Classic Reviews in Economic History

Abbott Payson Usher, A History of Mechanical Invention. New York: McGraw-Hill, 1929. xi + 401 pp. (Revised edition, Cambridge, MA: Harvard University Press, 1954, 450 pp.)

Review Essay by George Grantham, Department of Economics, McGill University.

How Economic Change Happens: Usher’s History of Mechanical Invention

Among the seminal works in economic history fewer are more perplexing than Abbot Payson Usher’s History of Mechanical Invention. The bland title offers no suggestion of a great ambition, which is nothing less than to establish logical foundations for an empirically based explanation of economic change, the prose is stern and unrelenting, and like a car that runs out of gas just before reaching its destination, the book simply comes to a stop with no conclusion,. Few historians consult the work today. Once ransacked for information on the early history of clocks, windmills, textile machinery, steam engines and machine tools, its encyclopedic function has been superseded by more accessible and up-to-date compilations. So why should we be tempted to study it now? What gain repays the effort required to master the technical minutia of several branches of mechanics and the erudite byways of classical and medieval scholarship? The main reason is that, along with Kuznets’ studies in historical income statistics, The History of Mechanical Invention is a founding text of a science dedicated to explaining economic change, what Usher called the “mutual transformations taking place between human societies and their environment.”

We begin with the pesky problem of how to tell that story. At the time Usher was composing the first edition of Mechanical Invention (1929), the narrative of general economic history was dominated by the “stages” approach, according to which the development of individual economies is displayed in a chronological sequence of conceptually distinct types. Conceiving economies as identifiable types goes back to generalizations proposed by the ancient Greeks to interpret the customs of the strange peoples they encountered in Asia and the European hinterland. In the eighteenth century the notion received a fillip from speculations attributing the evident segregation of human societies by type to adaptation to different geographical conditions. Adam Smith’s four-fold classification of societies into hunting and gathering, pastoral, agricultural and commercial economies is an unexceptional instance of this reasoning. As individual types were considered to reflect static environmental constraints, the typology contained no chronological implications, so that although Montesquieu, Smith and Turgot certainly believed that commercialized societies represented an “advanced” state of civilization, they held no strong view that it represented the latest phase of an historical sequence. Indeed, the conviction that men are physically and psychologically similar and the great prestige of Roman civilization stood in the way of a progressive narrative of social states. Reason is timeless.

In the hands of early nineteenth-century philosophers exalted by the Romantic concept of becoming, that static conception of social types acquired a temporal dimension. To position societies on the thin line of Time’s Arrow, however, implied that they are discrete entities historically expressing ontogenetic development that is independent of the particular environment. But exactly what force causes such entities to cohere and persist, and drives their historical development? The cause could not be definitely stated, any more than one could then explain ontogenetic development of living organisms.[1] Whatever it was — life-force, God’s will, the national Geist, it was ineffable. It could be felt, appreciated and asserted, but not explained. All that could be declared with any degree of confidence was that each society develops through a sequence of stages marked by increasing complexity of organizational forms, methods of production, degrees of regional and occupational specialization, and movement from small to large units of social and economic organization.

In the different versions supplied by successive generations of German historical economists and American Institutionalists the stages approach provided a serviceable framework for characterizing the range of societies revealed by geographical discovery and the general trend of European development since the early Middle Ages. It was a capacious tent within which several generations of economists and historians were able to get on with the business of investigating the evolution of the myriad institutions and activities that constitute an economy without having to worry much about what it all meant.[2] Yet despite conjectures that recall elements of the New Institutional Economics, the stages “theory” offered little in the way of a systematic interpretation of how particular societies interact dynamically with their environment. It did not explain how things change.

Much the same can be said of the empiricist tradition exemplified by Clapham’s Economic History of Modern Britain (1926). Making abundant use of contemporary statistical sources, Clapham aimed at correcting catastrophic accounts of Britain’s industrial revolution concocted from impressionistic sources by asking the quantitative questions, how much, how often, and for how long. The overall impression left by this monumental exercise in error correction was that one can draw few generalizations beyond the fact of the geographic diversity of England’s nineteenth-century economic experience. To the question, what happened in history, Clapham answered, many things to many people in many places. A thoughtful reviewer characterized the work as a tour book laced with numbers.[3] As Usher observed in a second review, that criticism was unfair. Clapham had an implicit model of England’s industrial transformation, but left it to the reader to parse it out for himself.[4] Yet, although honest enough, this leave-it-to-Beaver approach to historical synthesis was hardly the stuff of a science capable of building on past achievements. As Darwin had observed sixty years earlier, a fact is not neutral; it is either for or against an argument. Clapham refused to argue.

Neither Clapham’s sprawling narrative nor the ethereal holism of the stages approach adequately addressed the main question of how things change. Kuznets eventually resolved the problem posed by Clapham by reducing quantitative indicators of output to a scalar measure of economic activity that tracks the flows connecting income with aggregate saving and expenditure. That synthesis rests on well-understood accounting principles permitting one to speak intelligibly of an economy in time. It utilizes a measure that can be statistically decomposed to its proximate causes and even unto the causes of those causes. By contrast, synthesis offered by the stages vision of economic change achieved only an aesthetic coherence, where the significance of particular facts depended on their relation to an a priori compositional scheme. As Usher noted, the rhetorical persuasiveness is generally secured by “discreet omissions.” The notion that “history,” economic or otherwise can be described as the movement of a holistic entity implies the existence of an immanent principle determining the whole course of events, which makes it little more than a thinly disguised Natural Theology, where, as the Austrian novelist Robert Musil once said of allegory, “everything takes on more meaning than it honestly ought to have.”[5]

Particular Systems of Events

Usher’s answer to holistic history was to restrict analysis of historical causation to sequences of events for which temporal connections can be empirically demonstrated.

We ought not to say that the present is derived from the past and the future from the present. The proposition must be formulated in much more specific terms: every event has its past. The principle of historical continuity does not warrant any presumption about the relations among events occurring at the same time. This assumption is very frequently made, but it will be readily seen that it is not warranted (1954, p. 19).

Usher termed the sequences for which historical continuity can in principle be verified “particular sequences of events.” Such sequences are distinct from series of events resulting from similar responses to similar situations, such as the predictable responses of economic agents to changes in the environment, because a narrative adds nothing to our understanding of them. As Paul Veyne observes,

If the revolutions of people were as entirely reducible to general explanations as physical phenomena are, we would lose interest in their history; all that would matter to us would be the laws governing human evolution; satisfied with knowing through them what man is, we would omit historical anecdotes, or else we would be interested in them only for sentimental reasons, comparable with those that make us cultivate, alongside great history, that of our village or of the streets of our town.[6]

Objects of historical thinking acquire meaning from their place in a plot that explains them. Usher held that innovation was the critical element of such plots, because it adds something that could not be predicted from initial conditions and therefore has to be explained by links to events preceding it in time.

A particular system of events must therefore be shown to be a truly genetic sequence. It must rest upon one or more acts of innovation that have been preserved by tradition and developed by further innovations (1954, p. 48).

Invention, then, makes up an intrinsically historical element in a series of events. It cannot be predicted with certainty ex ante, but it can be explained ex post as a narrative of verified acts. The History of Mechanical Invention proposed just such a narrative.

How does one identify a particular sequence? What principles make them distinct objects of empirical investigation? Several alternatives suggest themselves. One might distinguish events by their goal or purpose. Usher doubted that this principle could be applied to technological sequences because there are too many ways to skin a cat. The boomerang, bow and arrow, and blow-gun all kill game at a distance, but because they do not belong to the same technological system of events, none could plausibly have evolved out of any of the others. The presence of a common scientific principle suggests a better defining principle, but general principles may be too broad to define relevant boundaries identifying the particular sequence. Steam engines and steam turbines both exploit the expansion of steam to transform heat into mechanical energy, but they employ radically different mechanical means of doing it. Reciprocating piston engines descend from a pneumatic technology that originated in the hand pump; the genealogy of the turbine starts with the horizontal water wheel. The same is true of the transmission of motion by gear trains. While the devices invoked common mechanical principles in watches, clocks, and heavy equipment, the particular problems facing inventors differed so much from one application to another that the historian needs carefully to specify the context to explain the path of invention in each class of application. The boundaries of particular systems of technological events are thus narrower than their underlying scientific principles might suggest. Usher believed that the determination of those boundaries is ultimately an empirical question, as the boundaries have clearly widened over time as a consequence of advances in pure and applied science.

Systems of events consume time. Economists are not much concerned about the logical status of time except insofar as it serves as the metronome for growth theory. Not everything happens at the same time, however, and particular systems of events unfold at different speeds. One does not see that bicameralism, coitus interruptus, the mechanics of central taxation, the detail of rising lightly on one’s toes when uttering a subtle or strong sentence (as M. Birotteau did), and other events of the nineteenth century must evolve with the same rhythm.[7]

Usher held that intelligible history is necessarily pluralistic. Particular sequences, which we currently call paths of temporal dependence, demand separate treatment to track down cause and effect. A subtler problem concerns the historian’s temporal perspective. Usher insisted that particular events should not be conceived as constituting the “end” of a sequence.

Historical sequences do not have terminal points. To understand the significance of Watt’s engine is to place it in a series of events that extend backward to sixteenth-century investigations into the vacuum pump and forward towards the Corliss engine.

The Emergence of Novelty

The heart of the matter is how new things happen. By what intellectual and social processes do new methods of production, new products, and new patterns of behavior become objects of choice in the stream of economic and social life?

Historians traditionally answered this question in two ways. The first was that inventions are inspired intuition given to exceptionally gifted persons. This approach stressed the discontinuity of inventions and the importance of a small number of inventors in creating the modern world. Usher deemed it “transcendental,” because in taking invention to be what amounts to a miracle, it puts the event logically outside time, so that it can have no mere historical explanation. The second approach took the opposite tack of holding that inventions occur continuously in small steps induced by the stress of necessity, somewhat like Darwinian evolution.[8] Usher termed this approach “mechanistic,” because it relegated the inventor to the status of “an instrument or an expression of cosmic forces.”[9] Neither the transcendental nor the mechanistic account of invention, then, was historical in the sense that explanation necessarily takes the form of a narrative. To the transcendentalist, inventions just happen (and we should all be grateful they do); to the mechanist, they occur automatically in the fullness of time. Neither explains how inventions happen.

Invention is an event in the mind, so an empirically grounded model of invention should be based on its cognitive properties. The properties that Usher found most useful in this respect are drawn from the findings of Gestalt psychology, which in the 1920s was a thriving field of experimental research. Gestalt psychology proceeds from the observation that the mind commonly perceives things as wholes rather than as a chaotic flux of sensory stimuli. That perception or gestalt, however, is not an ex post “interpretation” of the stimuli; it is how they are literally “seen,” what Wittgenstein called a “particular organization” of sensory (visual) experience.[10] The physiological basis of this well-documented phenomenon stems from evolutionary adaptations in neural circuitry that enhanced the capacity of early hominids to quickly extract signals from a perceptually noisy environment. As those adaptations took place prior to acquisition of language, gestalt perception does not obey the cognitive constraints of propositional logic embedded in language, but conforms to the spatial logic of pictorial composition, in where things take meaning from their “fit.”[11] Because of this a given stimulus can generate more than one true perception. For example, in the classic “figure-ground” form, we may see a black goblet against a white ground, or alternatively two white heads staring at each other across a black field, but never both at the same time. As the philosopher Russell Hansen put it, “There is more to seeing than meets the eyeball.”[12]

Usher contended that invention is seeing a “particular organization” of data present in the inventor’s mind. The gestalt paradigm opens the door to an historical treatment of invention, because what we see is influenced by our past experience, which is to say, our history. Darwin confessed that he saw the “plainly scoured rocks, the perched boulders, the lateral and terminal moraines” on his geological rambles through the mountains of North Wales, but he did not see what Agassiz had seen in Switzerland: that the eskers and eccentric boulders were the product of glacial transportation. [13] What we know limits what we are able to “see” at any point in time. That constraint imparts directionality to discovery because in time we come to know more things. But that directionality raises a further question. What happens when we see something no one has ever seen before, which by definition we do not know? In the figure-ground experiment, could we recognize the goblet rather than the faces if we had never before seen a goblet?[14] The inventor “sees” something no one has ever seen before; it has no referent. What exactly does the inventor recognize? What forces the data in his mind into a “particular organization” that makes sense?

Usher proposed that the inventor “sees” a solution to the specific problem occupying his mind at the instant of insight. The problem serves as a focal point for organizing bits of information into a pattern that potentially resolves it. Drawing on a graphical device used by gestalt theorists to illustrate the “law of closure,” Usher compared the moment of insight to mentally arranging a set of broken arcs into a circle, thereby satisfying the desire for completion stimulated by the problem. The event is emotional, which accounts for the common denial by cranks that their finding doesn’t work. Looked at in this way, invention is necessarily contextual, because in order to be solved the problem has to be specific enough to support a solution. When Watt was struck by the lightening bolt on Glasgow green, he was not pondering the general problem of conservation of heat; he was deliberating the concrete problem of its conservation in a specific Newcomen engine.[15] That specificity puts dates on the causal history of invention. Watt could not have posed his specific problem the way he did before 1760 because an adequate quantitative concept of heat had not yet been achieved. The balance cranes invented by Brunelleschi to hoist materials for the dome of Florence cathedral and that so impressed the young Leonardo solved the specific problem of how to safely lift stone, brick and bronze objects to the unprecedented height of 300 feet without knocking down the walls of the building it rested on. The use of pullies and counterweights goes back to antiquity; but their combination was something new made possible by a more complete mathematical analysis of the lever.[16]

In the instant of insight the elements of a potential solution to a problem come into a new relation. Extrapolating from K?hler’s experiments on cognition in higher primates, Usher posited that the elements must be actively present in the inventor’s mind for insight to occur. In the experiments, K?hler placed fruit just beyond a caged ape’s reach, placing a baton near the animal with which it could capture the prized object. In repeated trials he found that the ape solved her problem only when fruit and baton simultaneously lay within her visual field; otherwise she remained baffled and frustrated.[17] The experiment suggests that achieving a satisfactory solution depends on serendipitous concatenation of its elements. That condition imparts significant unpredictability to the achievement of an invention, as nature rarely arranges the elements to in a form revealing a satisfactory pattern.[18] There was a large measure of luck in Edison’s nervous fiddling with compressed lampblack while reflecting on his frustrated efforts to find a satisfactory filament for a light bulb.

Except in the rare instances in which inventors have left an autobiographical account of their work, the historian can rarely observe the actual moment of insight. What can be obtained from the documentation are the problems that were posed and the presence or absence of elements needed for their solution. This is usually enough to construct an explanatory narrative. Usher noted that “even at a level of incomplete verification, the historian can proceed to develop the techniques of analysis that will reveal the grosser features of the processes by which man makes himself.” The invention of printing provides a good, though complex example. The elements needed to resolve the general problem of “mechanical writing” included a suitable support (paper), suitable ink (oil-based), a press (the woolen cloth calender) and moveable type. All of these elements were available by the early fifteenth century, and were being combined to make inexpensive wooden block prints by the 1430s and 1440s. The general impediment to the using of this technique to print books commercially was its inferior cost-effectiveness as compared with that of books currently being produced in specialized workshops by hand. The specific obstacle arose from the need to produce type in large numbers, which meant casting metal pieces in molds capable of holding matrices of variable size, and finding suitable materials for the matrix and metal punches. To judge from an incomplete documentation, the synthesis of the various elements that solved this problem was a drawn-out affair lasting from the early 1440s to the 1470s, of which the decisive invention was the adjustable type mold. The invention of printing was not the product of a single mind or even a single firm, but can be seen as a collective effort stretching over a whole generation. Its timing seems to be dictated not so much by an overwhelming demand for printed material, which until the price of books fell was satisfied by the output of workshops, but by the convergence of independent strands of technological know-how that suggested the possibility of substituting machinery for men in making letters.

The gestalt experiments indicate that the process of invention is strictly sequential, in that a problem must be adequately posed and the materials for its solution assembled before insight can occur. Usher identified a fourth stage in the process. Just as a new scientific finding has to be integrated into the existing stock of knowledge, so technological insight has to be translated into a working model and scaled up (or down) to the size needed to perform the desired task. Not every insight is workable. It took Watt nearly a decade to transform his insight into a commercially viable steam engine, and had it not been for the skills of Matthew Boulton’s machinists and Wilkinson’s boring machine, the effort probably would have failed. Usher termed that stage “critical revision.” Like the other stages, it consists of many acts of problem-solving.

Because of the necessary sequencing of its events, invention uses up calendar time. At each stage problems arise that require to be solved by insight, making the system inherently indeterminate. At best, the historian can evaluate rough probabilities from objective constraints imposed by the definition of the problem and the availability of appropriate materials for its solution at a given point of time. Usher stressed that because it is drawn out invention is by nature a social process; nothing logically requires successive stages to be achieved by a single individual or within a single epoch. The idea of applying the principle of the Archimedean screw to propulsion of vessels through water was first raised by a scientist in 1729, but it took four decades of intense and expensive effort finally to bring the screw propeller to fruition in the 1840s.[19] Usher regarded such delays as the consequence of temporally definable “resistances.” In general, the resistances are not social or economic, but reflect difficulties with respect to adequate formulation of the problem, the absence of one or more of the essential elements to its solution, failure to achieve the insight, and difficulties of its implementation. All of these elements are in some measure subject to verification, and thus narrated. Each makes invention time-consuming and time-dependent.

Usher’s approach also supplied the means to explain the history of the economy. As noted above, optimizing adjustments by agents to preferences and material constraints do not represent fundamental change, because change comes ultimately from the introduction of novelty into a social system. Usher situated that introduction in man’s capacity for problem-solving, thereby linking narrowly economic history to the broader evolutionary history of mankind. That history is not ruled by a timeless algorithm, but like the history of biological evolution rests on specific events that can in principle be identified.

[[T]he act of insight does not rise above the contingency of our knowledge upon specific contexts. Because these activities are conditioned, analysis is possible; but because they are conditioned they must be conceived as contingent upon the relevant contexts. Acts of insight seek particular modes of action or thought as a means of achieving specific ends. They do not seek absolutes or eternal verities.

Problem-solving covers most spheres of life. Usher was particularly interested in the technological sphere; but the general approach applies to the more complex area of social problem-solving, of which the construction of economic and social policy are the most important examples. That history, however, is intrinsically more complicated and harder to pin down than the history of invention. Like most pragmatists of his day, Usher believed that the problems posed in this sphere were largely created by the technological changes that he regarded as having an autonomous history. They were not less important, for all that, just more difficult

The Proof of the Pudding

A model is only as good as its implementation. Usher implemented his model of invention through a chronological account of mechanical invention in Europe from classical antiquity to the mid-twentieth century. The selection of the mechanical band of the technological spectrum was strategic, in that the decisive technological breakthroughs driving falling transport costs and productivity growth from the seventeenth through the mid-twentieth century were mainly due to mechanization of operations previously carried out by hand and the invention of new ways of generating power. It was strategic for another reason: machines combine different techniques for transmitting and controlling motion. A study focusing on the history of specific syntheses held out the possibility of identifying the circumstances that led to the combining of “the simple but relatively inefficient mechanisms of early periods into the complex and more effective mechanisms of today” (1929, p. 67). A final practical reason was the comparative abundance of documentation.

The substantive chapters begin with a discussion of the difference between scientific and technological knowledge. Until the seventeenth century, science was, as it remains, an interpretation of the physical world.[20] But outside celestial mechanics, where the Ptolemaic system was used to calculate celestial positions, that interpretation was either too broad to identify technological opportunities, or too flawed to be of practical use. Drawing on Pierre Duhem, Usher argued that the chief impediment to scientific treatment of mechanics arose from the belief that the principles of force and motion are self-evident. “Attention was thus drawn towards logical demonstrations and mathematical theorems that involved pure reasoning rather than towards experimental study of the phenomena.” Invention of devices for transmitting rotary motion and lifting heavy objects thus rested on knowledge of the strength of materials apprehended through practical experience, just as in ceramics and metallurgy. It was only from the middle of the fifteenth century that computational methods began to be applied to these problems, and it was only from the middle of the seventeenth that they acquired the power accurately to predict moments of force. From that point on, progress in mathematical analysis of mechanical problems was rapid. By the eighteenth century mathematicians and engineers were applying Newton’s third law of motion and Hooke’s law of elasticity to calculate the strength of materials, and using the embryonic science of fluid mechanics to compute the pressure of water on water wheel paddles and turbine blades.[21] Fulton’s work on the application of steam power to water craft is an outstanding example of this work.[22] The contribution to invention was situated mainly in the stage of critical revision.

The next chapter inventories the state of mechanical technology in classical antiquity. Although classical scholarship has revised Usher’s understanding of draft animal harness, the diffusion of water power, and the extent of geographical and occupational specialization, his assessment of the possibilities for invention remains sound.[23] At the end of the fourth century BC, classical civilization knew the five basic machines: lever, pulley, wedge, winch, and screw, and by the Christian era understood how gear trains translate and transmit rotary motion. As noted above, scientific analysis of these devices was not much help in designing new devices, which meant that the opportunities to combine the elemental machines into more complex devices depended on opportunities that manifested in the more immediate perceptual field. The classical presses are a good example: the beam press utilized pulleys to raise the weighted beam, while the screw press combined beam and screw. These simple combinations were closely tied to an immediate economic context setting the problem to be solved. Thus, displacement of hand mill by the rotary quern and the beam by the screw press to in the second century BC responded to the immediate problem of efficiently meeting the demand for large amounts of processed foods created by the growth of cities and trade. One can see the same dynamic at work in the invention of equipment for transporting and shaping exceedingly heavy ornamental stones.[24]

The transition to greater input of conceptual knowledge in the inventive process explains the tectonic shift in the complexity of mechanical inventions between 1500 and 1700. Early machines synthesized information obtained by visual and tactile perception (and in the case of foods, by taste and smell). Such perceptual insights are typically apprehended at low levels of generality and have been achieved many times in many places. Parallel development of lithic technology in the prehistoric world is explained by the repeated discovery that siliceous stones flake predictably enough to shape into useful forms.[25] The same was true of crafts based on manipulation of physical materials. Getting beyond that immediate level of insight, however, usually required the input of more generalized knowledge. As machines grow more complex, the physical and conceptual elements involved in achieving solutions to particular problems multiply, but as general concepts in mechanics are not immediately perceived by the senses, they are less likely to be conceived, and thus culturally idiosyncratic.[26] At this point it makes sense to compare concepts specific to civilizations as an explanation of the divergence in technological development. Usher regarded formulation of generalized scientific concepts as part of a “round-about” process of invention, in which the problems addressed are not immediately directed towards achieving a practical result. Huygens analysis of the pendulum as a means of timing the escapement mechanism in clocks is a good example.

Chapters 7 and 8 document the medieval history of two distinct branches of mechanical invention dominated by the perceptual element. The first harnessed the power of water and wind to mechanize the operations of grinding, crushing, stamping, sawing and fulling; the other captured the potential energy of gravitational force to drive and time clockwork. Both developments worked out mechanical principles mostly implicit in machines present in classical antiquity. The development of water mills and wind mills is the best documented, the critical element being the gear train translating vertical rotation of the wheel to the horizontal plane of the millstones. Gearing had been used in devices employed to measure distance and angles, but its extension to heavy-duty work was something new. One can imagine, but never demonstrate, that the idea of the water mill was taken from the gear train utilized in the cyclometer. Following an argument advanced by Lefebvre des No?ttes, and since shown to be erroneous, Usher supposed that the diffusion of water power was retarded by the deadening effect of slavery on incentives to save labor. Archaeological evidence has since demonstrated widespread diffusion of water-powered grain mills by the second century AD, which speaks volumes to the value accorded to economizing labor in the most burdensome tasks.[27] It also speaks to the wide distribution of requisite carpentering skills. The smaller horizontal and generally larger vertical mills diffused simultaneously, their geographical distribution depending on the nature of the stream and the economic advantage of high volume milling. The increased incidence of vertical wheels after 1000 AD is best explained not by technological innovation, but by opportunities for scaling up milling operations created by the growing commercialization of corn farming.[28]

Growing commercialization in the twelfth and thirteenth centuries provided incentives to apply water power to other industrial activities. The most important uses required translating the rotary motion of the water wheel into reciprocal motion used to drive bellows, stamping devices, and saws. Although Usher considered the crank and cam to be medieval inventions, Ausonius’s fourth-century description of a water-powered device for sawing marble blocks in the Rhineland indicates its presence in Antiquity. As in other areas, the surviving documentation suffers from severe selection bias against evidence for its early use. Gear trains were adapted to other power sources where running water was unavailable or inconvenient. Of these the most complicated mechanism was the gearing for the windmill, which pivoted with the sail as it turned towards the wind. The most revealing aspect of the windmill, however, illustrates how purely perceptual knowledge produced inventions that achieved high levels of technical efficiency. When Euler, MacLaurin and Coriolis undertook mathematical and experimental studies of the optimal angle and shape of windmill sails in the eighteenth and early nineteenth century, they found that Dutch craftsmen had solved the problem as a practical matter by the seventeenth century.[29] As in the case of the watermill, the path of invention seems to have mainly reflected the accretion of experience under conditions of expanding demand for the apparatus.

Clockwork presents a different chronology. Timing devices controlled by the flow of water through a self-regulating float valve were more accurate than clocks whose timing was controlled by an escapement mechanism and remained in use down to the eighteenth century because they were cheaper to build and repair than the by then more accurate mechanical clocks.[30] While the invention of the escapement mechanism is obscure, its presence in clockwork is securely dated to the third quarter of the thirteenth century. Subsequent development of what was originally a massive mechanism exploited momentum of weighted bars or wheels to time the escapement and damp the recoil. Usher’s discussion of these points is highly technical and directed at questions of dating. In the broader history of mechanical invention the importance of clockwork resulted from its complexity, and demands for greater accuracy giving rise to a sequence of problems that were gradually resolved by scientists and craftsmen of the highest order. An important by-product of the construction of the early tower clocks was the transfer of knowledge of how to cut and design gears from the millwrights to blacksmiths. In the seventeenth and eighteenth centuries the demand for greater accuracy created opportunities to develop gear-cutting machines that gave solutions on a small scale and for work in softer metals to problems that were to emerge on a larger scale and in iron and steel.

The next chapter considers the place of Leonardo da Vinci in the development of mechanical invention. Leonardo’s role is both symbolic and real. As a symbol he marks the shift towards scientific analysis of mechanical problems (as an adult he taught himself geometry), and the use of scale models to test the apparatus (a procedure pioneered by Massacio to study pictorial composition). Of the 18,000 sheets he bequeathed to his pupil Francesco Melzi, only 6,000 have survived, and as they are not dated, it is impossible to determine the representativeness of the sample and the sequence of his thought. He invented a centrifugal pump, anti-friction roller bearings, a screw-cutting machine, and a punch to make sequins for ladies’ dresses. He conceived a machine to make needles, and in 1514 was given a room in the Vatican to construct a machine for grinding parabolic mirrors to capture solar energy for boiling dyestuffs. He expected to get rich from his inventions, and was alert to potential opportunities to substitute machines for labor. He was not confident in his Latin, and of Greek he had none. He sensed that mechanisms were subject to common principles, but did not have the training to bring the abstract concepts of force and movement into focus. His workshop method of jotting down rough notes and cases was not suited to sustained trains of abstract thought. But his capacity to imagine three-dimensional mechanical connections, which his artistic training permitted him visually to describe, was unequalled. His papers circulated widely after his death, and provided ideas and inspiration to inventors for nearly a century. Usher viewed Leonardo as embodying the shift from perceptual to conceptual invention in the practical sphere of mechanics.

Save for relatively isolated cases, mechanical innovation was empirical, realistic, and practical. Achievements of great consequence had been realized, but by a process in which the immediate end was ever in the foreground. It is only with Leonardo that the process of invention is lifted decisively into the field of the imagination; it becomes a pursuit of the remote ends that are suggested by the discoveries of physical science and the consciously felt principles of mechanics (1954, p. 237).

The remainder of the book, with the exception of the chapter on printing discussed above, traces out that subsequent history through a chronology of the development of textile machinery, clocks and watches, steam power, machine tools, and the development and exploitation of the turbine. As these developments are well-known there is no need here to review them here. In his account of particular inventions, often in eye-glazing and occasionally impenetrable detail, Usher was primarily concerned with showing the cumulative nature of mechanical achievement, much of it by unknown or relatively little known inventors. The development of textile machinery provided a well-documented case in point. While the increasing complexity of the material makes it difficult to reduce to an intelligible story following the lines set out in his model of invention, his broad conclusion was that the acceleration of invention in textile machinery was conditioned more by the nature of the mechanical difficulties to be overcome than economic factors. By the early eighteenth century the technical capacity and craft skills needed to overcome those difficulties were well in hand, as any visit to a well-appointed museum of technology will demonstrate. From that point on, progress depended on the way specific problems came to be posed, or not posed, and how the stage was set for insight. By the mid-eighteenth century, the increasing indirectness of invention and its rising cost made securing and protecting intellectual property rights increasingly important.

These factors are all evident in the development of the steam engine. Caus’s discovery that steam is evaporated water made it possible to conceive the possibility of extracting power from atmospheric pressure by condensing steam in a closed vessel. Exploitation of that insight raised a series of technical problems associated with positioning and controlling the valves regulating the flow of steam and water. Watt’s invention of the separate condenser was critical revision of Newcomen’s atmospheric engine. Translating that insight into a commercially viable machine raised new problems the solution of which largely depended on the skill and experience of Boulton’s craftsmen. The role of conditioning factors is illustrated by the serendipitous appearance of Wilkinson’s boring machine, which machined a cylinder four feet in diameter to tolerances no thicker than a dime. The development and diffusion of the steam engine in turn led to greater use of metal gears connecting increasingly powerful engines to increasingly heavy machinery, and as the speed and force of the engines increased, the resulting stress and friction induced intensive theoretical and practical study of the optimal shape and position of toothed wheels and pinions. The sequence thus illustrates Usher’s general model of mechanical invention as a sequence of problems raised and solved. We see in these developments a comprehensible narrative of how one thing led to another in the most critical region of the new technology.

The history of tools for shaping metal to high tolerances has a parallel history. The basic elements of the mandrel lathe, slide rest and lead screw were present by the end of the sixteenth century. In the eighteenth century the wooden parts were replaced by metal, increasing their accuracy and making it possible to machine heavier pieces of metal. Senot’s screw-cutting lathe (1795) displayed at the Mus?e du Conservatoire des Arts et M?tiers is an outstanding example of this development, and attests its international scope.[30] Usher argued that after the substitution of iron for wooden headstocks, the principal obstacle to the development of heavy-duty machine tools was the difficulty of obtaining accurate lead screws. Here the problem was well-specified, but achieving a solution required years of painstaking work. Maudslay invented a device to correct errors of one-sixteenth of an inch in a seven-foot screw, tested the result with a micrometer, and made further corrections until he achieved the desired result. Such accuracy was essential to achieve mass-produced metal parts at low cost, though as Usher noted, the applications were initially confined to narrow fields, most notably in the manufacture of wooden pulley blocks, and firearms. Of more initial importance was use of heavy machine tools to shape large pieces of metal to the fine tolerances demanded by working parts of steam engines and locomotives. By the middle of the nineteenth century that capacity was available to be applied to a widening range of mass-consumed products like agricultural equipment, sewing machines, typewriters and bicycles. By that date the process by which specific mechanical problems were posed, the stage set and critical revision of the resulting insight carried out had become largely autonomous. It is difficult to imagine what plausible reconfiguration of relative factor endowments could have significantly affected the ensuing wave of labor-saving innovation.

The final chapter sketches out the history of the turbine, of which the applications range from more efficient exploitation of the power in falling water to the exploitation of the energy in expanding steam and gasses. Although it runs parallel to the development of the reciprocal steam engine, the story of the contemporary development of the turbine is a “particular system of events” that is entirely distinct from it. As with machine tools, investigation of impulse motors can be traced back to the early sixteenth century. The technical problems to be resolved, however, were of the highest order of difficulty, involving the invention of materials capable of withstanding extremely high temperature and rotational friction, finding optimal shapes and positions of the tubes and vans for the different media that propelled them. All this took time. Mathematical studies of turbulence relevant to the performance of turbines date to the eighteenth century; the basic breakthroughs in design by Fourneyron and Burdin date to the 1820s and 1830s. By the 1840s the accuracy of machine tools was high enough to produce a tight fit between the rotor and its casing. Parts rotating at ten to thirty thousand rpm required grades of steel that became available only towards the end of the nineteenth century; in the case of gas turbines, the materials became available only in the 1930s. The history of turbines, then, encapsulates the general trend in mechanical invention from problem-solving directed at an immediate solution with means assembled in the perceptual field to problem-solving based on scientific analysis and assembly of materials from a wide range of sources. The point is that all of this took time, and although the rough outlines of a solution might be fleetingly glimpsed, the timing of its achievement could not be predicted. The first patent for a gas turbine was taken out in 1791; a practical solution to the problem of exploiting the expansive power of heated gas in jet engines was achieved only in the 1930s.

The development of the turbine leads the discussion to the generation and transmission of electric power. The potential of large heads of water and great heads of steam could not be exploited as long as it had to be employed in situ, because no establishment could take more than a small proportion of the total power available. The invention of the dynamo and means of long-distance transmission relieved that constraint. The early development of that technology was achieved between 1830 and 1880, by which time the crucial problems had been resolved. That history, too, represents a particular system of events. The history of internal combustion engines illustrates the same pattern. An early recognition of the possibility of using the explosive power of gas in a piston (Huygens, 1680, Papin, 1690), followed a century later by patented engines (Street, 1794; Lebon (1799), lack of success for an extended period of time due to the inaccuracy of machining, difficulties of controlling the timing of the ignition and opening and closing of the valves, followed by a successful inefficient engine leading to closer analysis of the sources of that inefficiency. The sequence plays itself out as a narrative. Usher observed that from a broad perspective the history of the individual sources of power revealed a tendency to develop all possible forms of application of a general principle. The result was that by 1950 the world possessed a set of power-generating devices that spanned the gamut of weight and power capacity.

The History ends with that observation. Over the course of more than 300 pages of substantive discussion, it gives an overview of the development of what was the central strand of technological development through the early twentieth century. It explains within the limitations of the documentation and the level of detail appropriate to a general overview how novelty emerged in the sphere of mechanization and the generation of power. Usher offered no conclusion to this work. Indeed, in the introduction to the second edition he noted that he deliberately avoided forcing the narrative into a preconceived mold. The History was not a test of the theory of emergent novelty, only an illustration. In his later work Usher returned to the question of how to combine the insights of economics with an empirical treatment of time. He argued that “any consistently empirical interpretation of history must find some adequate explanation of the processes of change.”[32] The great enemy to a rational understanding of the past in his time, as in ours, was radical idealism, which seeks to explain events by their presumed final ends or purpose.

Usher’s work raises a number of problems that have been imperfectly addressed. His insights on the nature of mechanical invention are generally accepted and have been extended by historians of technology and economic historians, but the model has not been generally applied to other spheres.[33] A significant obstacle to its implementation is the extremely high degree of technical detail required to give an adequate account of any particular technological development. While detail at that level is common in the fields of political and institutional history, the desire to read such accounts is an acquired taste, though perhaps no more so than in the arcane corners of art history. As a consequence, the deployment of Usher’s method by economic historians has tended to be illustrative rather than narrative and probative. The rhetorical difficulties turn on the audience to be addressed, and the level of generality required by the narrative. On the broader question of the role of time in economic processes, the picture is equally discouraging. The debate over the nature and significance of path-dependence touched analytical issues raised by Usher, but it was deflected by questions relating to dynamic optimality, which as Usher had anticipated, originate in a transcendentalist obsession with final ends. As a result, the question of what happened and how it happened got pushed aside by the question why it happened. “Why” questions are intrinsically non-empirical.

Usher’s focus on explaining the emergence of novelty as the special province of economic historians is nevertheless worth preserving. Bill Parker organized his lectures on economic history around the framework of challenge and response, which is just a broader way of identifying the history as a history of problems posed and resolved (or not). The problems are not just technological. The analysis of organizational and political responses to economic change can be carried out on lines similar to those that Usher considered workable for the study of scientific and mechanical invention. Some responses are comparatively easy to model using standard tools derived from the calculus of optimization; others require more contextual detail. A workable history, however, requires limiting the field to a “particular system of events” that permits a narrative account. An outstanding example of this type of economic history is Wright’s account of American slavery.[34] Since the early 1960s the main thrust of economic history was directed away from Usher’s concept of explanation by narration. The power of Kuznets’ categories to organize numerical data provided nearly two generations of economic historians with productive work filling in the gaps and running down the tangled chains of quantifiable explanation. But Kuznets took the technological revolution as a given; the modern economic epoch was its consequence. Yet in the end, to quote one of the less illustrious figures in American history, “stuff happens.” Part of the task of economic history is to find out exactly what that stuff was, and how it happened. Usher’s work is a model of that type of economic history, and also shows how difficult it is to successfully pull off.


I was distractedly browsing through my alumni bulletin this evening — checking the latest mortalities and other alumni affairs — when I came across the following passage in an article on Leland C. Clark (Antioch College 1941), who received the Frit J. and Dolores H. Russ Prize (the nation’s stop award for scientific engineering) in 2005 shortly before his death.

Here’s the story of his oxygen electrode invention.

Late one night, Clark — then in his thirties — was opening a pack of cigarettes while relaxing with colleagues after assisting in a by-pass surgery using his prototype heart-lung machine. Although the surgery had been successful, Clark knew that such procedures require precise monitoring of oxygen levels in the blood. But the platinum electrode he had originally designed wasn’t working well; red blood cells were blocking the oxygen molecules near the electrode.

What happened next was one of many shining moments in Clark’s career. “He was fiddling with his cigarette pack and suddenly got the idea that oxygen might permeate cellophane.” Soon thereafter, Clark tried moving the two electrodes close together, protected inside a glass tube by a cellophane membrane. The innovation allowed oxygen to enter and be measured with no interference from the red blood cells. To test the new oxygen sensor he needed to find a way to pull the oxygen out of a control solution to calibrate the sensor settings. He added glucose and the enzyme glucose oxydase, as a catalyst, and the oxygen was quickly removed.

Before long, however, he realized that by equipping his oxygen sensor with a thin film of the enzyme, he could read the decrease in the oxygen recorded in the presence of glucose. Suddenly Clark had a simple device for measuring glucose, also inventing the first biosensor for that purpose. Today, electrochemical biosensors have been designed to measure lactate, cholesterol, lactose, sucrose, ethanol and many other compounds.[35]

One sees here all of Usher’s stages in exceptional relief: the posing of the problem, the setting of the stage, the insight and critical revision, followed by extension into new problems and new solutions.


1. Perhaps no better example of that vision can be found than in following passage composed by the aged Friedrich Meinecke in its wreckage. “Behind the growing pressure of increased masses of population … stands the struggle for the way of life of the individual nations. By way of life we mean here the totality of the mental and material habits of life, the institutions, customs and way of thinking. All of these seem to be bound together by an inner tie, by some guiding principle from within, to form a large, not always clearly definable but intuitively understandable, unity.” The German Catastrophe: Reflections and Recollections. Boston (1950), p. 87.

2. Erik Grimmer-Solem, The Rise of Historical Economics and Social Reform in Germany, 1864-1894, Oxford (2001).

3. T. H. Marshall, English Historical Review 42 (1927), 624.

4. “The Application of the Quantitative Method to Economic History,” Journal of Political Economy 40 (1932), 186-209.

5. Cited by Veyne, Writing History: Essay on Epistemology, Middletown, CT (1984), 119.

6. Veyne, Writing History, 63

7. Veyne, Writing History, 26-27. The literary reference is to Balzac’s Grandeur et d?cadence de C?sar Birotteau.

8. Mokyr appears to adopt this perspective in his evolutionary interpretation of technological change. “Like mutations, new ideas, it is argued, occur blindly. Some cultural, scientific, or technological ideas catch on because in some way they suit the needs of society, in much the same way as some mutations are retained by natural selection for perpetuation. In its simplest form, the selection process works because the best adapted phenotypes are also the ones that multiply the fastest.” The Lever of Riches, New York (1990), 276. The proposition is defensible with respect to economic factors conditioning the diffusion of inventions. It does not explain, as Usher surely would have observed, how inventions happen. Mokyr’s concept of a unit technique or idea subject to selection bears an obvious resemblance to Leibniz’s monad, and the sufficient reason that generates in the fullness of time the “best of all possible worlds.”

9. Explaining technological change by Malthusian population pressure is an example of this kind of approach. For a recent example, see Oded Galor and David Weil, “Population, Technology and Growth,” American Economic Review 90 (2000), 806-28.

10. Ludwig Wittgenstein, Philosophical Investigations, Oxford (1972), 196.

11. See Norbert Russell Hanson, Perception and Discovery, Cambridge (1958), and more generally Wittgenstein, Philosophical Investigations.

12. Hanson, Patterns of Discovery, 7. A celebrated instance of dual perception was the inability of researchers to identify the cause of the potato blight, in which the fungus Phytophthorus infenstans was alternatively believed to be a cause and consequence of the disease.

13. The Autobiography of Charles Darwinz edited by Francis Darwin, New York: Dover Publications (1958)z 26.

14. Locke reports a conjecture made to him by a French correspondent who suggested a man cured of blindness might not be able to distinguish between a box and a sphere. That conjecture has been experimentally confirmed.

15. “I was thinking of the engine at the time, … when the idea came into my mind that as steam was an elastic body it would rush into a vacuum and might there be condensed without cooling the cylinder” (cited in Usher (1954; 71)).

16. Salvatore di Pasquale, “Leonardo, Brunelleschi, and the Machinery of the Construction Site,” in Montreal Museum of Fine Arts, Leonardo da Vinci: Engineer and Architect, Montreal (1987), 163-81.

17. In another set of experiments with chickens food was placed outside a rectangular enclosure having an opening on one side. The hens “solved” the problem of obtaining the food only when the food and the doorway were in their line of sight.

18. The one major exception may be the “invention” of agriculture in the Near East, which most likely occurred through an improbable sequence of climatic changes that induced incipient domestication in a handful of small grains and pulses harvested in naturally occurring stands. The term invention is inappropriate in this context. See David Rindos, The Origins of Agriculture: An Evolutionary Perspective, Orlando FL: Academic Press (1984), and Donald O. Henry, From Foraging to Agriculture: The Levant at the End of the Ice Age, Philadelphia: University of Pennsylvania Press (1985).

19. The chief obstacles were intellectual, one being disbelief that a device as small as a propeller could drive a large ship, and the other concerning the optimal shape of the device in the context of extremely complex issues with respect to fluid mechanics. The history is reviewed by Maurice Daumas, ed., A History of Technology and Invention, Vol. 2, New York (1972).

20. Prior to the seventeenth century it also interpreted the non-physical world, as the medieval enquiry into the physics of the Eucharist amply demonstrates. On this and other topics relevant to the present discussion, see Edith D. Sylla, The Oxford Calculators and the Mathematics of Motion, 1320-1350, New York (1991).

21. Maurice Daumas, ed., A History of Technology and Invention, Volume III, New York (1979), 25-27, 81-89.

22. Fulton made countless experiments calculating the resistance to paddlewheels of varying design and to the form of the hull in relation to the weight and velocity of the engine. His work was based on Colonel Mark Beaufoy’s experiments testing Euler’s theorems on the resistance of fluids. This was critical revision. H. W. Dickson, Robert Fulton: Engineer and Artist, London (1913).

23. See my manuscript, “Prehistoric Origins of European Economic Integration.”

24. J.B. Ward-Perkins, “Quarries and Stone-working in the early Middle Ages: The Heritage of the Ancient World,” Artigiano e tecnica nella societ? dell’alto medioevo, Spoleto (1971), 525-44; Valery A. Maxfeld, Stone Quarrying in the Eastern Desert with Particular Reference to Mons Claudianus and Mons Porphyrites, in David Mattingly and John Salmon, eds., Economies Beyond Agriculture in the Classical World, London (1991), 143-70.

25. Brian Cotterell and Johan Kamminga, Mechanics of Pre-industrial Technology, Cambridge (1991), 127-30.

26. Within restricted ranges of perception many mechanical concepts are indistinguishable. Where friction is present, the Aristotelian theory that constant force is needed to keep an object in uniform motion is observationally equivalent to Newton’s principle of inertia.

27. The archaeological evidence, which was not available to Usher, is abundant. For a compilation of European finds, see Orjan Wikander, “Archaeological Evidence for Early Watermills: An Interim Report,” History of Technology (1985), 151-79, and Richard Holt, The Mills of Medieval England, Oxford (1988). North African evidence is surveyed by David Mattingly and R. Bruce Hitchner, “Roman Africa: An Archaeological Review,” Journal of Roman Studies (1985), 165-213.

28. A similar transformation around the same time can be seen in the substitution in northern France of naked wheat (triticum aestivum) for spelt (triticum spelta), which being a bearded cereal costly to transport and difficult to mill was less suited to commerce. The displacement and the appearance of the vertical mill went hand in hand. See Jean-Pierre Devroey, “Entre Loire et Rhin: Les fluctuations du terroir de l’agriculture au moyen ?ge,” in J.-P. Devroey and J.-J. van Mol, L’?peautre (triticum spelta): Histoire et ethnologie, Bruxelles (1989), 89-105.

29. Daumas, History of Technology and Invention, Volume III, 20-22.

30. Galileo used water clocks in his experiments on falling objects.

31. Maudslay’s all-metal bar lathe is dated to the same year.

32. Usher, “The Significance of Modern Empiricism for History and Economics,” Journal of Economic History (1949), 149.

33. I made a preliminary stab in “The Shifting Locus of Agricultural Innovation in Nineteenth-century Europe: The Case of the Agricultural Experiment Stations,” in Gary Saxonhouse and Gavin Wright, eds., Technique, Spirit and Form in the Making of the Modern Economies: Essays in Honor of William N. Parker, Research in Economic History, Supplement 3, Greenwich, CT (1984), 91 214.

34. Gavin Wright, Slavery and American Economic Development, Baton Rouge (2006).

35. “Leland C. Clark Leaves a Medical Legacy,” Antiochian (Autumn 2006), 31.

George Grantham teaches economics and economic history at McGill University. He is the author of several works on the productivity of French agriculture in the nineteenth century, the macroeconomics of pre-modern agricultural societies, and the economic history of prehistoric Europe. He is presently applying Usher’s concept of a “particular system of events” to reconstruct the pre-modern history of European agricultural productivity.