Derek de Solla Price
The science/technology relationship, the craft of experimental science, and policy for the improvement of high technology innovation *
Research Policy 12 (1)
February 1984, 3-20
The author argues that advances in instrumentation and experimental techniques - what he calls instrumentalities - have been of major importance in stimulating and enabling both radical theoretical advances in fundamental science, and radical innovations in practical application. He supports his argument with historical examples, and concludes that explicit policies should be developed for the financial support of instrumentation. **
The purpose of this paper is to suggest some relatively small changes of emphasis in the historiography of science and technology rather than anything radically new. History of science is currently preoccupied with how its cognitive enterprise fits into the wider concern of intellectual history. Such a preoccupation makes little use of the ways new instruments and technical methods enlarge the universe of science. The proposed changes emphasize the explanatory power of such instruments and methods.
Correspondingly, the emphasis in history of technology changes from explanations based partly on the “application” of basic science and partly on the socioeconomic forces of markets for technology, to explanations based on the exogenous forces arising from changing crafts and instrumentalities in experimental science. In many ways this resolves a whole series of problems by mediating between the predilections of “internalists” and “externalists” in the history of science.
This, by itself, would constitute a reasonable revisionary program. However, beyond the intellectual quest of the historian, the mere change of emphasis opens up valuable new paths in the philosophy of science and exposes a certain incapacity inherent in the main lines that stretch from William Whewell to Karl Popper. It also resolves certain problems left unsolved by Kuhn’s analysis of scientific revolutions.
Even beyond this, the new emphasis allows us to analyze the relationship between modern basic scientific research and the high technology industries that are increasingly vital to our social, economic, and defense situations. In particular, it exposes grave and misleading shortcomings of the popular myth that basic science leads via “application” to technology. In itself this would be a valuable corrective to several errors of naive policy-making.
The theories to be outlined can probably be evaluated quantitatively as well as be argued for qualitatively. We might then gain valuable insights from the analysis of appropriate science indicators and citations, thus placing the matter of high technology innovation on a sufficiently empirical basis to let us derive policy options.
However science and technology are defined, the relationship is at the heart of a complex of deep problems that must be disposed of before
* This paper was prepared for a Workshop on The Role of Basic Research in Science and Technology: Case Studies in Energy R and D, organized by the US National Science Foundation, Washington, 12 and 13 March, 1982.
** The summary of this paper has been written by the Editors.
one can establish any main lines of historical explanation in science or in technology, and before one can engage in the rational discussion of science policy. It has long been commonplace to suppose there is some sort of direct transfer from science as a mode of knowledge to technology as a mode of know-how for making useful things and performing useful activities. Perhaps the most searching philosophical statements are those of Joseph Agassi.  On this view, the history of science is explained by reference to the utility of that knowledge for practical application. For example, mathematics had arisen through its usefulness in commercial transactions and for measuring agricultural lands; astronomy had been motivated by needs for calendars and navigation; much of physics by man’s quest for new forms of energy and communication; chemistry, by its capacity to supply fertilizers and explosives, dye stuffs, and medicines.
It is interesting that the utilitarian view of science just outlined has been accepted within political philosophies of both the extreme left and the extreme right. In Marxist accounts at the height of the Industrial Revolution, it was acceptable, and indeed illuminating, to regard science as a species of culture generated by man’s attempts to gain mastery over nature. Theorists of capitalist economies could use it as a neat “no-nonsense” stance rather than some effete “science for science’s sake” philosophy. Similar considerations still hold in an age dominated by sophisticated high technologies of weaponry and computerization, nuclear energy and genetic engineering. For strong reasons, the range of views from liberal to revolutionary emphasizes the dependence of science on its practical power over society and supposes that the democratization of science and the impetus for appropriate science and technology lie within the hands of each government. In their turn, governments conventionally look at national needs and use all the means at their disposal, including budgetary control over research funding, to design a scientific activity that fulfils particular national goals.
The utilitarian view of science looks very sensible and attractive at first sight. It also seems well supported by several case histories that spring readily to mind: the development of nuclear science by the wartime project to make an atomic bomb, the growth of petrochemistry by the requirements of an automobile industry, the discovery of vaccines by the urge to prevent dread diseases. Unfortunately, as the history of science has been subjected to increasingly deeper professional analyses, particularly within the last three or four decades, it has become increasingly evident that the utilitarian explanation for science is too simplistic and in many cases wrong.
Perhaps the most effective evidence springs not from any collection of particular case histories, but simply from appealing to almost any of the textbooks and monographs in the mainline of modern history of science and history of technology and considering their general plans and attitudes. Historical treatments of science consider the development of the various subdisciplines piecemeal by content and by period. Each section appears relatively self-contained, indicating a series of step-by-step cognitive advances in a tradition which Thomas Kuhn  calls “normal science”.
Recent techniques, using co-citation, allow this approach to be analyzed quantitatively. The step-by-step advance can be mapped in an amazingly accurate two-dimensional display in which each new scientific contribution is laid down in its right place, attached by co-citation links to the neighboring prior papers from which it springs, and knitted into a sort of research front composed of similar papers accumulating from the same background. Each paper in turn shows its locality by the place in the map and the company it keeps within its invisible college subspecialty domain, and each paper shows its strength or weakness by the extent to which it causes the map to deform and grow faster at its particular location. The entire design has some similarity to crystal growth, or to the piece-by-piece solution of some giant jigsaw puzzle proceeding outwards from a central core laid down at the beginning in the mid-seventeenth century invention of the scientific paper (if not before). Each annual co-citation map is laid down like successive skins of an irregular onion. Cut a section through the onion and it looks like a jigsaw puzzle with the recent action around the edges.
1. The Confusion Between Science and Technology in the Standard Philosophies of Science, Technology and Culture VIII (3) (1966) 348-366. Between Science and Technology, Phil. Sd. 47 (1980) 82-99.
2. The Structure of Scientific Revolutions (1962).
Fig. 1. Jigsaw Puzzle Model of the way the growing corpus of scientific papers fits together. The model shows nine successive stages illustrating the tendency of action to develop where there is action already (b-d). It shows the way in which some areas may become contained and fill in rapidly (e and f) and how islands may develop and require a distortion of the original structure before they can be fitted neatly into place (g-i). Source: Price 
A good history of science is, for the most part, an intellectual description of the dynamics of this road map jigsaw puzzle-like structure as it has grown with time, first at one place, then in another. It is generally recognized that the process of solving the puzzle, of extending the road map, appears to be transnational. Strategies are dictated by the opportunities provided by each scientific stage itself. At any given instance, certain opportunities seem to be ripe. A piece can only be fitted to the puzzle if the puzzle has grown to exactly the stage that will accommodate the new piece at its research front border. One has only a limited flexibility in putting down one piece rather than another in the short run, but each player has a feeling that the science is there to be “discovered” in its predisposed and presumably logical sequence.
All this is probably part of the great central scientific mystery that when scientists are creative, unlike artists and poets, they act as if what they create is entirely exterior to their personalities and their socioeconomic world, exterior to their philosophies and religions, beyond the constraints of language and of motivation. Each scientist may well have a very distinct personal style, but he or she assumes the objectivity of knowledge, in spite of the fact that each scientist’s perception of that objective knowledge may be different from all others. The micro-descriptions may be personal and societal, but the macro-description is sufficiently constant for one to hypothesize a single universal jigsaw puzzle and for overlapping discoveries and contested priorities to exist.
If such a scheme of cumulation were the entire story, there would simply be not much room for free will in science policy, let alone any appear to determination by the utility of the outcome in useful technologies. Kuhn leaves the very attractive loophole of a series of rare events involving revolutionary non-cumulative changes which he supposes emerge from time to time by brilliant cognitive leaps that change the paradigm of thought in the field in question. Typical revolutions would be that of Copernican doctrine, the genesis of quantum theory, and Einstein’s grand formulations of relativity.
This loophole in Kuhn’s formulation has proved very attractive to the social scientists, especially those seeking means of revolutionary change within their discipline. Nevertheless, the principal events treated thus in the natural sciences are obviously at the most subtle and sophisticated levels, as dictated by their exquisite difficulty and seminal importance. They are just not the type of event that can serve as even the thin end of a wedge to open the closed cumulation of science to the social, economic, and general utilitarian determination that such theorists seek.
Although the history of technology has also achieved much greater maturity in the modern period, there are still grave technical difficulties that prevent us from obtaining an analogous structural pattern. A great deal of what is written by professionals in the field is readily labeled as “merely antiquarian” and thereby dismissed as serious history. What happens in fact is that this solid scholarship is occupied with the translation of artifacts into words, somewhat analogous to the editing of texts that must precede the labors of the historian who wishes to use such information as a primary source. Beyond the antiquarian preparation of source material, there appears, but not nearly so clearly as in the history of science, some sense of the general nature of technological his-
3. Derek de Solla Price, Coping with the Biomedical Literature: A Primer for the Scientist and the Clinician, in: The Development and Structure of the Biomedical Literature, Kenneth S. Warren, ed. (New York: Praeger, 12-13, 1981.
tory. The same step-by-step cumulative progression can be seen: when one has designed six railroad signal widgets, the seventh is pretty sure to follow in an almost inescapable progression. What is impressive is that revolutionary changes, innovations without an apparent line of descent, seem to be a dime a dozen. Moreover, the pathbreakers do not have the fundamental and seminal importance that they have in the history of science. Some changes may be technically difficult - high pressure steam engines, atom bombs, and large computers - but others may be less sophisticated technically than the punch they pack in the socioeconomic arena: the sewing machine and the supermarket. Others may be interesting technically, but of little market effect and less impact on human destiny, like silly putty.
At all events, the history of technology, though a complex mixture of cumulative advance and unexpected innovations, all subject to considerable interaction with market forces, does not seem strongly dependent on science. There is no general way in which one can add technological footnotes to a step-by-step history of science, and there is no general way in which one can write a preface on the history of science for each chapter of a history of technology. Coupling may seem close in some fields and periods, as in early solid state physics, but these are local conditions and we can explain them elsewise.
From these arguments we reach a somewhat paradoxical conclusion: the historiography of “normal” science and of “normal” technology taken together leaves no room for the interaction between science and technology. Each seems rather complete within its box of endogenous forces. Normal science can be rather well accounted for by following its internalist history; normal technology, by its own straightforward cumulation of technical advances, together with the social and economic forces of the marketplace. If we are to look for other effects, they cannot be found in the externalist social history of science. That can inform us about scientists and their institutions but not about their transnational cerebral activity. In particular, if we are looking for a mechanism for an expected science/technology interaction, the only strong chance is to look in those parts of the advance of science and of technology that break from the pattern of cumulative step-by-step advance. In short, in science we must look to the process by which paradigms are broken, and in technology, for the source of unexpected innovations.
Several case histories within the development of science and technology may be of paramount importance for understanding the science/technology relationship.
It is almost trite to point out that writing was a rather late piece of technology, and that it is only after writing had been developed that we see the slow and steady step-by-step beginnings of mathematics and mathematical astronomy and then the other intellectual occupations that were the forerunners of modern science. Writing seems to have developed as a technique, probably beginning with the recording of names and numbers in Mesopotamia in the fourth millennium BC; it seems likely that it was a consequence of the ecology of the river valley civilizations and their need to store a superabundant harvest and then distribute stored food in the fallow period of the year. The same process seems to have occurred at a somewhat later date in the other major river valleys where high civilization emerged.
By then, humankind had had long periods of innovation and improvement in the techniques of weaponry, hunting, fishing, boating, pottery, weaving, building, etc. With writing as a prosthetic addition to genetic memory and generation-by-generation learning, remarkably high sophistication in arithmetic develops, then mathematics, then the application of mathematics to the analysis of regularities in astronomical phenomena.
It is only within the last few decades that we have fully understood the computer-like algorithmic mathematical methods of ancient Babylonian astronomy, achieved when the people who were to be Greeks were still inchoate. Two principal points emerge from this story. First, the huge and most ancient technological development of humankind could in no way be based on scientific knowledge, for no science had yet evolved. Clearly we must distinguish a variety of technology that is autonomous and not related to science; this is what is
customarily termed “low technology”.
Second, when science does develop, it is an elegant and lightly sophisticated mathematical astronomy that is pursued, because it was there and possible, because there were fortunately some problems that were intriguing and solvable, with an almost perceptible joi d’esprit in their very neatness. The astronomy that was born out of this goes far beyond anything that could be considered of practical use for the much more elementary problems of fixing a civil or religious calendar. One could always invoke some mysterious need to understand such matters for religious purposes or astrology, but there is no direct evidence to support such a conjecture.
It is not only in astronomy that this joi d’esprit emerges. Babylonian mathematical texts are full of school examples that purport to be about ditch-digging and the like but can be seen as thinly disguised occasions for the setting of quadratic and other equations, sometimes involving such practical absurdities as adding the area of a square to its length. If the first point is that a relation is lacking between the earliest science and its low technologies, the second point is that utilitarianism does not seem to motivate the earliest science and mathematics. The techniques of numbering and writing may well have arisen from needs of survival, but once there they rapidly became playthings too.
There is an historical aphorism that thermodynamics owes much more to the steam engine than the steam engine ever owed to thermodynamics. It has always been remarkably difficult to document clear cases where a new theoretical piece of scientific understanding was applied to bring forth a new technology. Almost all cases suggested have been ill-received by historians who know that the matter is much more complex than any direct application of new knowledge to practical innovation. On the other hand, the world seems full of well-established and easily recognizable cases of new understanding being wrung from the study of a newly emergent or even an old technology.
Since the seventeenth century, particular technologies have often been empirically investigated in order to gain a deeper understanding of them and, if possible, to increase their efficiencies. In the early days of the Royal Society of London there was such a concerted effort, leading, alas, to almost no effective conquests, so that, in a sense, there was a failed Scientific Revolution in this sort of study  as there were also losers, such as astrology. In the eighteenth century, further efforts to study industrial processes met with rather more success, especially in such bodies as the Lunar Society of Birmingham, and through the work of such pioneers as Wedgwood.  Later we have similar conquests of whole areas in practical metallurgy, with the steam engine, etc. The phenomenon is so usual that we do not have to coin a word for this sort of scientific study of technology: we call it “applied science”. In a strong sense, this term is a misnomer. The process is not an application of basic science to industrial needs. It is an attack by the methods of science on a particular technology. When we study the world of nature, the result is basic science. The English call it “pure” science, but that is a backhanded put-down, since by implication any other sort of science is dirty, sullied, and impure. When we study the artifactual world of techniques, the result is applied science. It can therefore be seen that the common view, in which basic science is directly applied to technology by inserting a stage labeled “applied science” is incorrect. The arrow of derivation runs from the technology to the applied science, not the other way around.
The most decisive historical case history for elucidating the science/technology relationship is one of the most important events of the scientific revolution, Galileo’s first telescopic observations. There is little doubt that the momentous discoveries published in the Siderius Nuncius of 1610 made Galileo’s reputation. The little book contains more important new discoveries per square inch than any scientific work before or since. The book created a craze for obtaining telescopes and repeating the experiments. It is these new discoveries that created the Copernican Revolution rather than any work of Copernicus himself; they created the passion for instruments and experi-
4. See K. Och, thesis, University of Toronto, 1981.
5. See Robert E. Schofield, The Lunar Society of Birmingham (Oxford University Press, Oxford, 1963).
ments that ushered in the age of modern science. For Galileo himself, I feel, it was the experience of the telescope that gave new meaning to the power of experiment. This also led to much greater public impact for his mechanical investigations that are often considered more fundamental because they lead directly to Newton and all later physics. Personally, I feel that if it had not been for the fame of the telescope, Galileo would have been just another late medieval mechanician.
How did it all begin? The telescope was, of course, a new technology arising from an already ancient craft of making eye-glasses. Not long before Galileo’s time, a technical improvement - the lens-grinding lathe, adapted from that for wood-turning - had made possible for the first time the production of deep dished concave lenses. These strong diminishing glasses were far beyond a power that was useful for the correction of vision. They became popular, however, because of their strangely illusory production of a tiny image of the microcosm, a sort of optical perspective in a world in which painters were just then strongly engaged in matters of geometrical perspective and the creation of illusion by their techniques. Once this sort of lens is available, there is only one combination of two lenses that presents itself to the eye immediately - the common Galilean telescope, which is formed by a strong concave lens held at arm’s length. Even when this is not in focus, the image of the clocktower leaps out at one, much magnified and nearer. We know now that there are two other two-lens combinations that would have worked - the Keplerian telescope and the compound microscope - but in both these cases, careful focus is critical. Unless conditions are just right, one sees only a blur.
We know now that the Galilean telescope was invented in this way in more than one place where lens-grinding was practiced.  The news reached Italy from the Low Countries where Sacharias Janssen and Hans Lipperhey of Middleburg had seized on a new-found phenomenon and realized that they had in their hands an ingenious new device that ought to be valuable. They took the route that is still common enough today of trying to sell it to the largest military spenders of their time, the Medici rulers. To this end they hyped their device as a militarily important invention, capable of spotting ships at sea and spying on armies at a distance. As it happened, this was poor technology assessment, for it was not until centuries later that the telescope assumed direct military significance, first for reading signals from ships at sea - remember Nelson using it to his blind eye - and then not until instrument-making produced field-glasses for general use in World War I.
Galileo was brought into the act as one of the first generation of new-style university professors who had to earn their living outside the cloistered professions. All his early activities had been full of the making and teaching of (rather trivial) instruments, and the seeking of patronage as an expert for hire. It appears that, with the stimulus of knowing that a tubular device with lenses at either end could make far objects seem near, he was very quick in duplicating the discovery and thereby reducing to nothing the technical property of the hopeful Dutchman.
At this point the unexpected happened. Galileo turned the device on the heavens and saw immediately the one spectacular view that is afforded by a very low-power, low-aperture telescope with a minute field of vision. It should be emphasized that using a telescope of this sort is rather like trying to peer through two keyholes in tandem several feet apart. What Galileo saw was a crescent moon. He must have noted immediately that apparent illusion that made the moon seem to have mountains and seas. I suppose it was not until he looked again a few days later and saw the moon under conditions of different illumination (it had moved around its orbit a little with respect to Earth and Sun) that the nature of what he was seeing clicked. If this was an illusion, it was acting just like the real thing. A quick order of magnitude calculation based on the shadow cast by the mountain soon made it evident that the newly discovered mountains on the moon had about the same height as mountains in the neighboring Alps. From that moment on, there was no doubt that what he was seeing through the new instrument was real. It quickly followed that he saw for the first time that Venus had phases and was illuminated by the Sun, an observation that made sense out of an outstanding difficulty in both Ptolemaic and Copernican astronomy. Furthermore, Jupiter had small planets revolving around it like a miniature solar system, and the skies were
6 See Albert van Helden, The Invention of the Telescope, Trans. Am. Phil. Soc. 67 (4) (1977) 67 pp.
full of many more stars than had ever been seen before by anybody. Later, he was to observe the rings of Saturn and the spots on the face of the Sun. All of these things were obviously “real” but hitherto beyond the reach of philosophy because they were beyond the reach of the senses.
This was a vital turning point for science. The dramatic new evidence that altered completely the nature of cosmology did so, not by any intellectual prowess on Galileo’s part, but by revealing new evidence, the existence of which had never been suspected. The telescope was not devised to seek such evidence, and it was not used primarily to gain more. Its purpose was to inject each new telescope owner into this world of what can only be called “artificial revelation”. The term is not used lightly. Galileo was not so conceited as to think that he was brighter than all previous authorities; he knew that he had been presented with decisive new evidence of the structure of nature.
It is worth pointing out, further, that the reaction of the Church was not one of blind adherence to ancient authority and holy texts. There may perhaps be some analogy with modern attitudes to evidence secured by those who ingest certain hallucinogenic drugs. We tend to decide that the revelations from LSD are due to a bending of the mind of the observer rather than to a new sensitivity to objective truth about the universe. Galileo as a good Catholic believed that he had been vouchsafed the means to these new truths. He had a “click”; he believed that what he saw was real and that the new evidence should be put at the service of the Church to attest to the glories of God’s universe. If the Catholic Church had not at that time been embattled by the Reformation and its political and economic pressure, Galileo might not have been beaten by a department of dirty tricks. Each accelerator, radio telescope, and science laboratory might now have a crucifix at the entrance, symbolizing a connection, rather than the split that then occurred, between scientific and religious knowledge.
The story does not end at this point. The popularity of the telescope caused such a boom in its manufacture that the craftsmanship developed. Growing interest led the industry a little later to produce other sorts of optical instruments, including the microscope. The fashion for instruments was extended to other devices that gave new and unnatural conditions rather than those which extended the senses, so we get vacuum pumps and electrostatic generators as well as thermometers and barometers. In the traditions of applied science, it soon happened that investigation of these instruments yielded new knowledge, such as the understanding of geometrical optics that followed the invention of the telescope and was part of the process leading to further instrumentation. The fashion for instruments that yielded artificial revelation was called the new philosophy. It was not, as has often been assumed, a means of thinking more logically or of testing each hypothesis for security, but the exploration of nature by observations that scholars and philosophers could not undertake before the dawn of this new craft of instruments.
It is reasonably clear how the craft expertise and success of the telescope led to that of the microscope. It is also clear that the new availability of this instrument led to the acquisition of new data, such as the existence of cells and spermatozoa, that changed the nature of biology as a field of enquiry. It is also clear in general terms that the microscopic and other optical techniques enabled researchers like Pasteur to solve deep biological problems and to generate techniques that in turn led to the cure of dread diseases. Such is a typical chain of causality deviously twisting a path between science and technology in that special craft of experimental science. It is this which we now suggest as typical of the fertile ground that is the main source of exogenous and unexpected non-cumulative changes in both the history of science and the history of technology.
The discovery of voltaic electricity follows a similar circuitous route as that which takes us from eyeglass manufacture to Pasteur’s microscope. Galvani was investigating the nature of the vital fluid that caused muscles to move. It was a piece of classical research into the nature of life. Because it was well-known that electrostatically generated shocks caused violent muscular reaction, Galvani, a professor of anatomy, was using his electrical machine on the easily available and dissected back legs of frogs. Noticing that the legs twitched without his turning the electrical mac-
hine, he was led to discover the cause in the contact between the wire inserted in the frog’s nerve and the laboratory bench rail in which they were hung. Volta, a physicist, now came into the act and showed that the frog’s leg was acting only as a detector and that what was really interesting was the previously unknown effect of a bimetallic contact. Rather quickly the new effect led to his voltaic pile, which was rapidly demonstrated internationally and created the same furor and rapid reaction as had the telescope.
The new effect, discovered around 1800, formed the character of most nineteenth century science and a good portion of its dominant technologies. Voltaic electricity in its beginning was not recognized as an energy source but as an agent of chemical change. It was the first new technique for changing chemical structure since the prehistoric advents of change by fire and water. Electrolysis could pull compounds apart. Within remarkably few years this force, rather than the other line of inquiry by Lavoisier that depended on the recognition of bases as substances (a change partly due to the mediation of the techniques emerging from the vacuum pump), made it possible for chemistry to have what is sometimes called its “delayed” scientific revolution. Less than one generation later, Davy had discovered half the elements of what had now become the periodic table. Within this same generation, Liebig was able to start the laboratory education and methodology that led, also extraordinarily quickly, to the discovery of fertilizers, synthetic dyestuffs, medicines of known composition, and explosives. The socioeconomic effects of inorganic chemistry are so enormous, particularly in conjunction with the post-Napoleonic reorganization of German universities, that they became a paradigm for a science-produced technology that changed the fates of nations, helped cause wars, and motivated new empires and sources of wealth. In just one more generation electricity itself, once its relation to magnetism was explored, became a similarly valuable force having major impact in its own right, and then we are off on the road that leads to Edison and Alexander Graham Bell, to electric light and power, and to the electrical transmission of signals. The twin sciences of chemistry and electricity lead to the perception of “better living through science” which is also central to the nineteenth century industrial revolution view of science as a provider of technologies. What is really going on is that the devious path between science and technology is leading to the discovery of new effects and their exploitation by suitable instrumentalities which in turn lead to new science on the one side and direct industrial application on the other. Only if one writes the history of science and the history of technology together with their common ground in the craft of experimental science can one perceive the strange paths that lead from anatomical experiments on the nature of life to such end-products as Maxwell’s electromagnetic theory, Liebig’s fertilizers, Edison’s electric light, and the modern pharmaceutical industry.
Plainly, there can be no route from basic research through applied science to technology. Basic and applied research are linked inseparably to technology by the crafts and techniques of the experimentalist and inventor.
Some general analysis of the role of instruments in the history of science is particularly important because so many misleading and naive ideas abound in the older and popular literature. 
Contrary to the popular notion, scientific instrument-making did not originate with the making of elementary measuring devices for such useful arts as surveying and navigation, time-keeping and positional astronomy. Length, mass, and time were measured in antiquity, but the making and use of the early instruments was a quite undemanding craft, principally because unaided human guesswork rivals and often exceeds that which one could achieve with simple instruments. The undoubted and impressive accuracy in ancient astronomy depends, as has now been shown several times, on purely qualitative observation, with no need for careful measurement and no thought of testing theories by such measurement. It is worth remarking that before the birth of probability theory, the differences between computed and measured values could not be handled. If the values differed, one was right and the other wrong. Not until about the time of Tycho Brahe in the six-
7. A further treatment of this can be found in Derek de Solla Price, Philosophical Mechanism and Mechanical Philosophy, Annali Dell’Istituto E Museo Di Storia Della Scienze di Firenze. 1980/1 (1980) 75-85.
teenth century did a feeling for “averaging” and permissible “deviation” enter the picture. Surveying did not become instrumentalized until the redistribution of monastic lands in the sixteenth century, and navigation by instruments was far less reliable than a seaman’s rule of thumb and chart knowledge until late in the seventeenth century.
The tradition of making complex fine instruments has a more noble and interesting beginning. In the early Hellenistic period, quite complex models were made to illustrate the complex phenomena of astronomy, then virtually the only science to have evolved to a competence visibly continuous with the science as known today. Astronomical theory frequently demands three-dimensional concepts, kinetic rather than static. Before diagrams and equations represented theories, there was quite sophisticated modeling by armillary spheres with engraved brass bands, by the use of gear wheels to make things go round in the proper ratios, and with engraved stereographic projections to view the sphere of the heavens on a flat disc. It is from instruments of this sort, sundials, astrolabes, geared planetaria and such, that a particular craft of fine instrument-making developed using the skills of the jeweler and metal-worker. Such devices and the craft proliferated and developed continuously from the ancient world, through Islam and into the European Middle Ages and Renaissance. On the way, the craft of the clockmaker developed from the geared planetaria, from which flowed the important set of peculiarly scientific techniques that later led to the general practice of machine-building. Clocks and sundials had the status of models rather than utilitarian time-tellers. As models they were philosophically tantamount to theories and very powerful in their influence on thought. In many ways it was the existence of the mechanical techniques that gave rise to a philosophy that could be called mechanistic.
In the sixteenth century, the instruments began to be of practical utility, and the craft of the instrument maker grew. By the first half of the seventeenth century, one could count about a hundred small workshops of such craftspeople in London alone. With the burst of popularity produced by the discovery of the telescope in the early seventeenth century, the trade prospered almost explosively, and the character of instruments changed from models to all sorts of new devices for extending the senses and producing hitherto unavailable effects. In the early eighteenth century almost all the things that could be easily done in this way had been done, and there was something of a slump in the production of new experimental science. The trade of the instrument maker was, however, enhanced in this period by a steady growth of a new sort of device, the special instruments made specifically for the teaching of students by demonstration. These entered educational practice in Holland and England and led to highly influential courses in experimental philosophy by such people as John T. Desaguliers and Willem J. s’Gravensande. 
The next major transformation of instrument-making came with the discovery of voltaic electricity around 1800. Almost immediately there appears a series, not exactly of instruments in the old tradition, but of unitary modules of apparatus that could be connected in various different ways as occasion demanded. In the associated science of wet chemistry, there is a similar change, in glassware and stocks of chemicals, rather than in metal work. These two sciences lead to a transition that takes us from instruments to apparatus and to manufacturers who replace the old craftspeople. The manufacturers flourish by equipping a rapidly increasing number of universities around the world that begin instruction in the new fields, first of chemistry, following Liebig’s pioneering in Giessen, and then, in the 1870s, in physics. The main thrust in such university laboratories is instruction in the experimental techniques, leading to research.
Tinkering with apparatus grew to considerable proportions, both within the laboratories by the new classes of technicians and the putative graduate students, and outside where chemistry, photography, electricity, and optics created new crafts of high technology for home experimentation such as that which made a scientist of the young Maxwell. Later techniques of glass blowing were germane to all the vacuum tube experimentation that leads to X rays and electronics. The whole tradition of early photography and radio
8 See G. E. Turner, The London Trade in Scientific Instrument-Making in the 18th Century, Vistas in Astronomy (Pergamon Press, Oxford) 20 (1976) 173-182, and Apparatus of Science in the Eighteenth Century, Revista da Universidade de Coimbra XXVI (1977).
seems to have been in the hands of ingenious amateurs rather than establishment scientists of the universities. It is interesting that the late nineteenth and early twentieth century phase of “sealing wax and string” in laboratories seems not so much a function of poverty (experimental science cost very little then) but of genius, experimenters like J.J. Thomson and Rutherford, who were not particularly clever with their own hands but had a dozen other pairs of hands to do their work.
The central issue in instrumentation is that there was built up in these laboratories a veritable armory of special knowledge of properties of materials and curious phenomena that could be played with. Any new technique served as grist for their mill of instructive experimentation to see what it would do and what was “the go of it”. A study of clouds led to the Wilson cloud chamber as a means of making radioactivity visible. The result was adventitious because no one was searching for such a method. It is the telescope syndrome repeated. And from time to time these neat effects would be raided by technology whenever a market potential became visible. It is that that rather strongly links the techniques to both scientific theory on the one side and industrial application on the other.
A huge industry of supply houses for apparatus components and instruments grew up to feed the university experimental laboratories in the period of increased professionalization and of the technological relations of sciences that were no longer “natural philosophy”. The transformation took place only as recently as World War I. The agent of change seems to have been the evolution of radio and electronics. Strangely enough, this is the line that seems to have produced the latest set of changes that have brought about the end of the old era and perhaps returned us to the main line at a new level.
Electronics had two different effects. First, it led to a new sort of big machine with Cockroft and Walton’s atom-smasher of 1930 and Lawrence’s cyclotron and the line of accelerators that rapidly became engineer-built installations that were virtually laboratories in their own right. A similar trend leads to the radio telescope as an analog of the old optical observatories, and to the computer as an independent installation. The other line of effects was that electronic craft became a special technique with its own craftspeople and a product of black boxes that continue the automaton tradition at a new high level, a simulation of intelligent processes far beyond those of arithmetical manipulations.
One would gain nothing by denying the existence of experimental tests that confirm or disprove a theory, as the philosophy of science commonly supposes. Tests deliberately constructed from available experimental techniques probably occur, but I feel they are not nearly so common as to be representative of laboratory work at the research front. They are certainly not the mainstay of seventeenth century activities in the heyday of the scientific revolution. Interestingly enough, they were particularly common in the early nineteenth century when the floodgates of electrical and chemical experimenting had been opened by the discovery of voltaic electricity, and that was precisely the period when William Whewell and others were laying down the foundations - often mythical - of the way that science was supposed to have worked. Such is the origin of that part of the classical philosophy of science, according to the school of Karl Popper.
A great deal of the actual work that goes on in all sorts of experimental laboratories consists in the discovery of new techniques for doing something or producing some new effect, then perfecting and extending the technique and using it on everything in sight. What happens mostly is that the result of such application of techniques yields only new results that fit very well with expectations derived from all previous understanding. The hope of the experimenter, however, is that from time to time, by luck and clever judgment, he or she will produce results other than those readily comprehensible within the paradigms of previous knowledge. In short, the experimenter hopes for (but does not always get!) a repetition of the Galileo syndrome when a new instrument yields a treasure trove of results that are not only unexpected but pathbreaking in their obvious significance. When that happens the new technique is a winner. Everything is thrown into making it more powerful, more general, and of wider application. It is interesting that although this is a consistent and normal pattern, we talk in terms of serendipity, lucky accident, and chance when describing
such events. The history of technology is notably convoluted,  and the craft of experimental science is perhaps its most devious branch. It lies far from the capabilities of straightforward technology assessment.
The important thing about these techniques of science is that they are not of themselves part of the knowledge system of science. They are clearly technology, an understanding of the way to do things, and often in their beginning, as with the telescope and voltaic pile, no one properly understands how and why they do work as they do, but only that they work and that they produce something new. We need a new term for these important techniques that help make new science. It will not do to call them instruments. Although the telescope fits this category, our term must let us include parts of the experimental repertoire that are labeled “effects”, such as the production of voltaic electricity, or the photoelectric effect, and such things as Cerenkov radiation or nuclear magnitude resonance. We must also include chemical processes, such as polymerization and Lowry’s method for protein determination, and biological processes, such as recombinant DNA that lead to genetic engineering. I advocate the use of the term instrumentality to carry the general connotation of a laboratory method for doing something to nature or to the data in hand.
Instrumentalities of this sort exist not only in the natural sciences but also in the social sciences and probably in mathematics. This rubric could also cover such things as a national census that yields new raw material for analysis, and public opinion polls, and various types of personal tests used in social scientific experiments. Similarly, though I argue the point with considerable diffidence, I would suppose that such major mathematical techniques as the differential and integral calculus, the solution of differential equations, summation of infinite series, quaternions, tensor calculus, and statistical techniques such as correlation coefficients, multidimensional scaling, and factor analysis should all be regarded as instrumentalities, even though they are intangible software constructs of the creative mind. If one construes such things as decisive technical inventions, very different from “theory”, that are then used for further work, the contribution of Newton, for example, to the foundations of mechanics can more sensibly be evaluated, as can the birth of the modern social sciences.
The genesis of these instrumentalities is not particularly well understood, partly because most previous historical research has focused on the cognitive history of scientific ideas, partly because there are few historians with the bench experience and the gut feelings of experimental craft to do the work, and partly because the stories about such work turn out to be highly intricate, involving relatively unknown craftspeople and technicians instead of well-honored scientists. A common feature of instrumentalities is that they are rarely accorded full recognition at birth: almost nothing would lead one to predict that a given technique would yield decisive results. One might never expect that an improvement in spectacle lens-grinding would change astronomical cosmology.
In essence, the results are quite unexpected, as when Rutherford is measuring the stopping power for alpha particles of a series of gases and realizes that something unexpected has happened only when he comes to testing nitrogen, observing for the first time what turns out to be induced radioactivity. The inventions of instrumentalities are precisely those that defy reasonable attempts to make a technology assessment. Of course, one can nearly always give very plausible arguments justifying the expense of building a new accelerator or radio telescope; but the true basis for the arguments is the hope that some discovery will come out of left field and do the unexpected.
There are a few excellent studies of the history of instrumentalities, although they make difficult reading.  The new movement among sociologists of science to use ethnomethodologies and to record what actually goes on in scientific laboratories throws a great deal of light on this particular area and focuses attention on the “playing” with instrumentalities that often dominates the general
9. See James Burke, Connections, which makes an excellent popular case.
10. A standard reference is E. Gerland and F. Traumuller, Gerchichte der Physikalischen Experimentierkunst (Leipzig, 1899), but this is both ancient and purely antiquarian. The best single monographic account I know is the excellent work by Thomas Park Hughes, Science and the Instrument-maker; Michelson, Sperry and the Speed of Light, Smithsonian Studies in History and Technology, No. 37 (Smithsonian Institution, Washington, D.C., 1976).
strategy of attack on scientific problems. 
Particular incidents in the history of science yield continuing evidence of instrumentalities. One is the case of Rosalind Franklin whose contribution to the Crick and Watson ‘double helix” story was the mastery she alone had of the technique of making good X-ray diffraction pictures from very small and badly crystallizable organic molecules, a technique she had learned and improved during her training in Paris. A single photograph from her was the vital evidence in this particular stage of the discovery.
Instrumentalities seem to be particularly dominant in the new biology. When we come to write an authoritative history of molecular biology and genetic engineering, we should be careful to give due credit to such techniques as recombination and hybridomas  and their inventors.
The fact that these newly invented instrumentalities move very often from being tools of the laboratory to a much wider commercial application is central to my argument. The process removes much of the mystery from having to assume some as yet undescribed application of the scientific understanding other than the direct use of its instrumentalities. It is almost trite to point out that if you wish to achieve some material effect, your tools, not the theories, are the instrumentalities. A theory cannot be used directly to move or change something. Sometimes the transfer to useful social application is immediately effective. Roentgen’s discovery of X-rays in 1896 was a typical accidental discovery of a new laboratory technique. Within a couple of weeks of the original discovery and its extremely rapid journal publication and transmission over the then efficient mail service, X-rays were being used by physicians to view broken bones in their patients all over the world. There is no truth to the frequent assertion that the time from invention to innovation has been decreasing steadily. The application of a new technique can still be almost instantaneous.
The most celebrated and crucial invention of modern times, the transistor, might also be viewed with this new emphasis. Instead of construing the development as a founding triumph of solid state physics in which a newly won theory of metallic conductivity is somehow “applied” to producing a new sort of semiconductor that is later produced, first experimentally and then commercially, one looks at the new techniques that were involved. There is the usual tortuous line of transfer from a technology in a quite different context, the gentle art of growing spectacular single crystals from molten substances. It had arisen in laboratory attempts to duplicate the natural conditions under which so many crystalline rocks and minerals had grown naturally. Gemstones offered a particularly strong assist, with longstanding attempts at making artificial diamonds, and eventual great success when it was shown that artificial rubies could be produced from furnaces with carefully controlled temperature gradients. The operation was so successful that it led to the production of most of the material for ruby-jeweled bearings in the watch trade of the world and thence to mass production of high-quality gemstones.
It was this technique that Bell Laboratories adapted to produce single crystals of metal to investigate electrical conductivity in the absence of the crystal boundary interfaces that were thought to dominate the electrical conductivity properties of ordinary bulk metals. When metals in single crystal form proved disappointing and unrevealing in spite of the prior inducement provided by tin whiskers that grew in switches and caused peculiar short circuits, the attempt was pushed toward producing the first pure single crystals of the semi-metallic substances, silicon and germanium, and it was from this that the unexpected conductivity properties were quickly recognized. The important point is that a piece of technique in this case yielded a substance in new form and became an immediate target for laboratory investigation and consequent major advances in theory, the bursting out of the whole central field of solid state physics. With the very same instrumentality, the opportunity arose immediately for making the transistors a longwanted device having practical commercial application.  Again, it was by no means im-
11. The best such recent work is that of Bruno Latour and Steve Woolgar, Laboratory Life, the Social Construction of Scientific Facts, Sage Library of Social Research, Vol. 80 (Sage, Beverley Hills and London, 1979).
12. See N. Wade, Hybridomas, the Making of a Revolution, Science 215 (February 26, 1982) 1073-1075.
13.For a fuller account of the preceeding events in this story, see Lillian Hoddeson, The Discovery of the Point-contact Transistor HSPS 12 (1) (1981) 41-76. The reference to the single crystals of germanium is fn. 112.
mediately obvious that the single crystals would lead to the transistor radio, let alone the computer. They arose only after heroic entrepreneuririg of the market conditions for the manufacture of the first generation of transistors. The paper by Jefferies  carries this story further by linking zone refining to vacuum switches, thus extending the example of tortuous and apparently serendipitous routes in experimental technique and technological innovation.
The central thought is, therefore, that the almost accidental generation of a newly invented instrumentality gives a means of doing something new in the laboratory and perhaps also conjointly in the world outside. In the laboratory the instrumentality has a chance to produce new phenomena that might well lead to breakthroughs in understanding. In the commercial world the same instrumentality, given the right sort of market manipulations, can create a new opportunity for application and fill a need that might or might not have been previously diagnosed.
Thus the dominant pattern of science/technology interaction turns out to be that both the scientific and the technological innovation may proceed from the same adventitious invention of a new instrumentality. In science the typical result of such a major change is a breakthrough or shift of paradigm. In technology one has a significant innovation and the possibility of products that were not around to be sold last year. There is therefore a correlation of sorts between the scientific and technological events. It is this, without doubt, that is the basis of the common but misleading presumption that somehow the scientific advance has produced the technological “application”.
It must be remembered that we are here dealing with only one class of rather important events, not with the entirety of scientific or technological change, which has already been admitted to proceed endogenously in step-by-step normal changes. It is the “revolutionary” developments, as noted by Kane in his workshop paper that are associated with basic scientific advance in this way. It is this association that makes the pursuit of laboratory instrumentalities in basic research so vital for innovation.
Even though we suppose that the instrumentality route may account for a great part of the science/technology interaction, we must not think of it as governing more than a small part of normal science and normal technology. This is true not only for basic science, which uses its entire achieved repertoire of instrumentalities to study and understand the world of nature, but also for the applied sciences, which use the same repertoire to examine the world of artifacts. The difference between basic and applied sciences is not one of method or applicability, nor of purity of purpose, but only of the subject matter under investigation. This is, indeed, the position that has been reached in scientometrics, from which it has often been confirmed that the accepted Frascati definitions on types of science do not relate well to differences of purpose but simply to demarcation of field. For this reason, all nations seem to have about the same mix of basic and applied science. Where differences exist, different countries have different mixes of technology. Some nations have a lot of agriculture and therefore a lot of agricultural applied research; similarly, countries that mine heavily have an applied science like metallurgy. This is something quite distinct from the science/technology link that produces high technological innovation.
Such is the power of instrumentalities, old and new, that they are probably also the chief agent for the sociological and substantive disaggregation of the chief scientific and technological disciplines into their constituent subdisciplines and invisible colleges. Scientists and engineers seem to be bound together in their invisible colleges, not so much by any communality of their paradigms, ways of thought, and cognitive training, as by a guild-like communality of the tools and instrumentalities that they use in their work. A high proportion of the world’s most cited scientific and technical papers come under the category of “method papers”, which act as surrogates for some particular instrumentality. When an organic chemist cites Lowry’s Method (which holds the world citation record), it is tantamount to declaring that a spectroscope or some other instrument was used. Such papers, by themselves and in combination with other instrumentality descriptors, can indeed be used to define the entire field that they blanket by being cited.
Some objective method ought to be discernible for identifying such an instrumentality from its
14. Jeffries (note: reference left incomplete by author).
citation pattern. I conjure that, in a two-dimensional citation mapping, a normal research paper can be mapped as a point in “subject space”, but an instrumentality paper, blanketing its entire area of a subdiscipline, behaves like an extended object. If one maps, not just papers, but patents as well, the patents will probably cluster even more around instrumentalities that blanket whole areas of “breakthrough”.
Another point deserves mention. When a new instrumentality surfaces in an academic context, the discoverer has a choice between scientific-style open publication, when it becomes a free good, or technology-style patenting, when it becomes a valuable good. This choice underlies a great deal of the tension in university/industrial relationships.
To apply the general theory of science/technology interaction to practical matters of national policy, we must first dissect away from the argument certain weighty economic matters that are not germane. The first of these concerns the term “development” which is customarily run together with “research” under its habitual wartime conflation of “R&D”, research and development. The two terms must be firmly disaggregated for any policy argument. Although development accounts for around three-quarters of the money and manpower in R&D, and basic and applied research only one-quarter, the manpower is very differently qualified and utilized, and the moneys are to be reckoned to quite different accounts.
For both basic and applied research, we are dealing with scientists and research-front qualified, rather academic, researchers, many of whom work in laboratories or (if they are theoreticians) around people who work in laboratories. Such people are the locus of instrumentality innovation. In the development sector it is common to use a quite different labor force, very similar to the people who are engaged in designing and making the prototypes and first-run productions of a new line of manufacture.
To caricature the point, when one spends research money it tends to go to professors and quasi-professors who happen to be without students. When one spends development money it tends to go to people with drawing boards and lathes and drill-presses who are like production personnel. I suggest that one can only make sense of input/output relationships in development expenditure by regarding it as a sort of overhead on production. When an industry is forever manufacturing new products that were not around to sell last year, the investment in development may be a very large proportion of turnover. Development is not thereby the source of the innovation but only a means of implementing an innovation already made. Innovation, according to our theory, proceeds from the new inventions within the craft of experimental science. It comes from that part of research, both basic and applied, that operates in laboratories with experimental instrumentalities.
A second matter to be dissected from the economic argument concerns funding and the social warrants for science. Analysts of Science Indicators ill-recognize that when money and other resources are spent on “research”, it matters a great deal what exactly that money and those resources go into. The point is obvious if one operates only within a particular nation, especially the United States with its very strong and explicit tradition for research funding. The merest acquaintance with other countries shows that this area cannot be taken from granted.
In many nations of the world, funding for research can include no provision for salaries to researchers. They are civil servants and their care and feeding is part of an establishment budget separate from that which enables them to undertake research along with, or instead of, their other functions within an academic or bureaucratic environment. In some nations, research funding may have a component, perhaps even a dominant component as a Canada, of graduate student support as assistants in research. In other countries such support is a matter of national manpower policy rather than research support.
In some countries, the U.S. practice is followed. In the research budget a certain, often quite high, proportion of “overhead” is included as a means of subsidizing the universities and other bodies for their indirect expenses in connection with the research. In the United States this arose historically from the process that generated the National Science Foundation as a continuation of the wartime high-level organization. It was also, however, almost a rescue device to enable the universities to
cope with the postwar flood of scientists being demobilized from a large-scale national effort in the atomic bomb, radar, etc. The consequences for the development of a new entrepreneurial spirit in scientific research has been particularly well, if controversially, documented and adversely criticized by J.R. Ravetz.  Some national science policies, as in Brazil in the early modern period, have no salary or overhead components and spend very little money on instrumentation or books because of currency control problems. The bulk of the research funding went either for such “development” as the purchase of government computers, or for tickets on the national airline to bring foreign experts and teachers in and to take Brazilians out.
In the United States, the moneys spent on research have all the major component applications. Some is salaries of principals; some is student assistantships and postdoctorals; some is institutional overhead. Only a fraction is spent on the actual expenses of research itself, including the vital part that pays the way for the all-important instrumentalities. It happens to be peculiarly difficult to get figures from the U.S. Government, or from any other government, that disaggregate the real expenses of alleged research support.
From the National Science Foundation there should be ample fiscal data analyzing the line items in research budgets, but the only published accounts I know  give inconsistent figures for 1971-74 that are probably, however, reasonably accurate. The salary and wage component was about half the expenditure; indirect costs, about one-quarter and rising rapidly (by now, nearer to one-half); and actual research expenses, the remaining quarter. The latter includes such components as travel, publication, and computer costs. Actual expenditure on laboratory facilities for the period was thus probably in the range of 10-15 percent of alleged research expenditure. It is probably rather less today and still declining.
Evidence of the decline is also afforded by Helen S. Milton’s Cost-of-Research Index 1920-1970  prepared for the Department of the Army. According to her data, the R&D cost per technical man-year, relative to the gross national product, began to fall in 1965 and has been decreasing ever since at about 2 percent per annum, indicating less and less spending on more and more expensive instrumentation. A comparable but more detailed study for the United Kingdom reached similar results.  Data from several laboratories indicated an increased cost of research due to sophistication of apparatus. Costs ran about 3 percent per annum in the period studied.
From these and all the other uncertain and often contradictory data, we would be well advised to secure all the indicators we can for the actual cost and expenditure for experimental facilities and probably also for the staff of technicians who are likely to be the retainers and communicators of innovation in the essential instrumentalities. Little of the data that would be needed to assess the actual national investment in laboratory instrumentalities and their technical personnel appears to have been published. We know neither magnitude nor trend. It is, however, easy to guess that the overall total is about 10 percent of the research funding, and therefore 2-3 percent of the R&D funding at the national level. It is almost certainly declining since 1965 in constant dollars and in dollars per scientist, in spite of a real cost increase in instruments due to greater sophistication. As to communication at the research front of the craft of experimental science, we have no studies of communication or mobility of flow among technicians. Some anecdotal evidence suggests that a new technique is very rapidly transferred from one laboratory to another, from universities to industries, and from country to country, because the major institutions for experimentalists have a large commuting circuit of visitors who are quick on the uptake in these vital matters. 
There is, fortunately, rather good cross-national
15. Scientific Knowledge and Its Social Problems (Clarendon Press, Oxford, 1971).
16. NSF Databooks for 1974 and 1975, NSF 74-3 and 1974-? table 9.
17. Research Analysis Corp., RAC-TP-430, July 1971.
18 A.V. Cohen and L.N. Ivins, The Sophistication Factor in Science Expenditure, Department of Education and Science, Science Policy Studies No. 1, (HMSO, London, 1967).
19. For further argument on this point, and evidence from the instrumentalities of petrochemistry, see Yakov M. Rabkin, Science and Technology: Can One Hope to Find a Measurable Relationship? Fundamental Scientiae 2 (3) (1981) and the references therein. For a further discussion of petrochemistry and a similar story for nuclear magnetic resonance, see John D. Symes, Policy and Maturity in Science, Soc. Sci. Inform. 15 (2/3) (1981) 337-347.
information regarding the scientific instrument industry and its rate of innovation. It has not been sufficiently emphasized previously that this industry has an importance for high technology innovation that vastly exceeds the relatively small economic volume of the instrument industry itself in domestic and foreign sales. The point is well made with a masterful marshaling of statistical data by Keith Pavitt.  Scientific instrument firms are quite often spin-offs from great national facilities in experimental science, such as the Cavendish Laboratory in Cambridge, England, and MIT in Cambridge, Massachusetts. Frequently also the mechanism for the entrepreneuring and expansion of such crucial high technology laboratories has been government procurement both in wartime and peacetime. One must suppose that policy in innovation and manufacturing in this industry has been motivated at least as much by the force of procurement as by any research funding.
A good measure of innovative activity in the instrument industry is to be had from the patent statistics in this class. I have shown in a previous study  that some countries are notably active or inactive in instrument patenting relative to their general patenting. Belgium, for example, has an activity 92 percent greater than expected; Japan, 32 percent greater; the Netherlands and Italy have a patenting activity 32 percent and 34 percent less than expected; and Canada, 28 percent less than one would expect. The United States is relatively normal in patenting scientific instruments in its own patent system, but the position might not be so sanguine if one analyzed U.S. activity in the other major patenting systems of the world.
A very detailed economic and organizational study of the world’s major manufacturing nations was published in 1968 by the OECD in Paris on the occasion of the Third Ministerial Meeting on Science. Their report on Gaps in Technology Between Member Nations contains a special volume for the sector report on Scientific Instruments, including a report on the U.S. industry. Their data substantiate many of the points already made and give more economic information than we can analyze even in the most general terms.
The economic problems of the United States are due in a large measure to a pronounced drop in the productivity of industry. A major dislocation of this sort may be more influential than the more visible imbalance between national revenues and budget expenditures in the internal manipulation of the economy. If one unpacks the concept of productivity, as has been done by M. Boretsky,  one finds that the balance is changing between low technology and high technology in determining the balance of trade and therefore the quality of life of this nation vis-à-vis the rest of the world. The United States has begun moving rapidly from a nation where most of the working population were in manufacturing to one in which they are predominantly in service industries, dominated by the information industries and their associated high technologies. Within a few decades, only a few percent of the labor force will need to work in agriculture, and not much more in low technology, to satisfy all national needs and perhaps even have some production for export. The overwhelming majority of workers are in service-related industries. It is from these that we must derive our export surplus to maintain a high quality of life.
Just as that has been happening, the United States reached a built-in saturation when the exponentially growing costs of scientific research investment in science outstripped any reasonable portion of the available budget allocation. This process was independent of any deliberate policy of a national budget. It mattered relatively little whether the government was kindly disposed to the support of scientific research. The crunch had to come some time within the decade, and it came with remarkable and unforeseen suddenness in 1965, spreading out rapidly to the other very developed nations of the world as their own exponentially growing science budgets increased beyond available funds.
We have thus been in virtually no-growth conditions for the support of research for the last 15
20. Summary of Main Findings, OECD Conference on Science Indicators, Paris, September 17, 1981.
21. The Analysis of Scientometric Matrices for Policy Implications, Scientometrics 3 (1981) 47-53; see especially table 2, col. XV.
22. Am. Sci. 63 (1) (January/February 1975) 70-82.
years. There has been indeed some measure of retreat. The actual loss has been about a 15 percent per annum decline in constant dollars, relative to the “normal” growth rate that the world had enjoyed for more than a century. The outcome pulls the rug from under our basis of innovation in the high technologies at the very time that it has become imperative to foster innovation and its consequent increase in productivity as a basis for our service economy. We must therefore use whatever theory we can muster to repair the damage, to increase high technology productivity in these circumstances of increased importance but of unavailable general funding.
Now that we have identified a dominant source for the vital innovation, we would be wise to give this source priority over the other expenses of scientific research and development. The route of action begins with a sensible fiscal accounting that makes clear what we are doing. A major step would be to disaggregate and treat quite differently the expenses of research and of development. Development should be regarded as part of the expenses of production, an overhead on innovative industry rather than an investment, and it should be taxed and funded on that basis, leaving policy to be dictated by the market and by government procurement. Research expenses need to be disaggregated into four separate components, each with its own policy: support of research manpower, students support in the breeding of new manpower, institutional support by overhead expenses, and the direct expenses of research with special attention to experimental costs.
The big question is, of course, whether we should try to modify the balances among these components. Anything that can be done to separate R from D, and to shift government funding away from D and into R, will automatically cause more innovation and less production of the thing already innovated. Since D is so much bigger than R, the nation gains. For example, if we suppose that in some typical area the R is 20 percent of the whole and the D 80 percent, the R-driven innovation can be doubled by reducing the D to 60 percent, and this is only a 25 percent reduction in production.
In the matter of the components of research funding, we also have considerable flexibility. It is revolutionary to suggest that the entrepreneurial difficulties in research can be met head-on by getting rid of large chunks of manpower support, or at least treating it quite separately from research support. But suppose we retreat partially to the pre-NSF conditions of having academics and physicians earn their keep by teaching and giving health service and require them to do research in order to have something to teach and deliver. Several other nations adopt the policy that in general it is not appropriate to buy research labor from people who already have a useful job in society. One needs to pay their additional research expenses, but it may not be wise to spend so much money motivating people to do research when one might get it, at least some of it, for free. The policy therefore would be to move money gradually away from salary support toward research expenses in apparatus, technicians, hardware, and buildings. We really need to separate people support from research funding and planning. Since it is obviously too late to cut the federal umbilical cord that links the welfare of our research institutions and manpower to funding, it would be wise to separate crucial decision-making that affects them from equally crucial, but different, decisions in research enablement and the provision of instrumentalities. By intertwining both jobs, we may be doing both badly.
The matters of student support and institutional overhead are separate issues from research funding, and they deserve to be treated as such. They are equally vital but they exist in a dimension different from the mainsprings of innovation, even though a large section of research publication is, in fact, performed by a transient population of graduate students and post-doctorals who are training for academic and other posts.
Lastly, the point must be made that we now have a useful means of fostering innovation that does not require government funding but relies instead on interaction between all places where a craft of experimental science is practices. Since so much in innovation depends on these adventitious inventions of new instrumentalities, we ought to do whatever we can to promote technology transfer among all sectors of activity: universities, government laboratories, and industry. Clearly we need to know much more about the funds actually spent on experimental costs and about the scietific instrument industry that exercises a leverage on innovation and scientific advance out of all proportion to its relatively modest size in econom-
ics and manpower. I applaud recent NSF program attempts to fund research instrumentation directly, especially since resources have deteriorated badly and will get worse without special intervention. It has been customary to view such interaction as being concerned mainly with the scientists and other professionals, but in the light of the present findings one might look rather at the interchange and availability of technicians, apparatus, methods, and the general gimmicks and notions that tend to be so important in the craft of experimental science. Transmission of craft knowledge to techniques rather than transmission of theory is probably the chief outcome of spending money to enable scientists to travel to meetings and otheri laboratories. It is highly desirable that all sectors get an opportunity to play around with any new gimmick or instrument, any novel material or effect, just in case that instrumentality will yield, on the one side, scientific advances, and, on the other side, unforeseen high technological innovatior leading to new markets.
Yale University. New Haven, CT 06520, USA