I am indebted to William N. Parker, James D. Foust, W. David Maxwell, and Judith Klein for useful comments on this paper. The research also gained from suggestions made at the Sixth Purdue Conference on the Application of Economic Theory and Quantitative Techniques to Problems of Economic History and at the seminar in economic history at the London School of Economics. Acknowledgment is made to the Ford Foundation through the Inter-University Committee on American Economic History and to Indiana University for financial support.

¹ Recent quantitative research has begun to fill this gap in our knowledge. See, for example, Parker, William N. and Klein, Judith L. V., “Productivity Growth in Grain Production in the United States, 1840–60 and 1900–10,” Output, Employment and Productivity in the United States After 1800, National Bureau of Economic Research, Studies in Income and Wealth, Vol. 30 (New York: Columbia University Press, 1966), pp. 523–82.Google Scholar

² Barger, Harold and Landsberg, Hans H., American Agriculture, 1899–1939 (New York: National Bureau of Economic Research, 1942), p. 27,Google Scholar and Strauss, Frederick and Bean, Louis H., “Gross Farm Income and Indices of Farm Production and Prices in the United States, 1869–1937,” U.S. Department of Agriculture Technical Bulletin No. 703 (Washington: U.S. Government Printing Office, 1940), p. 25.Google Scholar

Throughout much of the nineteenth century dairying was a primary activity on relatively few farms but it was a minor one on most. Nationally, dairying involved the labor of a large portion of the American farm population during the 1850–1910 period. In 1850 an estimated 500 million man-hours were expended on direct labor in dairying; by 1910 about 2.7 billion man-hours were so employed.

³ Bateman, Fred, “Improvement in American Dairy Farming, 1850–1910: A Quantitative Analysis,” The Journal of Economic History, XXVIII (June 1968), 255–73.CrossRefGoogle Scholar

⁴ Capital was also relatively “scarce” in nineteenth-century America, but mechanical devices can be capital-saving as well as labor-saving, of course, if for no reason other than the greater speed they afford. In fact, unless the elasticity of factor substitution is zero, any “technical advance” will produce absolute, though not necessarily proportionate, savings in both capital and labor requirements per unit of output. Furthermore, much farm capital was built by farm labor using large inputs of the abundant natural resources as raw material. See Primack, Martin L., “Land Capital Formation as a Use of Farm Labor in the United States, 1850–1910,” The Journal of Economic History, XXVI (Sept. 1966), pp. 348–62,CrossRefGoogle ScholarPrimack, , “Farm Construction as a Use of Farm Labor, 1850–1910,” The Journal of Economic History, XXV (Mar. 1965), pp. 114–25,CrossRefGoogle Scholar and Primack, , “Land Clearing Under Nineteenth-Century Techniques: Some Preliminary Calculations,” The Journal of Economic History, XXII (Dec. 1962), 484–97.CrossRefGoogle Scholar

⁵ Quantitative research into sources of productivity growth in the small grains corroborates this point. Parker and Klein, “Productivity Growth,” pp. 523–82.

⁶ Bateman, “Improvement in Dairy Farming.”

⁷ To what extent the dearth of scientific improvement resulted from lack of scientific knowledge, lack of economic incentives, or some combination of both is not completely clear. The earlier development of dairy breeds in Europe supports the hypothesis that the lack of economic incentives in the United States accounted for the slower development there. This no doubt is largely true. On the other hand, much of the European breed improvement had apparently no theoretical scientific foundation, but was simply based on empirical observation. English farmers, for example, evidently recognized the importance of a high-grade bull for the breeding process long before most American farmers. Of course, one can make a strong argument that this largely reflects the differing economic importance of dairying in these two countries. Frequently the potential benefit from better breeds imported from England was lost because American farmers simply were unaware of this empirical knowledge. See Chambers, J. D. and Mingay, G. E., The Agricultural Revolution (New York: Shocken Books, 1966), pp. 12–13;Google Scholar U.S. Department of Agriculture, Yearbook of Agriculture, 1936 (Washington, 1937);Google ScholarTransactions of the Wisconsin State Agricultural Society, 1852, Vol. II, p. 228;Google ScholarCorbin, John P., Practical Hints on Dairying (Whitney's Point, New York, 1871), p. 12;Google ScholarLeavitt, Charles T., “Attempts To Improve Cattle Breeds in the United States, 1790–1860,” Agricultural History; VII (Apr. 1933), especially pp. 64–5;Google ScholarPirtle, T. R., History of the Dairy Industry (Chicago: Majonnier Bros. Company, 1926),Google Scholar chap. II, and Ernle, Lord, English Farming, Past and Present, 4th ed. (London: Longmans, Green and Company, Ltd., 1927), chap. VIII.Google Scholar

⁸ But, “… while no milking machine yet invented has shown its practical value in a way that has led to its general use …,” said a USDA Bulletin in 1907, “… the prospect for the general introduction of milking machines seems to be so favorable [in 1907] that it was thought to be advisable by the Secretary of Agriculture to authorize an investigation or the fundamental problems involved in the use of such machines.” The result was publication of a detailed 55-page Bulletin by the Bureau in January 1907. See Lane, C. S. and Stocking, W. A., The Milking Machine as a Factor in Dairying: A Preliminary Report, U.S. Department of Agriculture, Bureau of Animal Industry Bulletin No. 92 (Washington: U.S. Government Printing Office, 1907).Google Scholar The quotation above is from page 10 of this Bulletin. The realization of the imminent introduction of machine-making at this same time is shown also by the issuance of the state experiment station bulletins in 1906. See Erf, Oscar, Milking Machines, Kansas Agricultural Experiment Station Bulletin No. 146 (1906)Google Scholar and Smith, G. A. and Harding, H. S., Milking Machines, New York State Agricultural Experiment Station Bulletin No. 353 (1906).Google Scholar

There were approximately 12,000 machines of various types in use on U.S. farms by 1910, 55,000 by 1920, and 100,000 by 1930. Only during and after World War II, however, did milking machines come into widespread use. U.S. Bureau of the Census, Historical Statistics of the United States, Colonial Times to 1957 (Washington, 1960), pp. 284–85.Google Scholar

⁹ Less seasonal than most farm enterprises, dairying remained in the twentieth century a large consumer of the farmer's time throughout the year. In 1937–1940, for example, American farmers expended an estimated 3.3 billion man-hours annually on milk cows alone, plus an additional 3.6 billion hours caring for dairy calves, heifers, and bulls. This amount exceeded that spent on cotton, corn or wheat, or oats during this period. Elwood, Robert E., Lewis, Arthur A., and Strubel, Ronald A., Changes in Technology and Labor Requirements in Livestock Production: Dairying, Works Progress Administration, National Research Project, WPA Report No. 2–14 (Washington, June 1941), p. v. (Hereafter cited as NRP Study).Google Scholar

¹⁰ Said the 16th Annual Report of the Bureau of Animal Industry in 1899: “In one respect, dairy labor is the same as a hundred years ago. Cows still have to be milked by hand … twice a day … This is one of the main items of labor in dairying …” U.S. Department of Agriculture, Bureau of Animal Industry, 16th Annual Report (Washington: U.S. Government Printing Office, 1899), pp. 261–62. (Also see fn. 8 above).Google Scholar

¹¹ This has significant implications, discussed below, for the course of dairy labor productivity since farmers typically were choosing relatively low-yielding animals even during the early twentieth century. See, for example, S. W. Mendum, “Cost of Milk Production on Forty-eight Wisconsin Farms,” U.S. Department of Agriculture Bulletin No. 1144 (Washington: Mar. 20, 1922), p. 11;Google ScholarSteven, D. H., “Economic Studies of Dairy Farming in New York: XI,” New York (Cornell) Agricultural Experiment Station Bulletin No. 562 (Ithaca: June 1933), p. 21;Google Scholar“Dairy Farm Organization in Southeastern Kansas,” Kansas Agricultural Experiment Station Bulletin No. 255 (Manhattan, Kan. Feb. 1931), p. 24;Google ScholarMoore, J. S., “Feed, Care and Management of the Dairy Cow,” Mississippi Agricultural Experiment Station Bulletin No. 259 (A&M College, Miss.: Jan. 1928), p. 10;Google Scholar Washington Agricultural Extension Service Bulletin No. 173 (Pullman: Nov. 1922), p. 45.Google Scholar Variations in annual input were largely attributable, claims the author of the last bulletin, to the organization of the work routine which reduced duplication and inefficiencies in labor use.

In 1910, the U.S. Department of Agriculture compared labor requirements with hand-milking and with machines on approximately 270 farms in New York, Ohio, Michigan, and Illinois. The results of the experiments showed that with a herd of 15 cows or less, a “fraction over seven minutes” was required to milk a cow by hand. With herds exceeding 50 cows, the required time was “slightly under seven minutes.” In most other nineteenth- and twentieth-century sources, the time to milk one cow by hand was found to be approximately 7 to 7.5 minutes, with milking time for herds smaller than 10 or 15 cows (the predominant size in the U.S.) near 7.5 minutes. This experiment further indicated that there are not substantial economies of size in large herds where hand-milking is used. Even with milking machines, the reduction in time as herd size increased was only 17 percent (5 minutes compared with 4.15), but substituting machines for hand-milking reduced labor by about 33 percent (from 7.5 minutes to 5); U.S. Department of Agriculture Bulletin No. 423, “Labor Requirements of Dairy Farms as Influenced by Milking Machines,” by Humphrey, Harold N. (Washington: U.S. Government Printing Office, 1916).Google Scholar

¹² The reasons for the extension of the annual period of milking and the sources used for estimating these data are discussed in Bateman, “Improvement in Dairy Farming,” pp. 268–72. Most of the agricultural literary evidence that formed the basis for the yield and labor estimates was found in the National Agricultural Library and its annex in Washington, D.C.

¹³ Three variables are involved: (a) length of milking season; (b) time required to milk one cow each day of the season; (c) total man-hours spent on milking. Variable (b) is based on nineteenth-century sources. Since (a) and (b) quantities are determinable for 1910–1930, it is possible to check roughly the validity of our (c) estimate for this later period to discover whether or not (b) had remained relatively unchanged as was assumed in the text above. Using 80 to 85 annual man-hours as the value of (c) and 325 to 330 days as the value of (a) produces a daily labor requirement of 15 minutes or 7.5 minutes per milking. (The value used for (c) was 82.5; for (a), 330 days). This would strongly suggest that although (a) changed between 1850–1860 and 1910–1930, (b) remained fairly constant (unless a milking machine was used). Since (c) must be estimated for 1850, one must have accurate values for (a) and (b). The value of 15 minutes was used for (b) with an historically supported value for (a).

¹⁴ One aspect of feeding which did not change during this period was its daily frequency. Few nineteenth-century sources, including state agricultural board reports, farm journals, and U.S. Department of Agriculture publications, found farmers who fed more than twice daily. Feeding more frequently than this was rare, and little variation is found on this subject. For example, the Transactions of the Washington Dairymen's Association claimed in 1903 that most dairy farmers fed twice daily. A New York farm journal said in 1861 that even “premium dairy farms” in that state fed only twice each day. The Dairy Farmer, I (Feb. 1861), 292.

¹⁵ For one product of these investigations, see District of Columbia Health Department, “Regulations for the Government of Dairies and Dairy Farms” (1897). This document contained seventeen sections regarding cleanliness of barns, cows, utensils, and milkers. Similar recommendations can be found in U.S. Department of Agriculture, Bureau of Chemistry Bulletins 81, 90, 99, and 105 (Washington, 1904–1907).Google Scholar

¹⁶ So strongly did a leader of the Western Holstein Dairy Association feel about the effect of sanitary rules on dairy costs that he stated in 1909 that, “… it means the time has passed when a poor man can start in the dairy business because he has to have everything reasonably sanitary and this all takes money and a lot of it …” Colorado Dairyman, I (Mar. 1910), 6–7.

¹⁷ NRP Study, p. 72.

¹⁸ Hopkins, “Some Correlations” (Part II), p. 228. Hopkins found, “Slightly more labor is required to feed and handle the milk of a cow of large capacity than to perform the same operations for one of low capacity, but most of this increase in labor may be attributed to the greater conscious effort to take the best possible care of the cows of a productive dairy as contrasted to the general lack of interest usually found to accompany poor cows.”

¹⁹ Concluded one writer after an extensive study: “There is a tendency for the amount of labor used per cow to increase, as more concentrates are fed …”; Hopkins, John A. Jr, “Some Correlations in the Cost of Milk Production,” Journal of Dairy Science, II (Mar. 1919), 83.Google Scholar Also see Delaware Agricultural Experiment. Station Bulletin No. 118. “… there is a tendency for the amount of labor used to increase as more forage is fed. This is partly, but not wholly, caused by the longer time required to handle the greater amount of forage.” Hopkins, “Some Correlations” (Part II), p. 213.

²⁰ Walter M. Wilcox and Emil Rauchenstein, “The Effect of Size of Herd on Milk Production Cost,” Journal of Farm Economics, XXX (Nov. 1948), 722. The authors say “without doubt, on the average, the larger dairy herds enjoy better management than small herds. This is true in each of the important dairy states as well as for the United States as a whole.” (p. 722).

²¹ This is not true of the actual (hand) milking of an animal, where the labor varies directly with the number of cows. There is a discussion of this in Minnesota Agricultural Experiment Station Bulletin No. 205 (St. Paul: Nov. 1923), p. 70. With machine-milking there is a time-saving in large herds.Google Scholar

²² See, for example, Minnesota Experiment Station Bulletin No. 161 (St. Paul: Aug. 1916), p. 29;Google Scholar Washington Experiment Station Bulletin No. 173 (Pullman: Nov. 1922), p. 47;Google Scholar Montana Experiment Station Bulletin No. 264 (June 1932), p. 20; New York (Cornell) Experiment Station Bulletin No. 554 (Ithaca: Apr. 1933), p. 62;Google Scholar New York (Cornell) Experiment Station Bulletin No. 562 (Ithaca: June 1933), pp. 17–18.Google Scholar

²³ All figures are from U.S. Bureau of the Census, Census of 1910, Vol. 5, Agriculture, p. 485. Even in the Western Dairy region as a whole, the average herd consisted of only 6.7 animals. Although relatively few American dairy herds attained a substantial size before 1910, apparently an ‘impression” of size diseconomies prevailed among New York's specialized dairy farms even during the mid-nineteenth century. This was discussed in The Dairy Farmer, Vol. I (Little Falls, New York, A. W. Eaton, publisher, 1861), p. 232.Google Scholar Said X. A. Willard, in this article: “The impression seems to prevail, that a herd of fifty cows is the maximum—above this point it is said that the expenses for labor—the inconvenience and many other causes, conspire to render the profit on the additional cows, much less in proportion. Most persons, therefore who have a large number of cows, and lands lying convenient for the purpose divide up their stock into dairies of forty and fifty cows.”

²⁴ See Bateman, “Improvement in Dairy Farming,” pp. 259–60. For a detailed discussion of the progress of dairy commercialization during this period, see Lampard, Eric E., The Rise of the Dairy Industry in Wisconsin (Madison: The State Historical Society of Wisconsin, 1963), Parts I and II.Google Scholar

²⁵ Average pasture season data were obtained from U.S.D.A. publications.

²⁶ For feeding, the lower limit in the sample was about 2.2 minutes daily per cow on northern farms during the early twentieth century. This represented an average for the entire year; thus one would expect the average during the pasture season to be less than this, the winter to somewhat exceed it. See U.S. Food Administration, Hearings of the Federal Commissioner for the Middle States (Washington, 1918), pp. 427–28. The yearly average was about one minute per day; the lower limit of 0.35 minute per cow daily was found in Michigan Agricultural Experiment Station Bulletin No. 277 (1916), p. 10. The average time for stable-cleaning in the early twentieth century was about 3.5 minutes per cow each day. The lower limit (an average for the entire year including the pasture season as well as winter) was 1.5 minutes. U.S. Food Administration Hearings, pp. 427–28, and New Mexico Agricultural Experiment Station Bulletin No. 215 (1933), p. 39. For a more extensive discussion of the lack of attention given dairy animals on U.S. farms and the domestic character of dairying at the mid-nineteenth century, see Lampard, The Rise of the Dairy Industry in Wisconsin, Chap. 3.Google Scholar

²⁷ Bateman, “Improvement in Dairy Farming,” pp. 257–58.

²⁸ No reliable estimates of changes in dairy feed quantities have yet been developed for this period. On the basis of preliminary work, however, it is possible to conjecture that the larger quantities and better quality of feed did not increase yields proportionately more than they increased labor, mainly because the new methods were more time-consuming than ones they replaced, (see fns. 16 and 17).

²⁹ See discussion in Bateman, “Improvement in Dairy Farming,” especially pp. 257–70.

³⁰ For basis of this assertion, see Bateman, “Improvement in Dairy Farming,” fns. 18 and 22, pp. 267–68.

³¹ They were responding on the demand side to the stimulus occasioned by the growth of incomes and local markets in the economy after 1850. There were also influences on the supply side other than with labor. The real cost of nonpasture feed was declining in some areas as productivity rose in grain production. (See Parker and Klein, “Productivity Growth.”)

³² The labor supply on many family farms consisted, of two groups between which existed a nonsyrnmetrical relationship of labor substitutability. Members of one group were capable of performing either heavy operations on major field crop production or lighter work on supplementary activities such as dairying; those in the other group (older family members and young children, for example) could perform only the lighter work. The second group was of fixed availability to the farm and “costless” from the viewpoint of production of the major field crops which required a different quality of labor and which generally accounted for most of the total output on mixed farms. The farm family's limited supply of money capital would therefore better be applied to substituting capital for labor in field crops—if a choice of where to make such a substitution had to be made—rather than in dairying because of the higher relative real cost of labor in nondairy activities.

³³ The ease with which dairy farmers could become victims of an illusion about dairy profitability was noted in an 1878 Kentucky agricultural report:

“In 1863–64, the associated dairy system had become an established fact [in the East]. Somehow the impression became general among the farming community that the dairy was reaping enormous profits.” Kentucky, First Annual Agricultural Report (Frankfort, 1878), p. 152. Also see comments in fn. 34.Google Scholar

³⁴ During the nineteenth century, state board of agriculture reports, U.S. Department of Agriculture studies, farm journals, and similar sources repeatedly decried the farm community's ignorance about dairy operations, about methods of efficiently organizing the dairy, and about performance records of dairy animals. For example, in the “Lectures at the Vermont Dairy School in January and February 1894”, it was said, “It is safe to say that one fourth of the milk cows of the country are kept at a loss …” Vermont Abstract of Lectures at the Vermont Dairy School, Lecture Course I-XXII, 1894. Also see Gurler, H. B., American Dairying (Chicago: 1894), p. 20,Google Scholar where he claims that given the average yield of American cows, there was no area where this level of output could be made profitable. Writing in 1899, Henry Alvord, an Agriculture Department dairy expert, felt that the minimum profitable yield of a dairy cow was 4000–4500 pounds annually. Sixteenth Annual Report of the Bureau of Animal Industry (Washington, 1899), p. 255.Google Scholar But even in 1910, the national average yield was only 3570 pounds, implying that given Alvord's standard, considerably more than half of American dairy cows were unprofitable. A similar minimum-profitable average yield was expressed in other sources. See, for example, The Cheese and Dairy Journal (Whitewater, Wisconsin, September 1904), pp. 1 and 7,Google Scholar where the comment is made that, “Many cows do not even pay for their food. One of the chief reasons … is because of the great number of scrub cows in the country …” Also see Collier, Peter, How To Make Dairying More Profitable (New York: 1890).Google Scholar He criticized the unprofitability of most American dairy animals, the farmers’ general ignorance of dairy management, and chided the dairymen for not running their dairies on a “business-like” basis. He sent out 1500 questionnaires “into every section of the state, asking for information which every dairyman should have been able to give …” One of the replies from a man who tried to get individual farmers to complete the questionnaires was: “… I find on inquiry that there is not one man in a hundred of the average farmers that knows anything knows anything about what his cows except that they feed them something and get some milk and butter. But I sincerely hope we may see a change in the management of cows for profit, and that the time will come when farmers will keep a record of their cows and know when they get a profit…” (p. 12).

³⁵ In his discussion of the high (relative to Britain) “floor” to United States manufacturing wages in the nineteenth century, H. J. Habakkuk points out that most American small farmers, unlike their European counterparts, received not only a wage payment for their labor services, but land rent and appreciation and a profit return. Habakkuk, H. J., American and British Technology in the Nineteenth Century (Cambridge: Cambridge University Press, 1962), pp. 13–14.Google Scholar Receipt of these.implicit returns, however, could also create a situation in which the returns might fall to zero or become negative with farmers absorbing the implicit losses unwittingly instead of reallocating their resources.