While agreeing with the main thrust of our article— that nineteenth-century Minas Gerais, with a large and expanding slave system not identified with the export plantation, constitutes a highly unusual case in the history of modern slavery—Professors Sienes, Dean, Engerirían, and Genovese offer a number of penetrating comments that invite a response. We will, in our reply, focus only on those aspects we consider most important to our conclusions, leaving unanswered, for the time being, several relevant questions raised by them.1

Our findings regarding the slave trade to Minas have convinced Sienes, who also indicates that the Mineiro exports could have paid for the imports of bondsmen. He questions, however, the accuracy of the underlying population data. In the opinion of a leading Brazilian demographer, “the inquiry of 1872 was a true census, conducted in a systematic and uniform manner … over the entire territory of the empire … .”2 The 1819 estimate was based on church records and is generally regarded as the best available for that time—almost a true census, according to Souza e Silva, a nineteenth-century expert. We know of six other estimates of the Mineiro slave population for this early period: four of them yield net imports within 2 percent of ours; the other two result in figures that are only 6 and 14 percent lower, respectively.3

Warren Dean objects to our finding of slave imports by the nonexport areas and suggests that “while the slave population of the coffee-producing Zona da Mata grew principally through importation, the slave population of the noncoffee region did so through natural increase.” That would have been the case only if the noncoffee slave contingent had had a rate of internal growth of about 1.39 percent a year for several decades. Since there were manumissions, its rate of natural increase had to be somewhat higher than that, practically identical to that of the Brazilian free population and well above those of all Western European countries in the mid-1800s.4 This is a question to be settled on empirical grounds, and much more research is needed. We can affirm, however, that Dean’s thesis contradicts all contemporary opinion and all known evidence. The rate of natural growth of the Brazilian slave population from 1819 to 1872 was lower than — 0.9 percent per year. For Minas Gerais this rate was estimated at — 3.67 percent per year, by W. L. von Eschwege, circa 1820, before the rise of the coffee industry.5 However crude this estimate may be, it suggests that a shift to a highly positive rate in the following decades would imply such a gigantic demographic transition that we find it hard to believe.

There is consensus among the commentators that the growth of the Mineiro slave population cannot be explained by the direct labor demand of the only plantation sector of the province, the coffee industry. Indeed, the evidence on this point is overwhelming. The dynamic labor requirements of the coffee plantation sector (the annual increase in the number of slaves needed, considering the growth of the output and a generous allowance for replacements) increased from the annual average of 105 slaves in 1819-30, to 285 in 1831-40, 661 in 1841-50, 1,142 m 1851-60, and 2,324 in 1861-73. The underlying productivity coefficient is based on the total number of slaves, of both sexes and all ages, existing at the 153 Mineiro coffee plantations of Laerne’s sample. That means, as shown below, that these labor requirements include not only fieldwork in coffee, but also harvesting and processing the crop, the production of food and some manufactures for the plantation population, domestic labor, fazenda maintenance, and probably the production of a surplus of several food staples. To account for the transportation of the crop to the seaport, using the extreme assumption that all muleteers were slaves, we have to add 20 percent to the labor requirements in the period up to 1860 and 10 percent in the years after that date.6 Land clearing and the planting of new coffee groves were not performed by slaves.

Recalling that Minas Gerais imported, in net terms, the annual average of 6,897 slaves in 1819-73, we see that the coffee sector was directly responsible for only 1.8 percent of the slave imports in 1819-30, 4.9 percent in 1831-40, 10.7 percent in 1841-50, 19.9 percent in 1851-60, and 37.0 percent in 1861-73. For the entire period 1819-73 the importation of slaves for the coffee industry represents 15.9 percent of the provincial total, and was certainly negligible in the years 1801-18, during which the province imported some 61,000 bondsmen. We can conclude, therefore, that the labor demand generated directly by the coffee sector does not explain more than 13.7 percent of the Mineiro imports of slaves during the first seven decades of the nineteenth century.

Engerman and Genovese are satisfied with our demonstration that “the provincial economy rested, for the most part, on subsistence production and local markets; that it nonetheless exhibited a steady demand for slave labor; and that free land and an untamed peasantry made possible a slave system largely outside the world market”; that is, everything that we set out to prove. What they question is the possibility of survival of such a system, had it not been part of a larger one that was integrated in the world market for plantation staples. They point out that, if it had not been so, the supply of foreign slaves, on which Minas depended critically, would have been “unthinkable” and that the system would have been ideologically and politically infeasible in a milieu hostile to slavery.7 In short, were it not surrounded by an export-oriented slave system, the Mineiro economy would have collapsed into some sort of natural economy or had to find some other form of coerced labor.

We have not addressed ourselves to this question in the original paper, but have no quarrel with their argument. In fact, one should even ask whether such a slave system would have arisen at all if it had not been preceded by a century of slave-based gold and diamond mining. It was the “gold cycle” that implanted the institution in the Mineiro heartland, created a propertied ruling class, as well as the urban concentrations and local markets. In its absence the pace and pattern of settlement would certainly have been entirely different, probably based in family farms and subsistence agriculture, with little or no slave labor at all, along the lines followed by other nonexport areas in Brazil and elsewhere in America. Given the sparseness of the native population, it is difficult to think even of the “hacienda” type of settlement of Spanish America.

Much less appealing is their suggestion that the explanation for the persistence of slavery (as opposed to nonslave forms of labor coercion) in Minas may “lie more in the juro-political than in the economic realm” and, particularly, the comparison they propose with the northeastern transition from slave to seigneurial labor relations. The difference between the Northeast (a plantation region) and Minas Gerais (and, for that matter, São Paulo, another plantation area) lies not in the juro-political framework, but rather in the kingdom of the Wakefield-Nieboer-Domar hypothesis.

There are fascinating, though not adequately tapped,8 indications that by the beginning of last century most agricultural lands in the Northeast had already been appropriated and that a large mass of free landless peasants had come into being. It was this circumstance—the existence of an economically dependent peasantry—that permitted the gradual abandonment of slavery (which started long before abolition) and the development of “seigneurial” labor relations that eventually evolved into a wage system. The process was accelerated in the second half of the century by the occurrence of severe droughts (and the possibility of exporting the slaves to the coffee areas), which, by shrinking the subsistence lands of the sertão, sent thousands of migrants to join the labor force of the coastal plantation areas. The comparison between the Northeast and the Center-South reinforces the idea that it is the evolution of the land-labor situation, rather than the presence or absence of the export plantation, that is crucial to the survival or disappearance of the slave system in a given region.

The boldest attempt to rescue the association between large-scale slavery and an export-oriented economy in the Mineiro case, comes from Sienes. He begins by questioning our low export figures. Since our data are based on fiscal records, Sienes reminds us of the possibility of smuggling. It seems very unlikely that anything but a negligible part of the provincial exports could have bypassed the fiscal stations. The Mineiro road system was inherited from the colonial days and was designed to prevent the smuggling of gold and diamonds. Portuguese colonial policy and the rugged topography of the region (particularly along the Fluminense and the Paulista borders) concurred to keep the number of outlets at a minimum and to make it one of the worst transportation systems in the country. It is easy to imagine what a smuggler would have to face, at a time when the main routes were considered almost impassable mountain trails. In the case of some commodities, like cattle, for instance, a contraband trade was physically impossible.

Furthermore, one must keep in mind that fubá is not exactly gold dust. The bulk of the Mineiro export trade in the nineteenth century consisted of products of very low value per weight or volume, the smuggling of which would have been uneconomical. The increase in the cost of transportation to bypass the toll stations would easily overcome the value of the tax avoided. We estimated that in 1859, for instance, the maximum profitable detour was 36.2 miles long for cotton cloth; 17.6 for coffee; 10.4 for cheese; 5.8 for bacon and tobacco; 4.1 for rice; 2.2 for beans; 1.2 for corn; and 0.8 for corn flour.9 This reasoning assumes, of course, that alternative routes existed, that the cost of transportation over them was the same as in the main roads, and that the risk involved had a zero price.

Sienes also suggests that our comparison between Minas Gerais and other Brazilian regions in 1869-73 is misleading because these data “catch Minas at a time when its noncoffee export sector was economically depressed.” There is nothing special about Minas in that period: as a matter of fact, it was chosen because of the availability of the data for the other regions. After the article was written, new research was completed, using Mineiro archival sources. It was possible to retrieve detailed export data for forty-two of the fifty fiscal years from 1839 to 1889 and to construct better series than any previously available.10

The new data confirm our previous findings. The real value of non-coffee exports11 remained stagnant during most of the century, oscillating without a trend around its 1818-19 level. Only after the mid-1880s did noncoffee exports show some sustained growth. The bulk of these exports consisted of a few commodities of low labor-intensity—like cattle, hogs, bacon, and cheese—plus cotton cloth and tobacco, all of which were non-plantation goods produced not only by slaves but, also, to a large extent, by free labor.12 The exports of agricultural food staples like rice, beans, corn, corn flour, corn meal, manioc flour, and sugar were trifling.

Per capita noncoffee exports decreased steadily, in real terms, as shown in Table 1. The years 1869-73 were not a time of depression in the noncoffee export sector: in spite of a strong long-term downward trend, the average per capita value in this period was higher than in the previous and in the following five-year periods by 6.1 and 36.0 percent, respectively.

In his effort to bring the Mineiro case back on the tracks, Sienes also postulates a regional division of labor, in which the slaves in the coffee sector specialize in the production for the world market, while those not engaged in that activity are responsible for the supply of food and other commodities to the export plantation sector. The employment of slaves in the noncoffee region of Minas would, therefore, be the result of demands generated in the coffee areas of Río de Janeiro, São Paulo, and the Zona da Mata (including their urban centers). This interpretation, though more roundabout than the simple slavery/export-plantation model, would, nonetheless reestablish the coffee sector (or the world demand for coffee) in its role of engine of growth or “dynamic center” of the entire system.

The trouble with this view is that it lacks empirical support. There is no doubt that almost all Mineiro noncoffee exports went to Rio de Janeiro and, to a much smaller extent, to São Paulo.13 Yet, as our export data show, the flow was so small that it could not have played the role proposed by Sienes.14 In the case of the basic agricultural food staples (rice, beans, corn, and manioc flour) the quantities exported were systematically less than one-tenth of one percent of the estimated internal consumption of Minas.15 In the cases of cattle, hogs, bacon, and coarse cottons this ratio was certainly much higher, but still internal consumption was much larger than exports and, as mentioned above, these are products with a low labor content and not easily identifiable with slave labor.16

Viewed from the Paulista and Fluminense sides the flows of most Mineiro products were not of much importance either, except, perhaps, with respect to coarse cotton cloth and some foodstuffs for the urban market of Rio de Janeiro. Around 1850, for instance, the Mineiro exports of corn, beans, rice, and manioc flour would represent the consumption of these articles by 890, 5,553, 309, and 2,554 persons, respectively. The highest of these numbers, that corresponding to beans, represents 1.2 percent of the combined slave population of Río de Janeiro, São Paulo, and the Corte.17 Of the basic items of the diet of the rural population (slave and free) of those provinces, Mineiro supply may have been of some significance only in the case of bacon, but even here its importance should not be exaggerated.18 The Fluminense and Paulista coffee plantations were largely self-sufficient in foodstuffs and other inputs, at least, during the first half of the century (the period of fast growth of the Mineiro slave population).19

After the end of the African slave trade in the early 1850s, there seems to have been a tendency toward specialization in coffee, which brought about sharp increases in food prices in both rural and urban areas.20 More research is needed on this topic, but this alleged trend should be viewed with caution. Writing in 1883, Laerne observed that, in spite of the planters’ “violent passion for planting,” a slave employed in coffee fieldwork “besides maintaining the roads and bridges, [had] to attend to the planting of maize, beans, manioc, batatas, in short, all that is needed on a fazenda for the consumption of the work-people.”21

It is also very interesting to note that during the 1850s, while food prices increased by leaps and bounds in the plantation areas and in the city of Río, the real value of the Mineiro noncoffee exports declined from an index of 100 in 1851-55 to 81.2 in 1856-60. The quantities of basic food staples exported from Minas either increased very slightly or, in most cases, declined in this period.

What about intraprovincial flows of commodities in Minas, from the noncoffee region to the plantations in the Zona de Mata? The sheer smallness of the slave labor force of the coffee sector, for one thing, would prevent the possibility of large such flows. It would seem very odd if the economic raison d’être of some 200,000 to 300,000 Mineiro slaves was the production of a surplus with which to support a few thousand bondsmen employed in the export sector. That would have been a truly infeasible economic system.

Furthermore, there is no reason whatsoever to suppose that the coffee plantations of Minas were less self-sufficient than their Paulista and Fluminense counterparts. The opposite seems to have been true. The Soledade plantation near Paraibuna, for instance, grew coffee, corn, rice, sugarcane, and cotton, and had sugar and farinha mills, as well as a distillery. Its two hundred slaves were clothed by the domestic looms, which also produced bagging for coffee.22 The Zona de Mata as a whole, and many of its plantations, were not just self-sufficient; they also produced an exportable surplus of several commodities besides their main crop.23

Fazendas outside the coffee region were even more diversified than those within it, but were oriented toward subsistence and local markets. The famous Fazenda da Jaguara combined a wide variety of agricultural crops with stock raising, sugar mills, tanneries, cotton weaving, and mining. The production of its five hundred slaves was not exported or sold to the coffee zone, but consumed at home and retailed in the neighboring town of Sabará.24 In the mid-1870s James Wells stayed at the Fazenda Mesquita, a small estate that had several slaves and was moderately prosperous. There he noted that

the farming operations cannot be considered as remunerative occupations, they afford just a bare homely subsistence; there is no rent or taxes or wages to pay, and the little surplus production of the farm, or the occasional sale of a bullock, affords sufficient means to purchase the few homely requisites that the farm does not produce, such as a piece of printed calico or sheeting, hats, a few iron kitchen utensils, or to engage a carpenter to repair any damage to the family chariot—the bullock-cart.25

In 1867 the typical Mineiro slave fazenda was described by Sir Richard Burton:

These fazendas are isolated villages on a small scale. They supply the neighborhood with its simple wants, dry beef, pork and lard, flour of manioc and of maize, sugar and spirits, tobacco and oil, coarse cloth and cotton thread; coffee and various teas of Caparosa and orange-leaf. They import only iron to turn into horse shoes; salt, wine and beer, cigars, butter, porcelain, drugs, and other “notions.” There is generally a smithy, a carpenter’s shed, a shoemaker’s shop, a piggery … and an ample poultry yard.26

The linkages from the coffee sector to the rest of the province were further weakened by the Zona da Mata’s proximity and commercial ties with Río de Janeiro. It was easier for it to obtain the goods it did not produce from the capital’s market than from the Mineiro hinterland. The (slave-based) iron industry is a case in point: it supplied most of the province with horseshoes, agricultural tools, and other iron goods; but southward its market went only as far as Barbacena, where the competition of foreign articles began to be felt.27

It is clear that the debate over the slave system of Minas Gerais or, in a broader sense, over the nature of nonplantation slavery is only beginning.28 Yet, the comments of these highly respected specialists convinced us that our paper achieved its main goal: to call the attention of students of slavery and of Brazilian history to this important area, the neglect of which has resulted in an enormous lacuna in our understanding of the Brazilian past and of modern slavery. We totally agree with the agenda set forth by them, which asks for more local studies and less emphasis on the plantation, and would like to add our own plea for more empirical research and fewer sweeping generalizations.


We refer the interested reader to Roberto B. Martins, “Growing in Silence: The Slave Economy of Nineteenth-Century Minas Gerais, Brazil (Ph. D. Diss., Vanderbilt University, 1980), for some of these answers as well as for additional evidence on the issues being debated.


Giorgio Mortara, “Demographic Studies in Brazil” in P. M. Hauser and O. D. Duncan, eds., The Study of Population (Chicago, 1959), pp. 235-236. In spite of the quality of its data, the 1872 census contains large errors in the summation of the parish figures to get the provincial totals. Our work is based on the correct sums of the parish data, while Conrad, Klein, Merrick and Graham, and others used the distorted provincial tables. A thorough revision of this census is under way at CEDEPLAR/UFMG, by Roberto B. Martins, Clotilde A. Paiva, and María do Carmo S. Martins, and a corrected version will be available in 1984.


Joaquim Norberto de Souza e Silva, Investigações sobre os Recenseamentos da Popolação Gerai do Império e de Cada Província de Per Se Tentados desde os Tempos Colonices até Hoje (Rio de Janeiro, 1870). See also, F. J. Oliveira Vianna, “Resumo Histórico dos Inquéritos Censitários Realizados no Brasil,” in Directoría Geral de Estatística. Recenseamento do Brasil Realizado em 1° de Setembro de 1920 (Rio de Janeiro, 1922-29), pp. 404-405. For the other early century estimates, see Martins, “Growing in Silence,” Appendix A.


The rate of 1.39 was computed assuming that the slaves not employed in coffee were 168,543 in 1819 and 354,688 in 1873. The Western European rates are in Simon Kuznets, Modern Economic Growth (New Haven, 1966), pp. 42-44.


For a discussion of this issue, see Roberto B. Martins, “Minas Gerais, Sécolo XIX: Tráfico e Apego à Escravidão numa Economia Não-Exportadora,” Estudos Econômicos, 13 (1983), 181-209, and Martins, “Growing in Silence,” pp. 171-178.


The duration of slaves in the field has been estimated from seven to fifteen years. We assumed that one-seventh of all plantation slaves had to be replaced every year. The number of new slaves needed in year t was then estimated as Rt — .86Rt-1 where R, is the labor required in year t. The labor requirements for transportation were drastically reduced by the opening of the União e Indústria highway in 1861 and the arrival of the railroad tracks to the coffee zone in 1869. The estimations of R, and of the transportation requirements are found in Martins, “Growing in Silence,” pp. 120-146.


This last difficulty has nothing to do with the nonexport character of the economy. Even a booming export system like that of São Paulo had to give up slavery when the institution lost its ideological and political support in the rest of the country and in urban areas.


Exceptions to this statement are Peter Eisenberg’s The Sugar Industry of Pernambuco: Modernization without Change (Berkeley, 1974) and the works of Jaime Reis.


We used the cost per ton/kilometer of muleback transportation over the route Rio Claro-Santos, estimated for 1859 by Warren Dean, Rio Claro: A Brazilian Plantation System, 1820-1920 (Stanford, 1976), p. 40. There is no reason to believe that this cost was lower in Minas Gerais. The source of the values of the export taxes is found in note 10, infra.


Roberto B. Martins and María do Carino S. Martins, “As Exportações de Minas Gerais no Século XIX. Versão Preliminar,” paper presented at the first Seminário sobre a Economia Mineira (Diamantina, Sept. 1982).


Throughout this article, “Mineiro exports" means all goods shipped out of the province, either for consumption in other provinces or for exportation to foreign markets.


The six commodities mentioned accounted for the following percentages of the value of noncoffee exports: 62.1 in 1818-19; 93.0 in 1839-48; 93-4 in 1850-59; 80.6 in 1860-69; 93.9 in 1870-79; and 94.1 in 1880-89. Raw cotton was important during a short period in the early part of the century. Lenharo’s contention, mentioned by Sienes, that “many Mineiro ranches in the Sul used slaves in large numbers” should be confronted with Auguste de Saint-Hilaire’s description of that cattle-producing area, where he observed that slaves were few, being outnumbered by free workers by three to one, and that the farmers and their families engaged in all the stages of stockraising, including the production of cheeses and bacon, and the transportation of these commodities, cattle and hogs on the hoof, to the Rio market. See Auguste de Saint-Hilaire, Segunda Viagem do Rio de Janeiro a Minas Gerais e a São Paulo. 1822 (Belo Horizonte, 1974), p. 54.


A considerable part of the Minas cotton cloth sent to Rio was reexported to the southern provinces. The contemporary observations of Spix and Martius and of Saint-Hilaire show that its market reached Río Grande do Sul and the River Plate area in the early nineteenth century. See Roberto B. Martins, “A Indústria Têxtil Doméstica de Minas Gerais no Século XIX,” working paper, CEDEPLAR/UFMG, 1982.


It would be incorrect to impute all noncoffee exports to the noncoffee region. A considerable, though impossible to determine at present, part of them was produced in the coffee zone.


Mineiro internal consumption was estimated using per capita consumption figures for São Paudo in 1836. The source and the defense of this procedure are in Martins, “Growing in Silence," p. 284.


The Mineiro textile sector employed slaves but was also a peasant industry, to a large extent. It exported between one-third and one-fourth of its output. Martins, “A Indùstria Têxtil Doméstica.”


We used the per capita consumption coefficients mentioned in note 15, supra. These coefficients are not available for the other relevant products.


Fresh meat and pork were seldom consumed by slaves or the rural poor. The jerked meat consumed on plantations was imported from Riío Grande do Sul or from abroad. Probably most of the cattle exported from Minas went to urban markets.


“Before the early 1850s both large coffee planters and small agriculturists had grown foodstuffs for personal consumption, with the surplus going to the local markets.” Stanley J. Stein, Vassouras: A Brazilian Coffee County, 1850-1890 (New York, 1976), p. 47. “The plantations of Río Claro were largely self-sufficient in foodstuffs, construction materials, draft animals, and some handcrafts.” Dean, Rio Claro, p. 38; see also pp. 39, 64.


Dean, Rio Claro, p. 38; Stein, Vassouras, pp. 47-48; Sebastião Ferreira Soares, Notas Estatísticas sobre a Produção Agrícola e Carestia dos Gêneros Alimenticios no Império do Brasil (Río de Janeiro, 1977), pp. 132-134.


C. F. van Delden Laerne, Brazil and Java: Report on Coffee Culture in America, Asia and Africa (London, 1885), p. 290.


Francis Castelnau, Expedição às Regiões Centrais da América do Sul, 2 vols. (São Paulo, 1949), II, 122–123.


In the early 1840s, the plantation of Capitão Francisco Leite, for instance, produced large quantities of cheese, sugar, and brandy, besides 11,000 arrobas of coffee per year. George Gardner, Viagens no Brasil (São Paulo, 1942), pp. 447-448.


See Martins, “Growing in Silence,” p. 320.


James Wells, Exploring and Travelling Three Thousand Miles through Brazil, 2 vols. (London, 1877), I, 163-165.


Sir Richard F. Burton, Explorations of the Highlands of Brazil, 2 vols. (London, 1869), II, 39-40.


See Martins, “Growing in Silence,” p. 312.


Regarding the relationship between slavery and the plantation in Brazil, we would call attention to Schwartz’s recent article on the slave-based economy of the Bahian Recôncavo in the late colonial period: the author reaches some conclusions strikingly similar to our own findings about nineteenth-century Minas Gerais. In spite of the fact that the Recôncavo was a major sugar plantation region during that period, Schwartz argues that aside from the planter class there also existed a broad minority of slaveholders, hundreds of individuals or families with one or two or five slaves, whose investment in slavery was quantitatively small but whose attachment to the institution was no less real. They were the majority of the slaveholders, and they controlled a substantial portion of the slaves." Moreover, according to Schwartz, the “majority of slaveowners were not active in sugar production but were, instead, involved in other economic activities or in urban settings, which had much lower concentrations of wealth in slaves." He concludes that, contrary to the prevailing historiography, “by the end of the colonial era, neither Brazil as a whole nor its Bahian Recôncavo was simply a sugar plantation writ large." See Stuart B. Schwartz, "Patterns of Slave-holdings in the Americas: New Evidence from Brazil," American Historical Review, 83 (Feb. 1982), 62, 84.