Derek J. De Solla Price
Is Technology Historically Independent of Science? A Study in Statistical Historiography
HHC: titling and index added.
Technology & Culture VI (4)
Fall 1965, 553-568
Physical Science and Industrialism may be conceived as a pair of dancers,
both of whom know their steps and have an ear for the rhythm of the music.
If the partner who has been leading chooses to change parts and to follow instead,
there is perhaps no reason to expect that he will dance less correctly than before.
Arnold J. Toynbee 
The definition of science and of technology is such a traditional preoccupation of historians of those subjects that few professionals or amateurs in these fields have escaped the tempting urge to begin their historical deliberations in such a fittingly scientific and orderly fashion. It is not my purpose to comment on definitions that have already been made, or indeed to add to their embarrassing profusion, for it seems to me that, even when the job is well done, it has been sterile and without fruit. So far as I can find, the only use to which most of these definitions have ever been subjected is that of deciding what range of subject matter should be included in a particular book or in a chapter of a book. In this respect it is somewhat similar to the recent concern of some of our colleagues with the problems of “periodization and classification.”  While such precise delimitation of the subjects may be useful and even admirable for splitting a complex whole into more easily assimilable parcels, it offers little by way of historical or scientific explication of the manner in which the development of science and technology has proceeded.
It is, of course, by no means clear that one should be able to produce anything much more useful or productive than this sort of definition.
DR. PRICE, Avalon Professor of the History of Science at Yale University, is the author of Science since Babylon (1961) and Little Science, Big Science (1963).
1. Introduction: The Geneses of Civilizations (“A Study of History,” 12 vols. [New York, 1962J),I, p. 3, n. 1.
2. See, e.g., the papers by Bonifati Kedrov, Eugeniusz Olszewski, E. Rosen, D. A. Wittop Koning, and S. V. Schoukhardine in Organon, No. 1 (Warsaw, 1964).
One could hardly expect, for example, to set up definitions or concepts that would be anything so potent as those of force and mass in terms of which Newton’s laws of motion can be formulated and an elaborate rational mechanics derived. It is, however, sufficiently reasonable to invite some speculation to see if there are perhaps definitions, concepts, or even models that would enable one to make hypotheses about the growth and structure of science and of technology and in particular about any relations between them.
In this light, it may first be remarked that one particular type of model is already so widespread as to be commonplace, so accurate as to give rise to a rich conceptual terminology, and so obvious that one does not often stop to think that it implies quite deep statements about the nature of our subjects. I refer here to the verbal-imagery models that scientists and technologists use in talking about their work. Perhaps the most common is that extended metaphor in which science becomes a geographical terrain that can be explored and traversed, divided into regions like countries and continents that have their frontiers and their borderlands, and conceived of as varying in terrain from difficult mountains and key passes to broad plainlands and fields. In other metaphors science becomes a tree having roots, a main stem, and branches that are subdivided into twigs; or it becomes a house of rooms which are being set into order and built; or a pyramid of bricklike discoveries and theories which are set, one on top of the other, by successive generations of workers at the research front. As for technology, it is often conceived of as something “growing out of” science, or “giving birth to” scientific interests, or in some sort of dialectical interaction with it, as implied by Toynbee’s suggestive image of a pair of dancers to the same music.
Now all of these pieces of imagery are in some degree valid models that can be tested to see if the analogies provide relationships that have not been deliberately built in the model or made obvious by its conception. Thus we may consider the geographical metaphor and ask if it is indeed true that science can be mapped in any way onto a plane. If it can, then two points near together, like New York and New Haven, cannot be at very different distances from a third point far away like Montreal. Is there any concept of “distance between scientific fields” that would make it reasonable to think of organic chemistry as being nearly adjacent to pharmacology but also to hold that neurophysiology was near to pharmacology but very distant from organic chemistry? If strong examples can be found to violate one’s intuitive expectations, then with such a metric or concept of distance one cannot map science on a plane, and to save all being lost, one might then go on to inquire
if what we really felt was that some sort of mapping in space of three dimensions or more would not involve such conflict.
It appears then that any such model for science or for technology can be questioned about the extent of its analogy, and if it is found that some of the analogies hold well, it is reasonable to suppose that we might be led thereby to extend and refine the model so that it explains features and phenomena that were previously unrelated. With this end in mind, I shall now proceed from a model of science, previously reported,  which achieves some success in improving qualitatively the sort of thing that is included in the geographical metaphor and which adds to it a quantitative treatment that makes it useful for analyses of manpower and literature. It will then be suggested that a similar model to this, with some necessary modifications, would be applicable to technology and give rise to parallel consequences in qualitative and quantitative studies. Comparing the two models one may then derive a little insight into the ways in which science and technology behave and, hopefully, some knowledge as to what is plausible in the nature of their interaction.
The model we present for science was originally developed in the course of quite different studies concerned with the analysis of scientific literature and manpower, both ancient and modern;  I shall not here report it in anything but the broad outlines that bear upon this new and unexpected application. The basic technique employed was originally that of a simple head count of men and the scientific papers they published, though this has been reasonably easy to elaborate without structural damage by paying attention to the way in which the papers and the men relate to each other, the journals in which the papers are published, and several indicators of the quality and scientific importance of papers, men, and journals.
It happens to be rather easy to define the journal literature of science; there are good bibliographical compilations that list by name the fifty thousand journals that have ever been published in the world and the thirty thousand or so that are alive now. If one is not satisfied by the subjective criteria for inclusion in such a list it is quite easy to make an operational definition that generates the complete corpus of literature
3. In my “Statistical Studies of Networks of Scientific Papers,” Symposium on Statistical Association Methods for Mechanized Documentation (Washington, D.C.: National Bureau of Standards, June, 1964), Science, 149, No. 3683 (1965), 510-15.
4 A general account of this earlier work is given in my Little Science, Big Science (New York, 1963).
by making use of the fact that authors of papers have for some time made a practice of citing previous publications that have been made use of in any paper they publish. Clearly the citing of a paper does not necessarily imply it has been read, and even more certainly not all papers read are cited, but there is some significance in such citations. Even if this is weak, one could start with one journal or group of journals and get, from the citations within its papers, all the other journals that contained papers to which reference had been made. These new journals in turn could be used to iterate the process until no new journals appeared in the list. In fact, such a list becomes sensibly complete after relatively few iterations, and if one is satisfied with less than perfection, as practical libraries, it can be found that a much smaller list than the world total can account for a very large majority of all papers referred to. Thus it is probable that 80 per cent of all scientific papers can be had from about one thousand journals, and that 80 per cent of all chemical papers would come from the biggest specialized journals totalling less than one hundred in number, and even that better than 95 per cent of the chemical literature could be had from less than two hundred of the most eminent journals. As well as generating a much more manageable (even if slightly incomplete) list of containers of the scientific papers, it might be noted that this iterative process can be extended to include all scientific books to which reference has been made.
With the journals and therefore the papers (and perhaps the books too) defined as a field, one may then proceed to the writers of these contributions. Instead of inventing a special term for a person who has ever published a scientific paper, I propose to use the term “scientist.” It is not an ordinary definition nor a familiar one, but it is one that can be used for further statistical studies. If one is dissatisfied that the standard is too low, it is quite trivial to compute from this the numbers of authors of two or more papers, or the number of authors whose papers have been referred to by others on one or more occasions. The calculation is not too difficult whatever demands of this nature may be made, simply because there is a rather constant and well-mannered distribution of the number of people writing various numbers of papers in a lifetime or in a year and of the number of papers referred to various numbers of times in similar intervals. These distributions are in fact exactly the ones that one would expect on the basis of simple hypotheses, and thus without any empirical assumptions any definitions of this form are easy to handle and transmute from one form to another. Obviously, however, a “scientist” in this sense has nothing to do with whether or not the person has been trained, whether he is employed as an academic
or industrial scientist, or whether he has done any amount whatsoever of scientific work, however significant, if it happens not to be published.
Having now defined the body of scientific literature and taken a scientist as a person who adds to it, we may next inquire about the relations between one person and another and those between one paper and another. Let us first remark parenthetically that our usage implies that the chief end product of a scientist’s work is the paper that he publishes, and this accords well with his obvious motivation to get his work into the eternal archive of the Literature. Above all, the scientist is a person who wants to publish; only secondarily does he want to read. His reading, or rather, his awareness of what others have written, is what shows up in the references of his papers, and it is to these that we may now turn for light on the linkages between men and fields.
On examination, references in scientific papers arise as a result of two processes. The first draws equally, and virtually at random, on all of the previously published literature of the field, any one paper being just as likely to be used as any other paper of equal merit. Papers differ markedly in their merits and therefore in the amount of citation to which they give rise, but on the whole this merit stays nearly constant with time or decreases only rather slowly. What happens in fact is that the average paper gives rise to about one citation per year, but this is because the number of papers available to cite it grows exponentially (doubling every ten years) while the chance of its being cited in any particular paper decreases by a similar factor of two in the same period. The second process is not at all randomly distributed; it shows strong tendencies to draw papers together in clumps by a network of multiple connections, and there is a strong trend toward links between papers separated in time by short intervals of only at most a couple of years.
The first process is a random raiding of the past literature; the second shows a marked research front of papers cumulating one on top of the other, knitted together by short range forces. The knitting seems to contain a lot of dropped stitches, which are what divides the research front into its clumps, these clumps seeming to correspond with the work of something like the order of one hundred scientists who probably constitute the peer group of a typical New Invisible College of all the people who really do the work at that particular segment of the research front.  Perhaps it is no accident that there have been, since the beginning of scientific journals, the order of one hundred scientists
5. There now begins to be a considerable literature on the phenomenon of the New Invisible Colleges; for a summary of some of this see Warren O. Hagstrom, “Traditional and Modern Forms of Scientific Teamwork,” Administrative Science Quarterly, IX, No. 3 (December, 1964), 241-63.
for each serial founded, presumably to minister to the needs of a new group.
I now suggest further that the first process, that of an unstructured drawing on the total archive, corresponds to the normal process of scholarship shared by the sciences and the humanities. A whole segment of learning is fed to the scholar by his training and reading; it matures, so to speak, in the barrel and then becomes the matrix for his own contributions. The second process I would regard as peculiar to the sciences and responsible for that mid-seventeenth-century development that made the sciences begin to burgeon at an exponential rate many times faster and characteristically more surefooted than any non-scientific learning. In fact, I think one might draw up a spectrum of subjects ranging from pure science to pure non-science by measuring and arranging in descending magnitudes the ratio of the research-front citation to the archival citation. There is little doubt that such subjects as particle physics and molecular biology would be on top of that scale and the taxonomic biological sciences toward the bottom. One might well find, however, that Assyriology and some parts of history, economics, and linguistics might be near the top, while other parts of history were near the bottom.
I hope I have now exhibited sufficient of this model of science and its underlying concepts to show that it can be made to correspond readily to the more familiar linguistic metaphors and that it can give usable statistics and other quantitative results as well as throw light on such qualitative issues as the difference between scientific and non-scientific scholarship. I have considerable hope that this model can be refined and analyzed, using such computer-handled data as are involved in the making of citation indexes, so that the field of science can be analyzed by its sections and relations and the entire literature and manpower monitored in its changing patterns on a continuous and historical basis. It may seem a little odd that all this might be possible without appeal in the first instance to any examination of the substantive content of any paper, but all one is doing is putting some credence in the author’s choice of journals and references as a means of fixing co-ordinates for his work. Inevitably the background noise is high, but the easy success of these analyses seems to indicate that the noise blurs rapidly, and even a small amount of orderly pattern shows up quite markedly.
The model just displayed involves no distinction between science and technology, though indeed it happens that many parts of the world
total of literature, as we have defined it, seem to be far toward the non-science end of the spectrum, showing very little of those short-range connections between papers that indicate one has the special scientific process of interaction with a close group of peers who are not merely contemporary but actually treading on each others’ heels. In small part, these non-sciences consist of the few special sciences involving such things as biological taxonomy, where preference is given to the citing of a first published description, and in another part they are obvious non-sciences that have been included by taking every paper in journals that publish or review humanistic as well as scientific scholarship. Unhappily it would appear that the great bulk of non-science emanates from journals that one would intuitively describe as “technical” rather than “scientific.” Certainly not all technical journals exhibit unstructured citation, but even in serials of this sort containing many papers of a “scientific” variety complete with short range referencing, it is obvious at a glance that there exists a quite different class of publication, almost bereft of bibliography, though sometimes citing a previous paper by the same author(s).
Such papers seem to report on the making of some new machine or drug or a testing of one previously described. It seems that after such a publication there is not habitually any further reference to the article or any direct building on it. That does not seem to be the purpose. This is all very queer, for there are many more engineers than scientists, and the great bulk of the journal literature is technical by any standards rather than “pure scientific.” One would suppose that if technology and science had similar structures one could find many peer groups of technologists cumulating their contributions. Quite clearly no such cumulation is evident in the literature in many of the chief branches of technology, though it must again be emphasized there exist fields that are traditionally “technology” which exhibit strong evidence of a structured cumulation of literature - electronics would be a good example.
Some light on this may be had by means of a cursory comparison of manpower statistics derived from the literature with those more conventionally obtained. If one takes such a field as physics there is reasonable agreement between the numbers of writers of papers and the number of people who claim to be trained and employed in that field. For chemistry the number of practitioners exceeds the number of writers - even if one counts all those who had ever written in the field and not merely all those writing now. For all engineering and similar technologies, the number of practitioners vastly exceeds the number of writers, even though the technical literature is voluminously greater
than the scientific literature. It appears then that, in addition to the small part of technology that shows typically scientific cumulation with research-front structure and the large part of technology that has no more cumulation than non-scientific literature, there must be an even larger body of practitioners of technology who are not producers of the literature.
Because of this it is certainly incorrect and contrary to intuition to attempt a definition of technology and technologists in terms of the technological literature. Clearly, the published paper is not, in general, the end product of a worker in a technological subject; he appears to be instead concerned chiefly with the production of an artifact or process. What then is the role of the literature in technology? I suggest that for the most part it is produced as an epiphenomenon. It comes about because many technologists have had scientific training and know full well the code of behavior of the scientist in which publication is not merely right and proper, but a high duty and a behavior expected by peers and employers. Since this activity does not, however, in most cases produce the highly interlocked research-front structure of science, it must be regarded in this light as a sham. Possibly the literature has some other function in technology, but clearly it does not feed on itself to grow at the mighty rate displayed by science.
Although the literature of technology does not cumulate in the same way as that of science, it is quite evident that it shares with science the rate of growth and surefootedness that causes it to burgeon at a rate outstripping all else in the last few centuries of civilization. For this reason I find it tempting to suppose that technology must have some mode of structured cumulation, similar to that of science, but not enshrined in the published literature. I suppose that at each part of a technological research front there exists and cumulates a “state of the art” that is familiar to a peer group of practitioners who are adding to this art in a structured way.  If we suppose that this non-verbal state of the art behaves in an analogous way to scientific literature, it would
6. A most telling and perceptive comment was made on this point in the discussion following presentation of this paper and in subsequent correspondence by Bern Dibner, Burndy Library, Norwalk, Conn. Dr. Dibner drew attention to the fact that there exists a special literature, quite different from normal technological journals, that strikingly reflects the state of the art in most fields of engineering. This special literature, also contained in the advertising pages of engineering magazines, consists of the catalogues and handbooks describing machines, components, tools, etc. Quoting Dibner: “Each engineer seems to resort to a dozen or several dozen catalogs which he uses for constant or occasional reference. These are furnished gratis by the apparatus manufacturers who include handy pictorial indices, color sectional dividers, and special binders to hold the printed pages together. Most industries being highly competitive, the most complete and logical catalog will draw [the prospect’s attention first and thereby gain the best chance of being specified. In addition to individual catalogs there are companies that gather and issue catalog material for entire industries such as electrical transmission, construction, aircraft, etc. Sweet’s and McGraw-Hill catalogs are big business. Engineering handbooks have a long tradition. Marks’ (mechanical), Pender (electrical), Knowlton (electrical), Carnegie (steel sections) are representative. There are thick handbooks for every specialty. Unlike company drawings, these catalogs and handbooks are the property of the engineer, are embellished with his notes and travel with him from job to job.”]
HHC: [bracketed] displayed on page 561 of original
appear that the process of normal growth is for small accretions to be made, bit by bit, with occasional nodal points that begin some new strip of the knitted structure. In general, new technology will flow from old technology rather than from any interaction there might be between the analogous but separate cumulating structures of science and technology.
At this stage we must draw more understanding from our separate models of science and technology before the nature of their interaction can be further examined. It has already been remarked that the traditional motivation of the scientist to publish is not shared in general by the technologist. One might even conjecture that the traditional motivation of the technologist is not to publish, but to produce his artifact or process without disclosing material that may be helpful to his peers and competitors before his claim to the private property of the technological advance can be established. In science the publication is the medium through which that private property is established, and approbation is secured to the extent by which that publication is helpful to others. In technology the publication may well jeopardize the private property unless there is a long gap in which to secure priority of production and the winning of a patent, and the longer the gap can be before someone else is helped to advantage, the bigger the gain in property of the innovator. Roughly speaking, science is a cumulating activity which is papyrocentric, while technology also cumulates, but in a papyrophobic fashion.
If we look at the consumption of literature, rather than its production, we find that again the behaviors of science and technology appear to be opposite and even complementary. In spite of the fact that they cumulate by building on the papers of their colleagues, scientists seem to have considerable resistance to reading more than they absolutely must. It is now well known from the work of Charles Y. Glock, Herbert Menzel, William A. Glaser and Robert H. Somers  from the American Psychological Association’s Project on Scientific Information
7. The Flow of information among Scientists (New York: Columbia University Bureau of Applied Social Research, May, 1958).
Exchange in Psychology  that at the present day a great deal of the communication job of a paper is made before publication in the open literature. In fields that are cumulating strongly, the news of research flows by personal contact and verbal report through the Invisible College and the surrounding peer group, and apparently this form suffices to a large extent. In consequence, by the time the paper is available in formal publication all the juice has been squeezed out by the peer group, and the paper is already well behind the research front of current work by the peer group. Though the process has become blatantly obvious only in recent times, it must have gone on to some extent throughout the period at issue, since the mid-seventeenth-century invention of the cumulating device of the scientific paper as a special sort of an atom of knowledge that could fit together with other similar atoms.
The literature consumption of the technologist is quite different. Though seeking to give little prior advantage to others, he is anxious to gain all advantage for himself by knowing all he can of the technical advances of others and of the scientific advances that might in his opinion have bearing on his technology. To put it in a nutshell, albeit in exaggerated form, the scientist wants to write but not read, and the technologist wants to read but not write. Less dramatically, I would like to split research activity into two sharply defined halves; the one part has papers as an end product, the other part turns away from them. The first part we have already identified with science; the second part should, I think, be called “technology,” though here there is more conflict with intuitive notions. By this definition, it should be remembered there is a considerable part of such subjects as electronics, computer engineering, and industrial chemistry that must be classified as science in spite of the fact that they have products that are very useful to society.
The first thing that may be said about the manner of interaction between science and technology is that it seems to proceed only slightly and with great difficulty through the literature. The technologist’s searching of the technical literature is doomed to failure because other technologists are not interested in helping him to their own disadvantage; in fact, it is for this reason presumably that one finds virtually no cumulation and few references in the greater part of the tech-
8. William D. Garvey, Belver C. Griffith, et al., Reports of the American Psychological Association’s Project on Scientific in formation Exchange in Psychology (American Psychological Association, December, 1963), Vol. I.
nical literature. Even if one looks at the patent literature it becomes evident, now that we have Garfield’s Citation Index,  that patents do not form a strongly cumulative network of citations. Few patents are cited more than once in any year, and those that are have at the most very weak chains showing linear arrays of patents improving on previous ones by the same person or group.
If the technologist searches on the other hand through the literature of science it becomes evident that this has not been published for his benefit but for the peer group at the front, in the first instance, and for the eternal archive in the second place. Because of this it must be presumed that science cannot flow into technology from the literature at the research front but rather from some position well behind this front. Basically, the technologist could only monitor the scientific research front if he were in his own right an active peer in an Invisible College. Short of this, he must retire to the level where such knowledge has been assimilated as part of the eternal archive. Does it happen then that elements of technology somehow arise and become attached to parts of the network of science behind the front? To use one of the familiar metaphors, is technology a sort of fruit that grows from the branches of science rather than from its delicate end twigs? Assuredly, if this happens it can only be a small part of the total activity of technology, for there would then be no obvious way in which the corpus of technology could form a well-connected and cumulating whole. I suggest that the metaphor of the fruit might well hold in some measure for those subjects like electronics which we have had to classify as scientific, but which clearly gives rise to a considerable body of useful application and artifactual end products. The greater part of technology, however, does not seem to have such strong links with any particular portion of scientific literature, nor does its own literature cumulate cohesively. We are therefore left with the image of a technology that cumulates at much the same rate as science, but not through its literature.
At this stage we may draw more consequences from the model. The complementarity of science and technology in its attitudes to the literature is a matter of considerable embarrassment for the technologist. The scientist need only find ways and means of getting his material published with the maximum possible dispatch, the technologist wants to monitor this literature. It is therefore reasonable to interpret the growing hue and cry about an information problem largely as the agitation of the technologist in attempting to deal with a scientific
9. See E. Garfield, “Science Citation index, a New Dimension in Indexing,” Science, 144, No. 3619 (1964), 649-54.
literature that has not been written with his interests in mind. The technologists are in a quandary, for by the nature of the beast they cannot or will not write the papers they need for themselves, and they do not want the scientific literature in the way in which it is automatically organized for use by peer groups at the research fronts of the Invisible Colleges. In the absence of any new and radically large-scale computer operations to form an automatic and up-to-date encyclopedia of the sciences, one falls back on a well-tried expedient to ameliorate and cope with this problem. The expedient is the natural process of scientific education in which all the knowledge won so far is packed down and fed to the student of science so that at his entry to the research front he is aware of the state of affairs without a complete monitoring of all strands of the literature. For some limited time thereafter he can cope with recent advances, but after a while he is capable, if at all, only within the bailiwicks of his own Invisible College.
It may well be that a similar process holds within technology and that the apprenticeship of the engineer, for example, acquaints him with the “state of the art” in the field of his own choice. More than this, it becomes evident that the process of training is also one in which considerable interaction can be effected between science and technology. During training and general education each scientist becomes conscious of the ambient states of the technological arts, and each technologist reaches a certain familiarity with the state of the sciences. Though there might well be some additions, greater or smaller, at subsequent dates, the minimum is available for purpose of interchange between science and technology. It can therefore be seen that without evoking the special mechanism of technological fruit on scientific trees, the cumulating bodies of science and of technology are each available quite readily to the other at a distance from their respective research front equal to about one generation of students. That is to say that technology can make ready use of the science that has been learned and packed down in a form suitable for students, and science can use the ambient technology that is readily familiar at a similar stage.
Only very rarely, I think, does a new piece of science give rise directly and quickly to technological repercussions. When it does, the effect is brilliant and startling, and the situation is of considerable historical importance so that the incident becomes glamorized and mythologized. 
10. In the discussion following presentation of this paper it was asked if there was not a strong body of evidence suggesting that the “lag time” between scientific advance and technological innovation had not been progressively reduced throughout the last few centuries so that now there was little lag and therefore strong interaction. In reply, it was again emphasized that the few recent examples used [in these studies tended to be like the cases of transistors and penicillin, grand innovations that break paradigms and happen only in those rare and highly atypical instances that correspond to Copernicus, Einstein, etc. in the history of science. If one excepts these traumatic phenomena and looks instead at the main bulk of “normal” technological advance, it seems doubtful if the lag time has ever differed from that taken to pack down the scientific knowledge from one generation of students to the next. Clearly the incidence of traumatic breaking of the paradigms in science and technology has increased exponentially (though much less rapidly than the growth in crude bulk), but the decrease of lag in normal innovation is most probably mythical.]
HHC: [bracketed] displayed on page 565 of original
There is here, I think, considerable analogy to the breaking of the paradigm a la Kuhn;  the normal progress of advance is one of steady cumulation from within, but the sudden and traumatic changes are of a different nature. It would therefore be just as wrong to see the history of technology as a series of spectacular applications of scientific discovery as it would to see the history of science as a succession of paradigm-breaking events.
At this point one might use the present state of the model to explain to some degree the peculiar difficulty there seems to be in writing the history of technology. There has been a tendency on the one hand to look only at the traumatic changes and see the great inventions which are probably not at all typical of the “normal” growth of technology, just as the Copernican Revolution cannot be counted as part of the normal growth of science. In this connection it is obvious that history written in terms of only such dramatic “Greats” can be, and usually is, highly misleading in spite of its popularity. On the other hand, the history of technology is made especially difficult because the subject matter of its cumulation does not consist of written papers; when such documents exist they are usually epiphenomenal in a way that documents in the history of science are not. There is a tremendous difficulty in translating to written terms the “state of the art” in ancient or in modern times. For this reason, so much of the history of technology necessarily appears to be antiquarian descriptions of artifacts and processes, displayed, as they must be, more like in a museum than as the narrative of conceptual change and cultural evolution that is possible in the history of science. Most certainly the social relations of technology are even greater than those of science, and so some sort of cultural and social history of technology becomes easy; but when one reaches below this to what is now frequently called the internal history, only that of
11. Thomas S. Kuhn. The Structure of Scientific Revolutions (Chicago. 1962).
science is already injected in documentary form in a way that mirrors the content of the science. The similar mirroring process in technology gives rise to the artifacts and processes, and it is necessary to transform this evidence into written form through the medium of descriptions which savor of the antiquarian.
The model of science and technology as a pair of complementary cumulating bodies of research accords rather well with Toynbee’s image of the two dancers. It may be remarked that the dancers are male and female, of sharply opposite attitudes to the production and consumption of literature, of very similar basic motivations in their quests for the private property of discovery. How much truth is there in the notion that the dancers move to the same musical rhythm? If the music be taken to be the needs of society, the wishes of governments, or the desires of men, there is, I feel, very little truth in it. The structure of science at least, as one can analyze it from the connectivity of its papers, seems to show clearly that new knowledge grows out of old at a very steady rate without even very much sensitivity to what one would suppose that societies and men desire. No amount of investment can be guaranteed to produce a cancer cure - not even to give one a 50-50 chance of finding one within ten years if it is not within the state of knowledge to do it. Science seems quite strongly to be its own sweet beast, and to manage it scientists have jealously guarded their right and duty to follow any inquiry wheresoever it might lead, whether or not the results are the ones they thought they wanted. Might it be that the cumulation of a technological art goes by the same sort of process and proves almost as intractable to the will of society and industry? If there is any sort of truth in this, it would make one doubt very heavily the utility of investment of any large sums of money for specific industrial research. It is all very well to pay for the general support of scientists and of technologists provided one is satisfied to take whatever advances happen to be produced, but there seems some inherent difficulty in supposing, according to this model, that anywhere but in very special fields would it pay to subsidize work to a particular technological end.
A particularly gloomy application of this principle is to be seen in an analysis of the biomedical literature Part of it seems to be normal cumulative science, another larger part including clinical papers has all the appearance of technology.  If these are dancers to the same rhythm
12. Later contemplation and discussion with colleagues have convinced me that it may be most fruitful to consider medicine as having scientific and technological components in the light of this discussion of their historiographies. As evidence of the identification of clinical practice with technology it should be noted that the [literature usually does not engage in strong interaction. Furthermore, it is a strong point that there may exist “schools” of clinical practice, as, e.g., those notable in the early history of psychoanalysis. The existence of such schools, as in the subject of philosophy, does not seem to correspond to that of New Invisible Colleges, monitoring a particular sector of research front. Rather, it betokens and corresponds to a set of people with the same fashions and modes, and indeed it may well be an embodiment of the technological state of the art. If this is so, it is an identifying symptom, for whenever we see Invisible Colleges we have research-front science, and whenever we see Schools it is research-front technology. In both cases there is strong communication and social interaction within the group, but only in the case of the scientific groups does the literature behave cumulatively in its strongly knit interaction. Just as in the main body of this investigation, one would suppose that clinical innovation is related to the medical science of the preceding generation, and the advances in medical science tend to grow from the common knowledge of clinical practice learned by the researcher in his medical training. Only in the traumatic and paradigm-breaking instances corresponding to the best known and popularly mythologized great changes in medicine could one find the anomaly of direct interaction between the research fronts of medical science and clinical practice.]
HHC: [bracketed] displayed on page 567 of original
they perhaps do it by the same rather weak and infrequent (though often spectacular) interaction that we have found above. In this case the popular and naïve view of technology as applied science is far too simple and directly misleading. Medical research is heavily supported in the strong belief that scientific advance will produce new cures and wonder-drugs. It is perhaps part of the special pleading that scientists have made, more or less successfully, since the days when Archimedes and Leonardo sought state support for pure scientific research by telling about the technology they could master to produce engines of war and other desirable things. There is a long-standing mystique and mythology which perhaps needs questioning by the deeper analysis of this science-technology relationship. Any excuse that gets support for science and technology is a good thing up to the point (which has now been reached in some countries) where society cannot afford to make greater investment and must then decide what to abandon and what to continue and expand. I am afraid that the naïve image of technology as “applied science” will be difficult to refine and understand in greater depth. But until we know what the rhythm is and how both dancers move to it we shall not have a proper understanding of the history of technology, and until we know that we shall not be able to make intelligent judgments in such critical areas as the support of science and technology and medicine by state and industry.
In summary, therefore, we can say the following:
1. Science has a cumulating, close-knit structure; that is, new knowledge seems to flow from highly related and rather recent pieces of old knowledge, as displayed in the literature.
2. This property is what distinguishes science from technology and from humanistic scholarship.
3. This property accounts for many known social phenomena in science, and also for its surefootedness and high rate of exponential growth.
4. Technology shares with science the same high growth rate, but shows quite complementary social phenomena, particularly in its attitude to the literature.
5. Technology therefore may have a similar, cumulating, close-knit structure to that of science, but of the state of the art, rather than of the literature.
6. Science and technology each therefore have their own separate cumulating structures.
7. Since the structures are separate, only in special and traumatic cases involving the breaking of a paradigm can there be a direct flow from the research front of science to that of technology or vice versa.
8. It is probable that research-front technology is strongly related only to that part of scientific knowledge that has been packed down as part of ambient learning and education, not to research-front science.
9. Similarly, research-front science is related only to the ambient technological knowledge of the previous generation of students, not to the research front of the technological state of the art and its innovations.
10. This reciprocal relation between science and technology, involving the research front of one and the accrued archive of the other, is nevertheless sufficient to keep the two in phase in their separate growths within each otherwise independent cumulation.
11. It is therefore naïve to regard technology as applied science, or clinical practice as applied medical science.
12. Because of this one should beware of any claims that particular scientific research is needed for particular technological potentials, and vice versa. Both cumulations can only be supported for their own separate ends.