MACHINE INVENTION DURING THE ENGLISH INDUSTRIAL REVOLUTION: EVIDENCE FROM THE PATENT ROLLS, 1711-1850
Richard J. Sullivan
College of the Holy Cross
Economic historians give much attention to the mechanization of industry, and regard it as one of the fundamental causes of the industrial revolution. Until recently, it was thought that an important aspect of mechanization was that more machines were being used in the production process. But if this were so, then industry must have invested so heavily in mechanical equipment that the capital-labor ratio would have risen, and studies of capital formation have shown that this did not occur significantly until after 1830 (Feinstein, 1981, p. 138-9). Thus mechanization is said to have influenced production processes through another route: from "about 1760 to 1830, increases in output per capita rested on providing the worker with 'better equipment' rather than with 'more equipment'" (von Tunzelmann, 1981, p. 155).
If industries turned to better machines in their production processes, then a subset, and possibly a significant subset, of those machines would have been recently invented. A hypothesis is implicit in the view that better machines were a principal aspect of the industrial revolution: inventors were busy inventing and improving machines for use in production processes. This paper examines the "better machine" hypothesis by investigating machine invention during the English industrial revolution. I tabulate the number of machine patents issued from 1711 to 1850, and use the data to measure the intensity of machine invention in the overall invention of an industry, where the intensity of machine invention is defined as machine patents divided by total patents relevant to the industry.
I find that there were many more patented invention for machines to be used in manufacturing processes. In the context of overall patented invention, however, machine invention was not extraordinary; non-machine invention was growing as fast as machine invention. The evidence suggests that more emphasis be placed on non-mechanical invention as a source of improvement to per capita output. I also find a mixed pattern to machine intensity of invention across industry, and argue that this is inconsistent with a macro-economic cause of increased inventive resources devoted to machine invention.
The theoretical foundation of my interpretation of measured machine intensity builds on Dutton's (1984) research, in that I assume that inventive resources were allocated in an invention market. Moreover, if patent data best reflect what Mokyr (1990. p. 13) has termed "micro-invention" (numerous inventions that have small individual but large cumulative impact), then patterns found in patent data will reveal responses by inventors to incentives. Inventors perceived relative returns to pursuing various inventive projects, and chose to extend their efforts towards those that promised the greatest reward. The mix of invention between mechanical and non-mechanical aspects of production will bear some relation to the potential rewards from invention, and as inventive projects are pursued, a rough equilibrium may be obtained so that the marginal returns from invention devoted to mechanical and non-mechanical aspects of production are approximately equal. The mix between mechanical and non-mechanical invention would not be fixed, but would change depending on factors such as the sophistication of knowledge relevant to mechanical and non-mechanical technology, as well as on factors influencing the demand for invention.
If it is true that better machines were fundamental to the changes seen during the industrial revolution, then the machine-intensity of invention must have increased across a significant portion of industry. By referring to a "significant portion" of industry, this hypothesis is meant to agree with the many writers who suggest that a macro-economic phenomenon occurred.
While patent data have come to be accepted as providing valuable information, they are flawed measures of invention and need to be interpreted with care. Though the level of patenting through time is of interest, it is the measured machine intensity, defined as the ratio of patented machine inventions to total patents relevant to an industry, that is the focus of this study. Ideally, this ratio would measure the actual machine intensity, defined as total number of patentable machine inventions to total patentable inventions in an industry. Most who study patent systems argue that mechanical invention is likely to be more patentable than other types of invention. If the propensity to patent machines is relatively high, patented machine inventions will be overrepresented among all patents, and the measured will overstate the actual machine-intensity of an industry.
Additionally, it can be shown that for various values of the propensity to patent, measured and actual machine-intensity can move in opposite directions. For a time-series of machine intensity within a single industry, where the ratio of the propensity to patent machines to its overall propensity to patent can be assumed to be relatively stable, it would take a very unusual set of parameter values for measured and actual machine intensity to move in opposite directions, so that we can be confident that changes in measured machine intensity reveals changes in actual machine intensity. For a cross-section of industries within a single time-period, where values of the propensity to patent could show considerable variation, small differences (of two or three percent) in measured machine intensity could be meaningless regarding actual machine intensity.
The source of information is a subject matter index of patents of invention (Woodcroft 1857). Two methods are used to identify machine patents. First, I include all patents listed in subject categories that are machines (saws, lathes, etc.). Second, for subject categories with both machine and non-machine inventions, I inspect a brief description of the patented invention, and if the description explicitly refers to the invention as being a machine, machine part, or as using mechanical principles, the patent is counted as a machine invention. After all patented machine inventions were identified, they were grouped into four categories: (1) production machines for manufacturing; (2) mechanical products (locks, pianos, clocks, etc.); (3) production machines for non-manufacturing industries; and (4) motive power. The result reveals the importance of mechanical principles in patented invention: for the entire period from 1711 to 1850, patented invention that involved mechanical principles accounts for 44.2 percent of total patented invention. Patents series relevant to specific industries have also been tabulated. Machine patents relevant to the various industries were found by cross-tabulating the machine categories with industrial categories.
Patented Machine Invention.
Aggregate statistics do little to support the idea that advances in mechanical invention were more important than non-mechanical invention, except for advances in steam technology. The 1761-90 period, which is arguably a crucial turning point in English inventiveness (Sullivan, 1989, and 1990, pp. 355-6), was also the period with the smallest share of machine and motive power patents out of total patents, with a share of 40.8 percent. While it is true that the number of machine and motive power patents increased steadily from 1711 to 1850, the share of total patents was relatively stable, suggesting that mechanical invention increased due to factors that increased inventiveness generally. Indeed, the share of patented machine inventions (excluding motive power) out of total inventions declines steadily, from 39.6 percent in the 1711-60 period to 33.3 percent in the 1821-50 period. While the large share shows how mechanical technology deserves special attention, the declining share of machine patented invention reminds us that we should not neglect non-mechanical technology, such as invention for chemicals, materials, preparatory processes, methods of manufacture, and non-mechanical apparatus and tools.
Table 1 presents the number of patented inventions, and measured machine intensity of invention, in 14 industries. For the entire 1711-1850 period, the Cloth Textile industry had both the largest number of patented invention, and invention that was the most machine intensive. But note that the industry with the second largest number, Chemical and Allied Industries, also had invention that was the least machine-intensive. Calculations show that there is not a significant correlation between the number of patents in each industry and the machine-intensity of invention. The most inventive industries were not necessarily mechanical, and machine-oriented invention did not necessarily imply large amounts of invention.
Machine-intensity of invention in the Cloth Textile industry, at 62.3 percent for the entire 1711-1850 period, contrasts sharply with that found for other industries, and attests to its unique technology. The contrast is more pronounced when comparing Cloth Textiles to other manufacturing industries, for the industries with the next largest values (for the entire period) of measured machine-intensity were Mining and Quarrying, and Agriculture, and were not manufacturing industries. The manufacturing industry whose patented invention was the next most machine-intensive was Paper, Printing, and Publishing, which reflects the printing press, and in the latter part of the 1711-1850 period, the paper-making machine, in its production process.
Similar analysis of the Construction, Other Manufacturing, and Instrument Engineering industries reveals machine-intensity similar in range and magnitude to the industries shown in Table 1. But these three industries have a common bond, in that they each include a mechanical product (locks, pianos, and clocks) among the patents counted in their total, and in each industry, the number of patented inventions for mechanical products exceeded the number for production machines. Mechanical invention was often oriented more towards improvement of products rather than production processes in industries where mechanical products were an important part of their output.
In both the Vehicles and in the Shipbuilding industries, tabulation shows a low percent of production machines out of total patented invention, suggesting dominance of non-mechanical processes, such as handicraft production, of vehicles and ships for the period from 1711 to 1850. But counting patented invention for production machines understates substantially the extent of machine invention in the Vehicles and Shipbuilding industries. Most important, application of inanimate motive power to vehicles and ships was the primary way that inventors in those industries applied mechanical principles. For example, in Shipbuilding, 47.4 percent of patented invention over the 1711-1850 period was aimed at improving ships' sources of power.
Although mechanical products and motive power were important applications of mechanical principles, it is production machinery that draws historical attention, for the machine, it is said, formed a nucleus around which production processes were revolutionized, particularly in manufacturing. Column (3) of Table 2 presents a tabulation of aggregated production machines for manufacturing industries. In the 1711-60 period, 55 production machine inventions were patented for manufacturing, with the number rising thereafter, until in the 1821-50 period, 1679 production machine inventions were patented. Inventive resources were increasingly devoted to improving machines used in manufacturing production processes. When set against all patented invention, the increase in invention of manufacturing production machines, however, is far more modest. As Table 2 shows, production machines for manufacturing accounts for 14.9 percent of total patents issued in the 1711-60 period, and the percentages are 15.7, 15.5, and 18.9 for the 1761-90, 1791-1820, and 1821-1850 periods. Only in the final period do we see a shift of any appreciable magnitude in the percentage, and even then, some 81.1 percent of patented invention was for inventions other than manufacturing production machines. Certainly, the mechanical aspects of manufacturing advanced considerably. But if advances in machine technology were a primary cause of technological advance generally, then it is surely going to show up in an increasing share of patented machine invention in overall total patented invention. Since the share increased only slightly, it is more likely that whatever caused the widespread increase in English invention was also responsible for the increase in aggregate machine invention.
Two additional observations show that mechanical invention had limited impact on manufacturing technology. First, of the 1976 patented inventions relevant to the Cloth Textile industry for the 1711-1850 period, 1232 were for production machines. Thus, production machines for Cloth Textiles alone accounts for 53 percent of the 2319 manufacturing production machines for the same period. To see more clearly the importance of Cloth Textiles, Column (1) in Table 2 presents the tabulation of its production machines, and Column (2) presents a tabulation of the production machines useful to the remaining manufacturing industries. While there was some ebb and sway between patented production machines in the Cloth Textiles versus the remaining manufacturing industries, invention for production machines for the Cloth Textiles industry clearly dominates invention for manufacturing production machines over the entire 1711 to 1850 period.
Second, excepting for Cloth Textiles, measured machine intensity of invention in most manufacturing industries was modest, ranging from 4.3 to 34.0 percent for the 1711-1850 period. Moreover, there was a wide variety of experience in changes to the machine-intensity of invention. In some manufacturing industries, invention became increasingly machine intensive; over the entire 1711-1850 period, industries that had a steady increase in measured machine intensity were the Chemical and Allied Industries, Clothes and Shoes, and Timber and Furniture, while over the 1761-1850 period, the Other Metal Goods, Other Manufacturing, and Instrument Engineering had increases in measured machine intensity. In some manufacturing industries, invention became decreasingly machine intensive; over the entire 1711-1850 period, the Leather industry had a steady decrease in measured machine intensity, while over the 1761-1850 period, measured machine intensity decreased in the Food Processing and the Coal and Petroleum industries. And, of course, some industries had mixed experiences (Metal Production and Fabrication; Paper, Printing, and Publishing; and Brick, Pottery, Glass, and Cement). The mixed experience in manufacturing left changes in the machine intensity of the manufacturing industries overall relatively modest.
Among the many changes associated with the industrial revolution, economic historians have placed much emphasis on the hypothesis that better machines revolutionized industry. The thesis is stated most clearly by Paulinyi (1986, p. 283) who wrote
The main element of the technical revolution, which we are accustomed to calling the Industrial Revolution, was the massive introduction of material-forming working machines, and it was able to change the existing technological system only because, thanks to numerous technical innovations, the barriers to the production of machines by machines had been overcome in the first thirty years of the nineteenth century.
The patent record suggests that the "better machine" hypothesis be relieved of the heavy burden placed upon it: in the context of overall invention, machine invention did not occur on a massive scale, and if it did, it was not before 1850, if at all. There was a large and growing number of patented machine and motive power inventions, but growth greatly favored motive power, and more specifically for steam engines and its application. The contribution of mechanical principles to manufacturing technology was important but not overwhelming. Even in those industries that experienced a rise in machine-intensity, the rise was gradual. A larger and larger number of patented production machines enlarged the pool of techniques from which producers could choose. But equally important were a larger and larger number of patented non-mechanical techniques from which to choose. Historians of this period ignore non-mechanical invention at the risk of a incomplete picture of changes during the industrial revolution.
The mixed experience of machine-intensive invention across industry suggests strong independent influences on the machine-intensity of invention within each industry. Macro-economic influences, such as significant advances in machine science, or widespread increases in demand, may have played a role, but left little discernible impact on the pattern of changes to machine-intensity across industry. The steam engine provides a good example of a narrow cause of an increase in machine-intensity, because until well into the nineteenth century, it had a direct influence on a limited number of industries, notably the railroad and shipping industries (von Tunzelmann, 1981, p. 157). In other industries, no single cause can explain the observed changes in machine-intensity of invention. Advantages of machines relevant to each industry, the development of non-mechanical alternatives to machines, changes in patterns of demand, and so on, determined the relative rewards to inventors in seeking improvements to the technologies of each industry. A mixed pattern of changes in potential rewards for machine invention contributed to the mixed pattern of changes in machine-intensity of invention.
The image left by examining patented machine invention is one of continuity within the context of revolution. The revolutionary aspects of English invention for the period from 1711 to 1850 was a widespread increase in resources devoted to invention beginning around 1760 (Sullivan, 1989), and a significant new technology, the steam engine. The continuity of this period lay in the types of technical problems to which inventors applied their efforts. The non-mechanical aspects of production processes received their due attention over the entire period. Inventors depended on the path of prior invention (David 1985), and worked within the context of existing techniques. As von Tunzelmann (1981, p. 160) writes, "pockets of extraordinarily labour-intensive operation continued in almost every industry throughout the nineteenth century." In some industries, this implied that hand techniques were improved. In those industries where machines would eventually prove to dominate the production process, the technological advance that led to the change was gradual, as improvements in non-mechanical techniques gave way to the mechanical inventions.