¹ Time on the Cross, Vol. I: The Economics of American Negro Slavery, by Robert William Fogel and Stanley L. Engerman. 286 pp. Boston: Little, Brown and Company. $8.95; Vol. II; Evidence and Methods—A Supplement, by Robert William Fogel and Stanley L. Engerman, 267 pp. Boston: Little, Brown and Company, $12.50.

² “The Jolly Institution,” New York Review of Books, XXI (May 2, 1974), 3.Google Scholar

³ Cf., e.g., the essays by Mintz, Sidney W., Harris, Marvin; Davis, David Brion, Sio, Arnold S., and others in Foner, Laura and Genovese, Engene D., eds., Slavery in the New World (Englewood Cliffs, N.J.: Prentice-Hall, 1969).Google Scholar

⁴ Curtin, Phillip D., The Atlantic Slave Trade: A Census (Madison: University of Wisconsin Press, 1969).Google Scholar

⁵ Cf. Davis, David B., “Slavery and the Post-World War II Historians,” Daedalus (Spring 1974), 9–10, for this imputation of purpose.Google Scholar

⁶ Cf. Easterlin, R. A., “Regional Income Trends, 1840–1950,” in Harris, Seymour, ed., American Economic History (New York: McGraw-Hill, 1961)Google Scholar; Wesley, C. H., Negro Labor in the United States, 1850–1925 (New York: Vanguard Press, 1927)Google Scholar; Evans, R. Jr, “The Economics of American Negro Slavery,” in Aspects of Labor Economics (Princeton: National Bureau of Economic Research, 1962)Google Scholar; Starobin, R. S., Industrial Slavery in the Old South (New York: Oxford University Press, 1970)Google Scholar; Gallman, R. E., “Trends in the Size Distribution of Wealth in the Nineteenth Century: Some Speculations,” in Six Papers on the Size Distribution of Wealth and Income, ed. Soltow, Lee (New York: Columbia University Press, 1969)Google Scholar; Soltow, L., “Economic Inequality in the United States in the Period from 1790 to 1860,” Jodbnal of Economic History, XXXI (December 1971), 822–839.Google Scholar

⁷ The wealth distribution obviously was more concentrated than it would have been had the blacks owned themselves (had they not been the property of others), and been able to accumulate other forms of wealth. But it is usually argued that the effects on the level and composition of effective demand primarily are those exercised by the distribution of income. Fogel and Engerman's (Vol. I, pp. 253–54) brief discussion of slaves' consumption needs and the market for industrial goods seems to accept this emphasis on income distribution.

⁸ “The principal cause of the persistence of the myth of black incompetence in American historiography is racism.” Thus begins the section of the text headed “Toward an Explanation for the Persistence of the Myth of Black Incompetence,” on p. 223 of Vol. I. Eight pages later: “What, of course, is common to Stampp and Elkins is agreement on the characteristics of slave behavior: slaves lie, steal, reign illness, behave childishly, and shirk their duties. Indeed, this characterization has been one of the enduring constants in the literature on slavery.” The innuendos carried by the wording of Fogel and Engerman's excruciatingly labored critique of Stampp's Peculiar Institution, in Appendix C, are especially unfortunate:

“Stampp's inadequate definition of black accomplishments under slavery may have been related to his approach to the issue of racism. While Stampp unequivocally rejected the contention that Negroes were biologically inferior to whites, he did not consider how [racism] might have affected the views of both the ‘eyewitnesses’ and the historians that he invoked as authorities. … Stampp reported as fact, untinged by racism, Olmstead's description of slaves as ‘chronic’ malingerers. . . .” (Vol. II, p. 234–35. Emphasis added.)

⁹ Time on the Cross, Vol. II, p. 16.

¹⁰ “Slavery and the Post-World War II Historians,” p. 11.

¹¹ The role of these new characterizations in the overall structure of Time on the Cross, as well as the evidence adduced for them, are discussed in David, P. A. and Temin, P., “Capitalist Masters, Bourgeois Slaves,” Journal of Interdisciplinary History (forthcoming, Winter 1975).Google Scholar

¹² C. Vann Woodward, observing that Fogel and Engerman's second volume, Evidence and Methods “is given over to documentation, defense of method, algebraic equations, tables and graphs, and computer language [of which we find none] beyond the comprehension of laymen,” voices this hope: “Their findings, their data, and their methods should have the most thorough and unsparing criticism…, especially [by] those who speak their own language.” “The Jolly Institution,” pp. 3, 6.

¹³ The Peculiar Institution, p. 28

¹⁴ It may be noted that Fogel and Engerman's concentration on large plantations might have been suggested by Stampp's assertion that, in the quality or variety of food given, there were no “appreciable differences” in the practices of large and small slaveholders. But why take this for granted when all else is being cast into doubt?

¹⁵ Cf. Bennett, Merrill K. and Peirce, Rosamond H., “Approximate Levels of Food Consumption Before 1909,” [Food Research Institute (mimeo), Stanford, April 16, 1962]—A Supplement to “Change in the American National Diet, 1879–1959,” Food Research Institute Studies, May 1961, pp. 95–119.Google Scholar The methodological supplement is referred to in the published Bennett-Peirce study which Fogel and Engerman cite in Vol. II, p. 98.

¹⁶ Since we are not told anything about the age- and sex-composition of the slave population represented in die sample studied by Fogel and Engerman, we cannot say whether there were proportionately more adult males—whose calorie requirements exceed those of women and children—than was the case for the 1879 U.S. population as a whole.

¹⁷ In this connection it would be interesting to know what proportion of the slave force (on the large isolated plantations comprising Fogel and Engerman's sample) were house servants, rather than field hands.

¹⁸ Cf. the data from Tigerstedt's Textbook of Physiology reproduced by Sherman, Henry C., Chemistry of Food and Nutrition, 7th Edition (New York: Macmillan, 1946), p. 190.Google Scholar

¹⁹ Sherman, Chemistry of Food and Nutrition, pp. 195–196, reporting on the findings of G. F. Benedict and his co-investigators.

²⁰ Cf. Vol. II, pp. 159–160. Using “Atwater's weights,” (we presume) the authors have adjusted their ($42.99) figure for per capita slave income slightly upward, to allow for the difference between the age-sex composition of the constructed family group and that of the representative plantation population—to which the average maintenance and extra income allowances refer. “Atwater's weights,” to which cryptic reference is made by. the section (4.11.1) we have quoted from Appendix B above, receive more explicit attention below.

²¹ See below for discussion of the differences in hours worked by the free and slave labor force. We reckon that a minimum adjustment on this account would indicate that the slave work force put in 16 to 22 percent more male equivalent manhours per worker than the free southern agricultural labor force. In addition to this one must consider the higher proportion of the women and children in the rural slave population that were part of the work force.

²² This imputation is based on the results of estimating a cross section Cobb-Douglas production function (for all southern farms and plantations in the Parker-Galhnan sample) in which there turned out to have been some significant scale economies. Cf. Vol. II, pp. 133, 143.

²³ Cf. Vol. II, p. 125 (sect. 4.11.6.2). The 49 percent figure in question appears to be correct, despite the fact that the algebraic formula for it is confusingly given (incorrectly) as: B/∑λ_tR_gt, where λ_t is the average survival rate to the t-th year, R_gt is the gross revenue in the t-th year and B is the “birthright.” Instead of B, elsewhere defined as a discounted magnitude [B = ∑λ_t(R_gt − M_t)/(1 + i)^t], in this context Fogel and Engerman obviously meant to refer to B° ≡ ∑λ_t(R_gt − M_t). Notice that, like the discounting factor (1 + i)− t, the survival frequency (λ_t) declines with increasing t and therefore it too accords heaviest weight to the (early) negative entries in the net revenue stream. We return to this point later in the text, following explanation of the way Fogel and Engerman have estimated the net revenues, (R_g − M)_t.

²⁴ The method of calculation outlined here (i.e., replacing the present value of the expected net revenue stream by the value of the birthright) is not the one described in the relevant portions of Appendix B, specifically Vol. II, pp. 119–120, including section 4.11.1 quoted above and the sections referred to therein. The method Fogel and Engerman describe is incorrect, but we have verified in private conversation with the authors that they followed the different procedure outlined here.

²⁵ Some students of the southern probate records casually assert that slave appraisals were usually low in relation to actual market prices. Cf., e.g., Postell, W. D., The Health of Slaves in Southern Plantations (Baton Rouge: Louisiana State University Press, 1951), p. 52.Google Scholar However, the important question here, which remains unanswered, is whether there were significant disparities between the relative age-specific appraisal values and the relative age-specific transaction prices.

More generally it is most plausible to think that the appraisers worked with conventional valuations which were both less. subject than market prices to regional variations, and less sensitive to changing expectations about the future course of market values. Perhaps this is why the dispersion of relative values drawn from estates in a region as large as the Old South is not greater than it appears to be from Fogel and Engerman's Figure 15, in Vol. I, p. 72. And perhaps, too, this is why they are able to report that an “age-price profile” for male slaves computed from the probate data during a period of rapidly falling market prices of slaves (1838–1843) “was quite similar to” the corresponding profile computed from data for the period 1850–1860, when slave prices were rapidly rising. Fogel and Engerman (Vol. II, p. 79) note this counter-theoretic “similarity,” describing it as a puzzle for which no explanation has yet been suggested.

²⁶ According to Appendix B (Vol. II, p. 80) the relative “price” observations for each age were first averaged, and a sixth-order polynomial (in age) was then fitted, by least squares regression, to the sample midpoint observations thus obtained. The estimated equation, however, is not reported.

²⁷ This is tantamount to saying that the risk averse individual is one for whom the marginal utility derived from income (or wealth) is diminishing. The measures of risk aversion proposed by Arrow and Pratt make use of this property of the usual bounded form of utility index, and thus incorporate the hypothesis that risk aversion increases with the level of income (or wealth). Cf., e.g., K. J. Arrow, Aspects of the Theory of Risk-Bearing, Yrjo Jahnsson Lectures, Helsinki, 1965, Ch. 2.

²⁸ What happens thereafter is considerably less clear, since the declining variance of the individual slave's survival may be offset by increased variance in the various dimensions of “performance” with the onset of senescence. The situation for females passing from adolescence into their reproductive years is more complex than that tor males: decreased variance of the a priori distribution of fertility can be offset by increased “risk” due to childbearing. In the present state of knowledge all these statements must be regarded as conjectural, but they equally serve to suggest the implausibility of assuming that the variance of the future net revenue stream is independent of the slave's age. We are indebted to Professor Warren C. Sanderson for impressing upon us the fact that it is upon just this implausible assumption that Fogel and Engerman must rely in order to infer an age profile of average net revenues from their age profile of slave prices and expected survival rates. Unless, of course, they attempt to defend their whole approach by maintaining that slaveholders gave no consideration whatsoever to risk.

²⁹ Cf., Vol. I, pp. 73–78, 82–83, for these inferences. The estimation method, discussed in Vol. II, pp. 80–81, involves using relative prices for adjoining ages and year-to-year survival frequencies to infer an underlying age profile in net revenues. The authors employ an iterative procedure in order to jointly find the uniform discount rate required to set the present value of the inferred remaining stream of average anticipated (i.e., survival adjusted) net revenues equal to the price of a slave at any given age. This internal rate is found to be 10 percent, or so we have reason to believe from die figure cited—in a different connection—in Vol. II, p. 78.

³⁰ The argurnent developed here extends to their inferences about females' earnings as well. Fogel and Engerman state (Vol. I, p. 83) that “on average, net income from childbearing was only about 10 percent of the total net income earned by women during their childbearing years.” They go on to argue from this that manipulation of the sexual lives of slaves in order to further raise fertility could not have substantially affected the overall net income derived from the ownership of females. But inasmuch as the derivation of these net income (revenue) figures abstracts from the differentially greater risk associated with the yield via the (sale of) birthrights of the offspring—compared with the risk associated with net revenues from participation in field work—the 10 percent figure cited is clearly a lower bound estimate of the magnitude in question. Whether or not the latter is relevant for the slaveholder's decision-making, as the authors' discussion would suggest, and whether or not their argument on this point is germane to the thesis that there was little interference by masters in the sexual lives of the slaves because it would not have been very profitable to do this, are important questions. But they raise issues quite distinct from those being considered in this review.

³¹ Cf. Vol. II, p. 125. The (discounted) measure corresponding to the first “moral” position is termed a “Robinsonian” measure (Vol. II, p. 87) to distinguish it from the second, “Marxian,” measure of exploitation—although there appears to be no precedent in either of the early or the recent writings of Joan Robinson to support the ethical propriety of considering only the present value of an expected future difference between the marginal productivity of labor and the real wage.

³² Vol. II, p. 120. Cf. Section 4.11.2.1 (pp. 120–22) for Fogel and Engerman's skeptical questions about the “traditional interpretation” of the slave experience in Jamaica, and elsewhere in the Caribbean and South America.

³³ It is closer because a dollar spent on childrearing is weighted the same as a dollar of earnings. In any discounted sum, contemporaneous expenses and earnings would also be weighted equally.

³⁴ The undiscounted rate of exploitation, is calculated as is understated unless ∑λ_tM_t has been understated by the same proportional amount as is B° (where λ_t is the probability of a slave living in age t, M_t is the maintenance cost of a slave of age t, and B° is the birthright calculated in the absence of discounting).

³⁵ Cf. Massachusetts Bureau of Statistics of Labor, 17th Annual Report (March 1886), Part III, “Food Consumption,” esp. pp. 262–267.Google Scholar

³⁶ It appears that Fogel and Engerman use Atwater's estimates of the calorie requirements of children and youths relative to requirements for adults doing moderate work, rather than expressing them as (smaller) fractions of the estimated calories required by adults engaged in vigorous labor.

³⁷ Appendix B (Vol. II, p. 139, Tables B.23 and B.24) discloses wide variation in the relative efficiency indexes which are presented as averages for slave plantations belonging to different regions and different broad size-classes. For example, the measured total factor productivity of plantations holding 16–50 slaves in the New South is 70.5 percent greater than that of “plantations” in the 1–15 slave category in the Old South. Thus, within the group of slave farms the difference from the mid-point to the extreme of this range is at least 35 percent, which is just as large as the relative productivity advantage that Fogel and Engerman report for their comparison of “all slave farms” and “northern farms.” This suggests that the latter productivity differential in favor of the slave mode of farming is not statistically significant.

³⁸ Vol. I, p. 192, emphasis added. Unfortunately, the text immediately goes on to restate this in the following way: “that is, on average, a southern farm using a given amount of labor, land, and capital could produce about 35 percent more output than a northern farm, or groups of farms, using the same quantities of these inputs.” This is a formally correct restatement if we take the qualifying “on average” to refer to weighted averages of all the individual southern and northern farms, the weights corresponding to their contributions to the respective aggregate agricultural outputs of the two regions. Appendix B Tables B.21 and B.22 (cf. Vol. II, pp. 135–137), and the accompanying discussion make it clear that the relative efficiency index derives from an aggregate total factor productivity calculation for southern and northern agriculture. But in describing the derivation of parallel average productivity measures for the farms in different regions and size classes within the South (Vol. II, pp. 138–140) Fogel and Engerman do not explicitly say whether the averages are weighted or unweighted. And as their southern production in this case comes not from the full Census of Agriculture for 1860, but instead from the Parker-Gallman sample, it also remains unclear whether the weighting—if weighting was implicitly or explicitly employed—followed the representation of farm types in all southern agriculture, or just in the 5,230 farm sample drawn from counties that produced at least 1000 bales of cotton in 1860. For further details of this sample, cf. William N. Parker, ed., The Structure of the Cotton Economy of the Antebellum South (Washington: Agricultural History Society, 1970), esp. the articles by Gallman, Foust and Swan, and Wright.

³⁹ It already has been noted above that Fogel and Engerman reject as tainted by racism all contemporary statements, and more recent historical interpretations, which would suggest that slaves were less competent, and less diligent workers than free northern farm laborers. The fact that the slaves were black and the free workers were white obviously complicates the question, but the race issue should not be allowed to completely obliterate more general predispositions to regard slaves as inferior workers. As Davis, David B.(The Problem of Slavery in Western Culture [Ithaca: Cornell University Press, 1966], pp. 59–60)Google Scholar points out: “The white slaves of antiquity and the Middle Ages were often described in terms that fit the later stereotype of the Negro. Throughout history it has been said that slaves, though occasionally as loyal and faithful as good dogs, were for the most part lazy, irresponsible, cunning, rebellious, untrustworthy and sexually promiscuous.” Moreover, it is one thing to argue against a presumption of inferior work performance by black slaves, but it is quite another to establish their superiority. Discounting the contrary qualitative evidence will not suffice for the latter purpose.

⁴⁰ In the essay on “Plantations” in The International Encyclopaedia of the Social Sciences, W. O. Jones maintains that such economic advantage as the modem plantation possesses derives from “its ability to mobilize unskilled labor”; the plantation succeeds by substituting supervisory and administrative skills for skilled, adaptive labor, making most of the availability of a labor force whose principal skill is “to follow orders.”

⁴¹ One should not be tempted into speaking of “the superiority of black labor in the context of slavery,” for slavery clearly was neither a sufficient nor a necessary condition for eliciting a response to the intrinsic motivation of group competition and work rhythm. Both phenomena have been remarked upon by ethnographers and students or work organization among free non-industrial peoples. Rhythm is sometimes reported explicitly as an “incentive” in tillage labor, where a line of workers may reap or hoe in unison; in the Dahomean dokpwe, and the Haitian combite, we are told, special drummers and songs are used. Gang competition as a work “incentive” is more widely observed and certainly not restricted to peoples of African stock. In rice planting among the Betsileo, for example, the women attempt to plant shoots faster than the men can prepare the field ahead of them. Cf. Udy, Stanley H. Jr, Organization of Work (New Haven: HRAF Press, 1959), pp. 114–115.Google Scholar

⁴² Cf. Higbee, Edward, American Agriculture: Geography, Resources, Conservation (New York: Wiley and Sons, 1958), p. 27Google Scholar, for maps of frost-free periods based on the records for the years 1899–1938. According to the discussion of farm hours by Blodgett, J. H. (“Wages of Farm Labor in the United States,” U.S. Department of Agriculture Bureau of Statistics, Miscellaneous Series-Bulletin No. 26 [Washington, 1903] pp. 23–24)Google Scholar, however, 3 to 4 months were lost for field work in the North, whereas perhaps 2 months were lost in the South—typically 6 weeks in the hot months when cultivation ceased as corn and cotton crops mature during the lull from active growth, and 2 more weeks in winter. (It is noted that the traditional week of freedom afforded slaves at Christmas coincided with the winter suspension of field work.) Putting the difference between the time lost in the north and in the south at 6 weeks, and taking the 6 day week as normal for slaves in the antebellum period, we can think of latitude putting 36 extra field work days at the disposal of the southern plantation. No extra allowance is considered for differences in field work time due to suspensions of activity enforced by rain, etc.

⁴³ The 10 percent extra southern labor input figure is based upon the estimate of 36 extra field days discussed in the previous footnote. It was developed in the following steps: (1) The 36 extra field work days was translated into 367.2 hours, using 10.2 hours per day. The latter is the average amount of daylight in the latitude of Charleston, S.C. during the December-January period when the net loss of time occurs for the North (Cf. American Almanac, 1888). (2) Outside the winter months the length of the northern period of daylight is greater. In June-August, when the difference is greatest, farms in the latitude of New York City had approximately 54 extra hours of daylight compared with farms in the latitude of Charleston, S.C.—according to the sunrise-sunset intervals published in the American Almanac for 1888. (3) According to the W.P.A. National Research Project field survey conducted in 1936, average daily chore hours per worker in northern wheat- and corn-region farms exceeded those on southern Eastern and Delta cotton-region farms by 3.5 hours during the winter quarter. We add 1 hour of chores to the assumed 10.2 hours per day of field work, bringing the southern work day to 11.2 hours for the winter quarter. It is therefore appropriate to subtract [3.5 × 36 =] 126 hours of chore work from the 313 [= 367 − 54] net additional hours afforded southern farms by the latitude difference. This gives a remainder of 187 extra hours, which represents 9 percent of the 2163 hours estimated as the annual per man input of labor in Northern corn regions by the National Research Project Survey for 1936. Cf. Hopkins, John A., Changing Technology and Employment in Agriculture (U.S. Department of Agriculture, Bureau of Agriculture Economics, Washington, D.C., May 1941), pp. 23–27.Google Scholar An alternative procedure, using the figures supplied by Hopkins and assuming that due to differences in weather as well as in climate (associated with latitude) the South enjoyed 60 extra days of field work, yields a 14.8 percent adjustment in the per worker input of labor time for the South relative to the North. Oh the strength of this we round the previous, conservative 9 percent figure upward, to 10 percent.

⁴⁴ Their text says that plantations' comparative ability to fully utilize potential labor arose, “not because slaves worked more hours per day or more days per week than free farmers. The best available evidence is that both slaves and free farmers averaged approximately 70–75 hours of work per week during the peak labor periods of planting, cultivating, and harvesting. Nor does it appear that slaves worked more days per year” (Vol. I, p. 208). We are unable to find any citations supporting these assertions in the appendix B Notes to Chapter 6. Moreover the statement about length of the average work day carried the significant qualifying phrase: “during the peak labor periods.” For evidence and arguments contradicting Fogel and Engerman on this point, see below.

⁴⁵ Cf. R. Ransom and R. Sutch, “The Impact of the Civil War and of Emancipation on Southern Agriculture,” Table A, p. A.5. We are grateful to Professors Ransom and Sutch for permission to use the data presented in this unpublished paper in computing the per worker male equivalent hours figures cited here. The fractional weighting of women's and children's labor-time differences gives a particularly conservative measure in this case, because the proportional decline in the hours worked by secondary family workers in the transition from slavery to freedom was more pronounced than the proportional decline in the case of males. For all members of the mack work force the absolute reduction in estimated annual labor time runs in the range of 500–600 hours. These estimates are consistent with Charles Seagrave's estimates of declines of between 9 and 74 days worked per year by Louisiana class 1 field hands in the immediate post emancipation period 1864–67. Cf. Seagrave, C. E., “The Southern Negro Agricultural Worker: 1850–1870,” Unpublished Ph.D. dissertation, Stanford University, 1971, pp. 71–72.Google Scholar

⁴⁶ The calculation is made as follows: Relative to free northern firming the southern slave farm sector labor input index is to be raised by (1.10)(1.16) = 1.28 or (1.10) (1.22) = 1.34. Fogel and Engerman (Vol. II, pp. 126–127) adopt the Cobb-Douglas form of aggregate production function and therefore compute the index of total factor inputs as a weighted geometric average of the labor, land and capital input indexes. Their weight for labor—as previously noted above—is 0.58, so the adjustment factor by which the relative total input index for slave agriculture (vis-à-vis northern agriculture) should be multiplied is (1.28).⁵⁸ = 1.154, or (1.34).⁵⁸ = 1.185. These ratios, divided into the ratio of southern slave factor productivity to northern agricultural factor productivity (which Fogel and Engerman report as 1.40) yield the partially “corrected” versions of the latter relative productivity ratios which are mentioned in the text below: 1.40/1.185 = 1.18, and 1.40/1.154 = 1,21.

⁴⁷ After these two procedures are described, at the beginning of the second paragraph in Vol. II, p. 138, there follows this statement: “This adjustment in the southern labor input without a corresponding adjustment in the northern labor input biases the relative advantage of the South downward.” On careful reading it becomes clear that this applies only to their adjustment for labor force participation by women and children in southern agriculture, and not to the “equivalent worker” adjustments.

⁴⁸ Cf. Vol. II, pp. 132, 135, and Tables B.20 and B.21. The ratio of southern to northern land inputs is reduced by a factor of 0.40 (from 1.257 to .505) by the value-weighting of acreage. Since land gets a .25 weight in the geometric averaging of all inputs, the latter index is only reduced by the factor (.40²⁵ = ) 0.796 on this account. The effect on measured total input productivity is calculated as l/(.796) = 1.256.

⁴⁹ Thus, were resident southern landowners preponderant in the market for southem rural land because they had better access to information than non-residents, and were they more risk-averse as a class than northern rural landowners (because those having access to mortgage finance tended to be the larger, wealthier planters, and the degree of risk aversion increases with wealth), it would follow that the market price established for the same future expected stream of land rental yields would be lower in the South than in the North. Note that higher southern mortgage rates, and/or shorter time horizons on the part of investors in southern land, would have parallel effects biasing Fogel and Engerman's index of the relative “quality” of southern land inputs in a downward direction. We are not prepared to assert that all these awkward conditions obtained, but the authors have neither discussed the possibility nor adduced any evidence to suggest that such was not the case historically.

⁵⁰ As an alternative to the strategy considered in the text discussion, Fogel and Engerman might have attempted to apply a common national set of relative average prices in weighting improved and unimproved acreage in the two regions. In the same spirit, it would be possible to compute two separate indexes of relative unit values for all the improved lands within each region, and then set the. median values of the two indexes equal.

It should be clear that the procedures sketched here would, and probably should be employed to remove biases in the measurement of land inputs due to the effect of varying transportation conditions oh land price levels within the South—and not only between the South and the North. It remains unclear to what extent the apparent factor productivity advantage reported for free and slave farms of all sizes in the New South, vis-à-vis corresponding farms in the Old South, actually derives from the lower level of land prices associated with inferior transport conditions in the former region. Cf. Vol. II, p. 139, Table B.23, for these comparisons. Fogel and Engerman's text discussion of intra-South factor productivity differences (Vol. I, pp. 192–195) does not mention their finding that these extend to free farms as well, but the latter suggests that the greater relative efficiency of New South plantations compared with those of the Old South cannot properly be taken as solely reflecting regional differences in slave plantation characteristics and management.

⁵¹ Cf. Easterlin, R. A., “Farm Production and Income in Old and New Areas at Mid-Century” (unpublished manuscript, 1973)Google Scholar, forthcoming in The Old Northwest: Essays in Economic History, D. C. Klingaman and R. K. Vedder, eds. The comparison cited by Fogel and Engerman (Vol. II, p. 134) is for the entire South against the entire North in 1840. This yields a (.938) correction factor which would not have quite as large an effect in reducing the relative southern output. (Still, 7 percent represents a fifth of the 35 percent relative efficiency advantage they find for all southern agriculture.) In 1840, however, the transport conditions in the Old North-west and the trans-Mississippi west were much inferior to those prevailing in the southern Gulf Plains and Delta regions, and correspondingly the general level of local prices was higher in the latter than it was in the North Central census region. By 1860, on the other hand, these conditions had been dramatically transformed by the building of railroads and the fuller development of the northern water routes leading to the eastern seaboard. (Cf., e.g., Fishlow, Albert, “Antebellum Interregional Trade Reconsidered,” in Andreano, R. L., ed., New Views on American Economic Development [Cambridge: Schenkman Publishing Company, 1965], pp. 187–200.)Google Scholar For this reason we think the comparison of northern and southern price structures in 1840 based on conditions prevailing along the eastern seaboard—which we have referred to in the text—is more apposite for the purpose of adjusting the 1860 regional agricultural output ratio.

⁵² Note that if a higher relative price for cotton ruled than that indicated by the slope of PP in Figure 1, S-producers would move towards greater specialization in growing cotton and the comparison of aggregate outputs evaluated at the new (steeper sloped) price-line would show an even greater advantage for southern agriculture. It should be apparent from the text discussion that Fogel and Engerman's “efficiency” measures for different regions and farm-types within the South are subject to the same distorting influences caused by crop-mix variations.

⁵³ Cf. Vol. I, pp. 95–96 especially, where they refer to an equation (eq. 3.10 in Appendix B, Vol. II, p. 62) for predicting the course of slave prices on the basis of cotton prices and other information. The justification for this specification clearly has to be that there was enough land in crops other than cotton, in the short-run, and enough land clearable at constant costs per acre, in the long-run, to warrant supposing that slave prices would reflect expected (and actual) cotton price changes. Otherwise the latter would tend to be capitalized into movements in southern land prices.

⁵⁴ Despite gang wages 2.11 to 2.42 times as large as the earnings of sharecroppers (on 4 Louisiana plantations during 1865–1866, say Fogel and Engerman, “planters found it impossible to maintain the gang system once they were deprived of the right to apply force,” and sharecropping became the predominant mode through which southern Negro labor was mobilized. (Cf. Vol. I, pp. 238–239; Vol. II, p. 160.) Yet, as the source they cite for these Louisiana earnings differentials makes clear, during the two years in question the weather was miserable and the sharecroppers took a particularly bad financial beating. Moreover, it is noted there that the high wages set in these Louisiana parishes by the Freedmen's Bureau did succeed in recruiting black gang labor: in May, 1866 only 21 percent of the Negroes under the supervision of the Bureau in Louisiana were sharecroppers instead of working under various types of wage contracts. This “anomalous” preponderance of the system of paying wages for gang labor seems to have persisted in the Louisiana sugar-growing regions. Cf. Seagrave, C. E., “The Southern Negro Agricultural Worker: 1850–1870” (unpublished Ph.D. dissertation, Stanford University, April, 1971), pp. 41–42, 53–54.Google Scholar

⁵⁵ Vol. II, p. 19, emphasis added. Undoubtedly Fogel and Engerman are right in identifying the validity of this methodological claim as “the real question.”

⁵⁶ Vol. II, p. 222. Kenneth Stampp is charged with having thus “inadvertently obfuscated” the true immorality of slavery, and with having failed “to stress that proof of good treatment was insufficient to remove the moral brand” from the institution.

⁵⁷ It is the thesis of Fogel and Engerman's Ch. 5 (“The Origins of the Economic Indictment of Slavery,” Vol. I, esp., pp. 158–161) that the “moral purity” of the eighteenth-century radical Quaker position against slavery subsequently was diluted by the addition of an “economic indictment” constructed to meet the propaganda needs of nineteenth-century abolitionists. But as an essay in intellectual history this seems quite incorrect. Cf. D. B. Davis, The Problem of Slavery, pp. 291–309, 316–317, on the elements of rational analysis in Quaker theology, reflected in a willingness to examine Biblical texts in the light of reason and human standards of justice, as well as for the inclusion in early Quaker tracts of arguments concerning slavery's social and economic consequences. Davis (p. 317) explicitly cites an American Quaker pamphlet of 1713 as anticipating Hinton Rowan Helper's (1857) warning that Negro bondage would promote economic divergences and political conflicts between rich slaveholders and poor whites.