But the fence removed herders and dogs as well as predators, and over the course of the summer, “the sheep gradually became accustomed to free, unmolested grazing, and forgot the habits learned when herded” . They formed smaller bunches, both while grazing and also to sleep at night, and they moved more lightly over the landscape. “The result was that little or no damage was done to the forage crop in this way. The entire crop was eaten and not wasted” .By weighing twenty lambs “of average size” at the beginning and end of the experiment, and collecting similar data from bands herded on the range nearby, Jardine was able to compare animal performance under the two systems. In 88 days inside the fence, the pastured sheep gained an average of 20 pounds, whereas those from a band on the outside gained only 15 pounds, on average, in 96 days. Overall losses among four outside herds ranged from 1.4 to 3 percent, compared with just 0.5 percent for the fenced herd. Finally, the fenced sheep required only 0.0156 acres per sheep per day. Three open range herds, roughly estimated, required 64-123 percent more acreage, which Jardine attributed to the effects of trampling. “It is safe to conclude that range grazed under the pasturage system will carry 50 per cent more sheep than when grazed under the herding system, where the band is driven to and from camp each day” . In the closing pages of his report for the 1908 season, Jardine took up the economic question: “Will the pasturage system pay?” Here he faced a difficulty, because the coyote-proof pasture had been very expensive to construct: $6,764.31, to be precise, blueberry package including more than $2,000 in materials, $1,000 in transportation costs, and $1,000 to clear heavy timber from portions of the fence line .
This amounted to nearly $850 per mile . Jardine chose not to use these figures, however, arguing that the location was exceptionally remote and heavily timbered. Using more general estimates, he calculated that “the cost on most grazing lands will approach very closely $400 amile” . He then tabulated the financial benefits in increased carrying capacity, heavier sheep, reduced losses to predation, and lower labor costs. Not counting increases in the lamb crop and in the amount and quality of wool , Jardine arrived at an annual return of $746.50, based on a herd of 2,200 sheep in a 2,560-acre pasture for three months. Thus, an initial investment of $3,200 , at 8 percent interest and including maintenance, would pay for itself and begin yielding dividends after six years. He did not include the costs of the hunter in his analysis. The experiment was repeated in 1909 with 2,040 sheep enclosed in the pasture for 99 days, during which time only four perished. The results were nearly identical to those from 1908, although Jardine’s report was more detailed and more emphatic in its declarations of the virtues of “the pasturage system.” The hunter killed one grizzly, two badgers, seven brown bears and seven coyotes , and just one brown bear and three badgers managed to breach the fence . The sheep again displayed a gradual tendency “to depart from their old habits and accommodate themselves to the freedom of the pasture,” so much so that by the end of the season “it was almost impossible to keep them close bunched without a dog” . Losses to poisonous plants, as well as predators, were higher among sheep herded outside the enclosure, and the pastured sheep were again heavier at the end of the season than the herded sheep .
Acreage required per sheep perday was 52-90 percent higher outside the fence, although Jardine attributed some of this to poor quality herders . The issue of herder skill presented a puzzle, which Jardine acknowledged but failed to address directly. “A first-class herder will work all the time” and seldom use a dog, resulting in “quiet, scattered grazing that may approach the pasturage system in efficiency” , whereas “a lazy man…will wear out his dogs, worry the sheep, and destroy the forage” . With respect to weight gain, “there is as much difference in the results obtained by a first-class herder and those obtained by a poor herder as there is between the results under the pasturage system and those secured by the good herder” . And while “range grazed under the pasturage system will carry from 25 to 50 per cent more sheep than when grazed under the herding system,” it was also possible “that an excellent herder can, to a considerable extent, allow his sheep freedom and keep them quiet, thereby increasing the carrying capacity of his range. No doubt there are herders who do this” . All told, “the carrying capacity of the same range utilized by different herders may vary at least 25 per cent” . These were potentially troublesome admissions to make, for both scientific and economic reasons. Herder skill was clearly an important variable in sheep performance, but it was one that Jardine could neither measure nor control. Removing herders might thus be seen as necessary to a properly “scientific”assessment of range grazing. And it could clearly affect any calculation of the economic rationality of building fences and controlling predators. What if better training for herders were a more economical solution?
Instead of confronting these issues, Jardine reverted to general claims about labor needs under the pasturage system, superseding by half the estimate that King had given to Coville two years earlier: “It is probable that one energetic man…can properly care for four inclosures similar to the experimental coyote-proof pasture,” meaning “one man would care for from 8,000 to 10,000 head of sheep” . Notwithstanding these problems, Jardine reached the same conclusions in 1909 as he had the year before. “When left unmolested by herders and dogs in an area protected against destructive animals, a band of ewes and lambs will accommodate themselves to the freedom of such a system and will separate into small bunches, coming together occasionally but again separating. With few exceptions they will graze openly and quietly” . Losses will be slight, weight gain and wool clip will improve, carrying capacity will increase by 25-50 percent, and labor costs “will not exceed 25 per cent of the cost under the herding system” . Although he did not present numbers for 1909, he again concluded that “the increase in carrying capacity and decrease in expense of handling in pasture during the lambing season will justifly the cost of construction necessary to inclose [sic] the entire allotment” . The Wallowa experiment was hailed as a remarkable success, and the Forest Service quickly embraced it as guidance for the administration and management of rangelands generally. It appeared to solve numerous problems and satisfly everyone, provided one ignored or excluded the herders. In his “Annual Report to the Forester for the Fiscal Year 1909-1910” for the Branch of Grazing, Potter described the results as “very gratiflying” and summarized them as follows: “The primary objects of the experiment have been accomplished, i.e., it has been demonstrated that the grazing capacity of the Forest lands can be largely increased by improved methods of handling stock, and that the increased cost of such methods, if any, is offset by increases in the number and weight of lambs raised, heavier wool crops, and reduced losses from predatory animals” . Notably, Potter omitted any reference to labor costs in his summation. He suggested that the results be applied “to spring and fall or yearlong ranges” in other national forests. Elsewhere in his report, Potter tabulated the accomplishments of Forest Service personnel assigned to predator control: 269 bears, 129 wolves, 148 wolf pups, 1,155 wildcats, and more than 7,000 coyotes killed in the 11 western states, “an increase of 109 per cent over the number of animals destroyed last year” and representing “a total saving to stockmen of considerably more than one million dollars per year” .A critical physical geography of the Coyote-Proof Pasture Experiment reveals two important insights. First, blueberry packaging the institutional context in which science is practiced may be at least as important as the experiments and findings that the scientists produce. Livingstone has shown the importance of place to the conduct of science, and the geographical particulars—both social and ecological—of the Wallowa site attracted Coville to locate the study there. But the larger context was a national one, and in this case, not only did it shape the questions that were asked and define the terms of success, but the success caused the institutional context itself to change, setting in motion a path dependency for subsequent research and management of rangelands elsewhere.
The particularities of the Wallowa site were abstracted away so that the results could be taken as relevant to rangelands throughout the western US. Second, although the findings of the experiments were presented as scientifically robust, ecological facts—in which predators, livestock, fences, and vegetation interacted in measurable ways—the methods were weak and the ultimate metric for evaluating success was actually an economic one. Costs and returns, and thus profit on investment, determined whether fencing and predator control were worth implementing. In this calculus, the decisive factor was neither fences nor livestock performance but rather the labor of herders. The high cost of fencing could be justified economically only if the fences greatly reduced the need for herders—and in the absence of herders to protect livestock, predators would have to be rendered effectively insignificant. What’s more, even the economic analysis was flawed, with key costs either minimized or excluded in order to reach the desired conclusion. By today’s scientific standards, the Coyote-Proof Pasture Experiment was far from impressive. The methods varied in several ways from one year to the next, and they were reliant on qualitative assessments or herder accounts for some important data. The findings were confounded by sheep breeds, herder practices, vegetation types and other variables, and no direct attempt was made to assess the impacts of the pasturage system on vegetation. Moreover, there were no real controls by which to judge the relative effects of the enclosed pasture, the absence of predators, and the absence of herders or dogs as factors influencing the dependent variables . Estimating the carrying capacity outside the fence was imprecise, since those herds were not confined within fixed boundaries. Finally, the economic calculations excluded the costs of the hunter, and didn’t account for the actual costs of the fence. Virtually every finding looks suspiciously similar to the expectations that Coville carried into the experiment in 1907. Moreover, Jardine’s reports were not subject to peer review, but only to the scrutiny of his superiors in the USDA, who themselves appear to have pre-judged the results. There is almost no chance that the experiment would be recognized as publishable, or even scientific, if it were conducted today. The experiment succeeded not on the basis of its scientific rigor, but instead because it lent authority to ideas that were already viewed favorably within the institutional context that gave rise to it. Coville and Jardine produced a set of knowledge claims that appeared to conform to scientific norms of experimentation: the deliberate manipulation of objects, organisms and people, and the careful recording and interpretation of actions and reactions among them. Most of these primary data were of a broadly ecological character, and thus appeared as apolitical and “objective.” The results were translated into economic terms to assess the practicability of implementing a similar management regime on Western rangelands as a whole, and if the economic analysis was at once flawed and decisive, this contradiction would be resolved by government largesse: the cost of both fencing and predator control would be subsidized by an array of federal agencies over the decades to come. Predator control occurred throughout the national forests and beyond, much of it organized and funded by the Bureau of Biological Survey under federal legislation passed in 1914; fencing was underwritten by the Forest Service, the General Land Office’s Grazing Service, and the Civilian Conservation Corps, whose crews built thousands of miles of fences between 1933 and 1942. Much simpler and cheaper, four-strand barbed wire fences were used—rather than the elaborate Wallowa design—as hunting, trapping, and poisoning reduced predator populations region-wide. If both fencing and predator control were already planned or ongoing, and might well have proceeded without “scientific” support, then the greater significance of the Coyote-Proof Pasture Experiment lies in the ways it altered the institutional context itself, and thereby the trajectory of rangeland administration and research.