,” the Mineiro slaveowners actively clung to the institution: in the 1870s and 1880s, Minas held the dubious honor of having the lowest manumission rates in Brazil.⁵⁴

Why did Minas cling so tenaciously to slavery, for so long? We do not think it was sui generis, a perversity that defies explanation. In fact, the Mineiro case challenges only the theory that the export plantation and the slave regime must always have marched hand in hand, and that the latter could not survive without the former.

The survival and growth of a large slave system in Minas Gerais shows that the crucial condition for the existence of slavery was not the production of plantation crops for exportation, but, as Wakefield, Nieboer, Domar, and others have argued, the existence of free land.⁵⁵ There was no voluntary supply of wage labor in Minas Gerais because the agricultural frontier was wide open and the free peasants had plenty of land from which to obtain an independent subsistence. In this context, a class of nonworking landlords could have survived only by continuing to rely on coerced labor.

Throughout the century there were plenty of people in Minas Gerais, but, to the exasperation of would-be employers, braços for hire were chronically in short supply. The free peasants did take on occasional jobs, such as land clearing, or, by the more peripatetic ones, the work of cowboy or tropeiro; they would not be persuaded, however, to engage themselves permanently to work for others, “with constancy and in combination.”⁵⁶ There was no voluntary supply of labor because most lands had not been privately appropriated, or, as Marx put it, the process of primitive accumulation had not been completed. A large portion of the vast provincial territory was then in a state of absolute wilderness. The population was sparse: even in the settled parts of the province, in the late nineteenth century, one could travel for days without meeting a soul or seeing a house. A large portion of the land did not have private owners; even in the appropriated parts, ownership was sometimes more nominal than effective, because property rights often were not enforceable.

Land was so plentiful as to be exceedingly cheap in many settled regions. Saint-Hilaire estimated that good lands in the São Francisco Valley cost about one-three hundredth of the price of the same area in the region of Bologne, in France, which was “famous for its barrenness.’’⁵⁷ In the 1870s, Wells reported from northern Minas Gerais, that “land can be obtained here in… almost any quantity and quality, forest, grass or scrub, for a mere bagatelle.”⁵⁸ Beyond the frontier, land was cheaper than that: it was free. Another observer remarked, still from northern Minas, in 1879: “In fact, land has very little value here. The proprietors give it free to the poor peasants for their plantings, and the latter can also obtain as much as they want by clearing the forests.”⁵⁹ As late as 1920, President Arthur da Silva Bernardes described Minas Gerais as a state “short on labor and long on land.”⁶⁰ All a peasant had to do was to move into an unclaimed spot, build his hut, and plant his roça.⁶¹ It was practically impossible to prevent squatting, even in appropriated lands. The peasants could be expelled from their plots, but “who is to stop these inhabitants of the forest from going a few leagues farther on, from disappearing into the forest and making their clearing when the fog comes, so that the police won’t see the smoke, and then remaining there two years planting and contenting themselves with their manioc bread?”⁶² In 1845, it was estimated by Francisco Diogo Pereira de Vasconcelos that 45 percent of the area of the province was held by squatters.⁶³ The open frontier included vast areas not only within Minas, but also in the neighboring provinces, such as Goiás and Mato Grosso.⁶⁴ Every peasant family could take a piece of land, cultivate it, and subsist without having to sell its labor to others. In the absence of some form of coercion, the economy could only (as its free sector did) collapse into a constellation of family-sized independent units. Nobody—planter, farmer, industrialist, or miner—would have a servant “to make his bed or to fetch him water from the river," let alone to till his fields from dawn to dusk, day in and day out. “Why would a free man submit to work all year around for a stranger, if he lives in a country… where any land can be cultivated and nobody has to work more than four weeks to obtain his subsistence, without giving up his freedom?”⁶⁵ This statement was made in the 1820s by Eschwege, but this reasoning remained true until the end of the century. In the 1870s and 1880s, every free peasant still possessed what struck one observer as “a veritable three acres and a cow.” “Every one, no matter how poor, has somewhere a patch of Indian corn.” This was true even for village dwellers, for among them, too, “almost every household has its roça ... in the near neighborhood, a pig, and fowls in the yard.”⁶⁶ The peasants were found “destitute of even an iron spoon or fork,” observed James Wells, “or means for purchase; their little labor is expended in raising a few vegetables, fishing, and in building a poor hut .... Yet withal, they are the most independent of all peoples, proud of their right to do nothing, and they do it most effectually.”⁶⁷

Free land, and its implication—the economic independence of the juridically free peasant—explain the continued attachment of the propertied class to slavery. It meant that the slave continued to be the only resource available to all those, planters or others, not willing to earn livings by the sweat of their brows.

We thank Professor Joseph L. Love for his many suggestions and encouragement, as well as Professors Amilcar V. Martins, Francisco Iglésias, Maurício B. de Paula Pinto, Peter Eisenberg, Eul-Soo Pang, José Murilo de Carvalho, Augusto César de Oliveira, Mr. Hélio Gravata of the Arquivo Público Mineiro, and Douglas C. Libby for kindly making available to us some of the materials used.

Author notes