Oso 9780198849070 Chapter 11

OUP CORRECTED PROOF – FINAL, 11/12/2019, SPi
Downloaded from https://academic.oup.com/book/40405/chapter/347258580 by National Cheng Kung University Library user on 29 September 2022
PART IV
T H E P U R S U I T O F S C I E N C E BY
O T H E R ME A N S : ‘ A P P L I E D’ A N D
‘POPULAR’ SCIENCE
10
Technology and the Limits of Scientific
Theorizing
In this real world of sweat and dirt, it seems to me that when a view of things
is ‘noble’, that ought to count as a presumption against its truth, and as a
philosophic disqualification. The prince of darkness may be a gentleman, as
we are told he is, but whatever the God of earth and heaven is, he surely can
be no gentleman.
William James, ‘What Pragmatism Means’ (1904)1
The successful aeroplane, like many other pieces of mechanism, is a huge
mass of compromise.
Howard Wright, ‘Aeroplanes from an Engineer’s Point of View’ (1912)2
It has been an unstated premise in the kind of understandings of science that we
encountered in Part III that science is an essentially theoretical enterprise. Scien-
tific theory is taken as a given, complete in itself and autonomous. On this
conception, science, in the form of scientific theory, has two wholly dependent
subsidiary exercises: ‘applied science’, offering applications to practical questions,
and ‘popular science’, offering simplifications for a wider public. We are now
about to see that technology, broadly conceived, cannot in fact be conceived as
‘applied science’, and indeed is not something deriving from science. In some
respects, it is independent of science, in other respects it is an integral and
inseparable part of science, practically and conceptually, but in no respect is it
simply an offshoot of science. In Chapter 11 we shall see that cognate consider-
ations hold for ‘popular science’: in some respects it is autonomous with respect to
theoretical reasoning, whereas in others it forms an integral part of the scientific
enterprise. In both cases, there are deep and complex conceptual issues that bear
directly on the legitimation of the scientific enterprise. Accordingly, they are
relevant to how we are to conceive of the association between science and
civilization. The way in which the values of a ‘pure’ theoretical science bear on
1 William James, ‘What Pragmatism Means: Lecture Two’, in William James, Pragmatism: A New
Name for an Old Way of Thinking (Buffalo, NY, 1991), 34–5.
2 Howard Theophilus Wright, ‘Aeroplanes from an Engineer’s Point of View’, Aero 6 (1912),
374–80: 374.
Civilization and the Culture of Science: Science and the Shaping of Modernity, 1795–1935. Stephen
Gaukroger, Oxford University Press (2020). © Stephen Gaukroger.
DOI: 10.1093/oso/9780198849070.001.0001
292 Technology and the Limits of Scientific Theorizing
the understanding of modern civilization is not, and could not be, the same as the
relation between civilization and a complex—and for all intents and purposes
uncontrollable—mix of technology, scientific theory, and popularization. Yet it is
precisely to this mix that we must look for what drives any sense of what modern
civilization owes to science.
The upshot of this is that it is not possible to consider the issue of the unity of
science from the middle of the nineteenth century without raising the question of
the developing relation between science and technology, or without raising the
question how we assess claims about the civilizing roles of science and technology.
We shall see that we cannot simply collapse technology into science, but nor can
we collapse science into technology. Either of these would efface crucial differ-
ences in modalities. Not only were the claims made for the civilizing effects of
science generally different in kind from those for the civilizing effects of technol-
ogy, for example, but the way in which they were promoted, despite areas of
overlap, differs significantly. Science can be put into opposition with religion, for
example, whereas technology cannot,3 and this becomes a very distinctive feature
of the promotion of science in the nineteenth century. At the same time, we have
to reconsider how metascientific assessments are generated, in particular whether,
rather than scientific achievements, it may (to the extent to which we can separate
these) be that it is technological successes that are shaping them. Certainly, it
would seem in some cases that there is a commitment to values that have come to
be associated with science, such as objectivity, meritocracy, and freedom from
superstition. But while some of these, such as freedom from superstition, may be
the values of science, others, such as objectivity, while they seem like the values of
science turn out on closer investigation to be values of practical and experimental
procedures that can equally be those of engineering and technology.4
At the same time, there can be no doubt that when people considered the
benefits of science in the period with which we are concerned, what they
invariably had in mind were not any identifiable benefits of an increase in
theoretical understanding of the natural world for example, but improvements
in living and working conditions which they associated with technological innov-
ation. Yet what was explicitly promoted was ‘science’, and it was absolutely crucial
that it was science that was given ultimate responsibility for the manifest benefits
that accrued, because it was science, not technology, that was considered to
exercise civilizing effects on the population. It was science, not technology, that
claimed to offer a coherent ordered view of the world and our place in it. But is
this separation between science and technology, with a subsidiary role for the
3 In the mid-nineteenth century, Pope Gregory XVI, as part of his opposition to science and
modernity, had blocked the development of railways and the introduction of gas lighting in the
streets of Rome: see Evans, The Pursuit of Power, 151–2. Here technology is treated as a
manifestation of science, however: as a symbol of science, rather than something in opposition to
religion in its own right.
4 See Lorraine Daston and Peter Galison, Objectivity (New York, 2007).
Technology and the Limits of Scientific Theorizing 293
latter, sustainable? Can we even imagine a world in which science was pursued
completely independently of technology, a world in which there was no ‘applica-
tion’ of science at all, just ‘pure’ science pursued for its own sake? Such a world
would seem to be that envisaged by Sarton when he writes that ‘the chief aim of
scientific research is not to help mankind in the ordinary sense, but to make the
contemplation of truth more easy and more complete’;5 or that of Henry Carhart,
professor of physics at the University of Michigan, when he claimed that ‘the
quality of mind that discovers laws of nature is of a higher order than that which
makes application of them’;6 or that envisaged by Charles Eliot, president of
Harvard University, when he claimed that the goal of science had nothing to do
with its practical applications, but the fact that science ‘enables and purifies the
mind’.7 How, we may ask, would this ‘pure’ science differ essentially from
philosophy?
With the technological transformations of scientific practice from the mid-
nineteenth century onwards, accompanied by technology’s integration into sci-
ence, the latter loses the kind of theoretical purity that many of its advocates
claimed for it. In the wake of the Great War, the idea of the unity of science gets
caught up in deep and intractable questions about the extent to which it is
possible to devise a notion of ‘pure science’ that can be separated from techno-
logical and other issues. The theoretical purity of science was seen, by many
philosophers of science for example, as the timeless essence of science (even if they
would not have put it in quite those terms), and giving it up was met not so much
with resistance as with complete denial: with the result that the legitimation of
science effectively becomes a matter of assimilation to philosophy, as if science
were ultimately a form of conceptual analysis and derivation from basic concep-
tual truths. In this way, perhaps, it could avoid the stigma of being a tool of war,
the image that it took on for many in the wake of the Great War.
Given this context, we might begin to wonder whether, as at first seems to be
the case, the new post-war projects for the unity of science, such as those of the
Logical Positivists, were simply ignoring the developments over the previous
twenty or so years. Perhaps instead they were acutely aware of these developments
(how could they not have been?), and were responding by trying to repurify
science. In either case, the autonomy of science had now come into question in a
way that these projects gave every appearance of refusing to address. Yet the very
nature of science was at issue. As early as 1900, one of the pioneers of ‘technical
mechanics’, August Föppl, was asking whether mathematics was auxiliary (Hilfs-
wissenschaft) to technology or a foundation (Grundwissenschaft) for it, and urging
5 Sarton, The History of Science and the New Humanism, 188.

6 Henry S. Carhart, ‘The Educational and Industrial Value of Science’, Science n.s.1 (12 April
1895), 393–402: 399.
7 Quoted in Kevles, The Physicists, 24.
the former.8 The question became an especially pressing one in the disputes over
aerodynamics in the first half of the century, as we shall see below. More recently,
Peter Janich has summed up what is at stake in his claims that, ‘in place of the
musty ideology of the researcher who unravels nature’s secrets, the physicist will
understand himself to have just one task: enabling technology’,9 and that natural
science ‘is to be understood as a secondary consequence of technology rather than
technology as an application of natural science’.10
It goes without saying that the issues we encounter in exploring the relationship
between science and technology depend on what we assume as the model of
science. I am taking the physical sciences as the model, because this has been very
much a feature of the developments that we have been concerned with. But this
model of scientific understanding is not universal, and the elevation of physics to
this role, especially in the mathematical form that it had taken by the nineteenth
century, is possibly quite unique. Medicine has certainly had a more prominent
role in Islamic and Chinese cultures, for example. I have no doubt that consid-
eration of medicine, whose aim is a practical one, namely that of the prevention
and cure of illnesses, would have given us a far more straightforward route to
understanding the relationship between theoretical, experimental, and techno-
logical practices. But following such a route would not have alerted us to those
peculiarities of science in Western culture, such as the concern with the unity of
science, that have gone to the core of its identity as something that is to be
identified with civilization. It is these features that contribute to making Western
science, as we have come to understand it in the wake of the eighteenth and
nineteenth centuries, such a distinctive form of practice, and it is these, rather
than some generic notion of science (of the kind one finds in theories of ‘scientific
method’ for example), that we need to understand.
In what follows, I want to explore two questions central to understanding the
nature and role of technology in the nineteenth and the early decades of the
twentieth century. First, there is the problem of how technology engages with
science. Here I shall be arguing that, to the extent to which science and technol-
ogy can be integrated, what might once have been thought of as scientific
developments should in fact be conceived in terms of a mixture of theory,
experiment, and theory-free invention.11 This unstable mixture inevitably confers
8 August Föppl, Vorlesungen über technische Mechanik (vol. 1, 2nd edition, Leipzig, 1900), ch. 1.
9 Peter Janich, Zweck und Methode der Physik aus philosophischer Sicht (Konstanz, 1973), 17.
10 Peter Janich, ‘Physics—Natural Science or Technology’, in W. Krohn, E. Layton Jr, and
P. Weingart, eds, The Dynamics of Science and Technology (Dordrecht, 1978), 3–27: 13.
11 On invention, I simply draw attention to the comments of Elmer Sperry, one of the most
successful and prolific inventors of the turn of the century, and the father of cybernetic, or feedback
control, engineering. Distancing himself from any Eureka-style picture of the inventor, Sperry lays
out his completely matter-of-fact approach: ‘I would study the matter over; I would have my
assistants bring before me everything that had been published about it, including the patent
literature dealing with attempts to better the situation. When I had the facts before me, I simply
did the obvious thing. I tried to discern the weakest point and strengthen it; often this involved
Technology and the ‘Unruliness’ of Science 295
on ‘science’ a disunified, unruly, modular character. Second, I want to look at the

different values of science and engineering: at the relative standing of scientists
and engineers in the nineteenth and early twentieth centuries, and differences in
approach to problem solving. In particular, we shall be looking at a case where
science and engineering have a problematic fit: physics and engineering ap-
proaches to aerodynamics in the early decades of the twentieth century. Here
we shall see how the separation of ‘pure’ scientific and ‘practical’ engineering
concerns was unable to stem this unruliness, despite the claims of a rigorous
foundational approach by those in a tradition of mathematical physics. I argue
that the association of science and engineering can be so close that we must take
seriously the non-discursive products of science such as machines. Once we
consider not just the function and construction of machinery, but also the
operation of machinery, we encounter questions very different from those that
concern us in the study of ‘pure’ science, but to which we need to be attentive.
TECHN OLO GY AND THE ‘ UN RU LIN ES S’ OF SCIENCE
I referred earlier to an unruliness about scientific practice, something that is at

odds not only with the epistemological reading of science but with the whole idea
of the unity of science. In this latter respect, it is closely associated with the
modular nature of the sciences, and with a pluralism in the types of explanations
offered, most obviously, but certainly not exclusively, in the life sciences. This
makes it especially important that we probe the sources of this unruliness, and we
can use two issues to focus the questions. The first is the internal structure of
modern science, an internal structure that was formed in the seventeenth century,
when it comprised speculative systematic natural philosophy, experimental nat-
ural philosophy, and quantitative natural philosophy. The second is a nineteenth-
century development, the merging of science and technology. In contrast to the
idea that the connections between science and technology come via ‘applied
science’, I will be arguing that the way to make sense of this merging is by
exploring the connection between a largely autonomous experimental component
in science and practices of ‘invention’ in a broadly technological context. Whereas
the ‘applied science’ approach leaves science untouched by technology, exploring
the connections between experiment and technology allows us to understand how
technology can be integrated into scientific practice in such a way that the latter is
shaped structurally by technology. Moreover, it allows us to do this without
reducing the one to the other, and this is particularly important, as there remains a
degree of incommensurability between the production of knowledge and the
alterations with many ramifications which immediately revealed the scope of the entire project.’
Quoted in Hughes, American Genesis, 20.
production of useful objects and processes, even if there exists a significant grey
area, and perhaps even some overlap, in the middle.
The tripartite composition of science that developed in the seventeenth
century—comprising speculative natural philosophy, experimental natural phil-
osophy, and quantitative programmes (mechanics, astronomy, optics, acous-
tics)—has been discussed in detail in earlier volumes,12 and we touched on it in
Chapter 4, but its development in subsequent centuries needs to be noted briefly
in order to orientate our discussion. Confining our attention to the physical
sciences, because these formed the model of science for almost all the figures
with whom we have been dealing, the first point to note is that the three
components were never fully integrated. Initially, for seventeenth-century mech-
anists, epitomized in Descartes, speculative natural philosophy and a quantitative
programme were considered to go hand in hand. An account of the size, speed,
and direction of motion of the micro-corpuscles constituting material things was
considered to be all that was needed for a full explanation of the physical
behaviour of macroscopic bodies. Experimental natural philosophy, initially in
Boyle’s pneumatics and Newton’s work on the formation of the optical spectrum,
was regarded as falling outside the scope of natural philosophy proper because the
explanations offered were phenomenal and not causal. But in the eighteenth
century, things began to take a new direction as the role of mechanics was
rethought. The success of Newton’s Principia encouraged the claims of rational
mechanics: an axiomatic, analytical form of mechanics, which had aspirations
ultimately to cover the whole of physics, but which did not operate in terms of a
reduction of the macroscopic to the microscopic realm. Micro-corpuscularianism
went rapidly into decline, and with it the kind of speculative natural philosophy
that had provided seventeenth-century natural enquiry with a grand sense of
purpose. Experimental natural philosophy, by contrast, provided a very product-
ive route in physics, chemistry, and the life sciences in the eighteenth century, and
by the beginning of the nineteenth century it was forming a fruitful union with
analytical mechanics, although both retained a significant degree of autonomy.
Systematic natural philosophy, revived briefly in Classical German Idealism, was
largely discredited after the 1830s, but subsequently, in the work of the Neo-
Kantians, as we have seen, the legitimatory programme was taken out of scientific
enquiry as such and pursued at a metalevel, as it was in positivists such as Comte
and Mill.
The numerous later experimental programmes that had evolved from experi-
mental natural philosophy played a central structural role in the functioning of
scientific practice. The crucial point about experimental practice is that it is
inherently piecemeal: it is the epitome of modularity and pluralism. Its central
role in scientific practice reinforces the latter’s ‘unruly’ structure. We have already
12 Gaukroger, The Emergence of a Scientific Culture, chs 8–11; The Collapse of Mechanism, chs
3–5, 8.
looked at the role of autonomous models, in the form of paper-tools, in chemistry.

At the same time, Galison, as we saw earlier, shows how material tools, in the form
of advanced technologies, shape accounts of the behaviour of micro-particles, and
cannot be lifted out of their very different instrumental contexts when we come to
the problem of reconciling these accounts: they remain in place as integral parts of
the respective approaches. And just as Galison provides an illuminating account of
the profound difficulties in distinguishing between science and technology in
modern particle physics, likewise Hugh Aitken, in his study of the radio industry,
shows how, in order to secure unimpeded information flow, engineering scientists
are needed to devise an intermediary language to allow communication between
scientific and technological concerns, so that information—which travels in both
directions—takes a useable form.13 More generally, Ronald Kline has noted that
‘historians have shown that it is often difficult to distinguish between science and
technology in industrial research laboratories; others have described an influence
flowing in the opposite direction—from technology to science—in such areas as
instrumentation, thermodynamics, electromagnetism, and semiconductor the-
ory’.14 Channell sums up the situation in these terms:
As historians began to examine the history of technology they found little evidence for a
strong dependence upon science. A detailed historical analysis of such major technological
inventions as movable type printing, the mechanical clock, guns and gunpowder, metal-
lurgy, the steam engine, textile machines, machine tools, railroad, and the automobile led
to the conclusion that such inventions depended little, if at all, on scientific knowledge,
skill, or craftsmanship. Historians of technology also began to challenge the common
assumption that the Scientific Revolution of the sixteenth and seventeenth centuries has
been primarily responsible for the Industrial Revolution of the eighteenth and nineteenth
centuries. Almost every important technological development that contributed to the
Industrial Revolution, such as Abraham Darby’s production of iron using coke, Richard
Arkwright’s textile machinery and Thomas Newcomen’s steam engine, owed little to any
scientific theory or discovery. Even when some connection between technology and
science could be identified, the connection many times turned out to be either indirect
or much more complex than the applied science model indicated.15
In the light of this, we need to address the second issue, the relation between
experiment and technology: or, more specifically, the relation between scientific
experiment and a category of technology that includes extra-theoretical forms of
practice. Just as we should not think of experiment as simply an adjunct to
scientific theory,16 so we should not consider technology as simply applied science.
13 Hugh G. J. Aitken, Syntony and Spark: The Origins of Radio (New York, 1976), 1–20.
14 Ronald Kline, ‘Construing “Technology” as “Applied Science”: Public Rhetoric of Scientists
and Engineers in the United States, 1880-1945’, Isis 86 (1995), 194–221, 195.
15 David F. Channell, A History of Technoscience: Erasing the Boundaries between Science and
Technology (London, 2017), 10.
16 This is clear as early as Galileo’s treatment of falling bodies: see Gaukroger, The Emergence of a
Scientific Culture, 413–20.
The category of ‘applied science’ is a nineteenth-century invention, and the

distinction between ‘pure’ and ‘applied’ is a nineteenth-century one. The term
first appears in English in Coleridge’s 1817 Treatise on Method, an introduction to
the Encyclopedia Metropolitana.17 As with much in Coleridge, the idea can be
traced back to Kant. His distinction between ‘pure’ and ‘applied’ science is one
between synthetic a priori truths, which Coleridge links with the divine, and
merely contingent truths based on empirical evidence.18 Babbage subsequently
draws on this distinction in his book-length entry on machinery for the Encyclo-
pedia Metropolitana, defining the applied sciences as deriving ‘their facts from
experiment; but the reasonings, on which their chief utility depends, are the
province of what is called abstract science’.19 The idea of ‘applied science’—which
had been in competition with the slightly different ideas of ‘practical science’ and
‘science applied to the arts’ in the first half of the century—had become a standard
one by mid-century.20 Note however that ‘pure’ and ‘applied’ are context-
dependent, and shifting, terms. While mathematics was treated as the model for
physics in the early decades of the nineteenth century for example, optics,
acoustics, and mechanics were deemed ‘mixed mathematics’, by contrast with
‘pure mathematics’, and of decidedly inferior standing as a result. Yet later in the
century, for the defenders of the purity of science, these areas became archetypally
pure mathematical physics, and they were contrasted favourably with applied
science, without any change at all in the type of content of the disciplines.
Robert Bud has distinguished three components in the idea of applied science
in the nineteenth century:
First, it became a category incorporating both machines and a kind of knowledge about
them. Second, it entailed the claim of an historical relationship between carefully nurtured
scientific knowledge and praxis which had already been the source of the nation’s
industrial success. Third, it expressed an expectation about the future, that in years to
come, science would provide a key source of national wealth and prosperity.21
The second and third components, which are very distinctive of the second half of
the nineteenth century, are particularly closely related. The second depends on a
convenient swapping of the aims of science and technology. As Peter Dear puts it:
The authority of science in the modern world rests to a considerable extent on the idea that it
is powerful, that it can do things. Artificial satellites or nuclear explosions can act as icons of
17 Samuel Taylor Coleridge, General Introduction; or, Preliminary Treatise on Method (London,
1817).
18 His ‘applied science’ is a translation of the term ‘angewandte Wissenschaft’, coined by Kantian-
inspired German scientists in the late 1780s and early 1790s. See Robert Bud, ‘ “Applied Science”:
A Phrase in Search of a Meaning’, Isis 103 (2012), 537–45: 538–40.
19 Charles Babbage, On the Economy of Machinery and Manufactures (London, 1832), 379–80.
20 Bud, ‘ “Applied Science”: A Phrase in Search of a Meaning’, 541–5.
21 Robert Bud, ‘ “Applied Science” in Nineteenth-Century Britain: Public Discourse and the
Creation of Meaning, 1817–1876’, History and Technology 30 (2014), 3–36: 4.

science because of the assumption that they legitimately represent what science really is: in
such cases, the instrumentality of science stands for the whole of science. Conversely, when
appeal is made to science as the authority for an account of how some phenomenon or object
really is in nature . . . it then receives back from its presumed instrumental effectiveness an
image of truthfulness that this instrumentality has already been accepted as confirming.22
The conflation of aims of science and technology paves the way for the third
component: the idea that a science-led form of technology is what had delivered
the industrial, modernizing benefits, and that it was in this that the future of
civilization lay. As we have seen, Thomas Huxley was adamant about the civilizing
power of science, its ability to save us from ‘another flood of barbarous hordes’,
and its necessary role in securing ‘physical and moral well-being’. In an 1880
lecture at Mason Science College (the future University of Birmingham) he
pointed out that ‘the practical man . . . may ask what all this talk about culture
has to do with an institution, the object of which is defined to be “to promote the
prosperity of the manufactures and the industry of the country”’. He continues:
I often wish that this phrase, ‘applied science’, had never been invented. For it suggests that
there is a sort of scientific knowledge of direct practical use, which can be studied apart from
another sort of scientific knowledge, which is of no practical utility, and which is termed
‘Pure science’. But there is no more complete fallacy than this. What people call applied
science is nothing but the application of pure science to particular classes of problems.23
This characterization of applied science, which as Gooday has noted became a
touchstone in the ensuing debate about what kinds of science ought to exist and
ought to be recognized as ‘authoritative’,24 was nevertheless at odds with how
many had seen the matter. The Victorians did not necessarily consider applied
science as an applied form of pure science at all: for many it was an entirely
autonomous domain of practical knowledge. For Huxley, by contrast, the purity
of science was central if it was to play the civilizing role that he expected of it. For
him, the crucial distinction was in fact not so much that between science and
technology but that between science and barbarism. But Huxley’s confidence in
the dependence of technology and invention on science was manifestly misplaced.
There are many cases—such as the magnetic compass needle, oxidizing agents,
steam power, or penicillin—where the device or substance appeared before, or
independently of, any scientific theory that might account for it.
22 Peter Dear, ‘What is the History of Science the History Of? Early Modern Roots of the
Ideology of Modern Science, Isis 96 (2005), 390–406: 404.
23 Thomas Huxley, Science and Culture, and Other Essays (London, 1881), 1–23: 19–21. Cf.
Joseph Henry’s comment in his 1850 retirement address to the American Association for the
Advancement of Science: ‘We leave to others with lower aims and different objects to apply our
discoveries to what are called useful purposes.’ Quoted in Nathan Reingold, ‘Joseph Henry on the
Scientific Life: An AAAS Presidential Address of 1850’, in Nathan Reingold, ed., Science, American
Style (New Brunswick, NJ, 1991), 156–68: 159.
24 Graeme Gooday, ‘ “Vague and Artificial”: The Historically Elusive Distinction Between Pure
and Applied Science’, Isis 103 (2012), 546–54: 550.
STEAM FROM
BOILER
K
B F G
D H
C L
E
OVERFL O
WATER
FEED
W
Fig. 10.1. Giffard’s steam injector
A striking nineteenth-century example is the Giffard steam injector (Fig. 10.1),

devised in 1858 by Henri Giffard, an engineer whose main interest was in the
construction of steam-powered dirigibles. Giffard sought a feed apparatus for his
dirigible that would not be subject to friction, by contrast with force pumps,
which were hindered by friction, thereby absorbing power from the engine. It
worked by delivering cold water to a boiler against its own pressure, using the
boiler’s own exhaust steam, and by the early 1860s it had completely replaced
mechanical pumps. From the point of view of physics, however, the device
presented a seemingly intractable problem: the process looked, per impossibile,
to be a case of perpetual motion, and physicists struggled to understand how the
injector worked.25 The task was to reconcile scientific understanding with an
established body of technological knowledge, but it was fifty years before a
satisfactory thermodynamic explanation was offered.26 Giffard’s steam injector
was a case where the technological development preceded the scientific under-
standing. As Mokyr notes, the Industrial Revolution of the first half of the
nineteenth century created a chemical industry without chemistry, an iron
industry without metallurgy, and power machinery without thermodynamics.27
25 See Eda Kranakis, ‘The French Connection: Giffard’s Injector and the Nature of Heat’,
Technology and Culture 23 (1982), 3–38.
26 It came in Henri Poincaré, Thermodynamique: Cours de physique mathématique de la faculté des
sciences de Paris (2nd edition, Paris, 1908), 323–34.
27 Joel Mokyr, ‘The Second Industrial Revolution, 1870-1914’, in V. Castronovo, ed., Storia
dell’Economia Mondiale (Rome, 1999), 219–45 (http://www.faculty.ecn.northwestern.edu/faculty/
mokyr/castronovo.pdf).
A related kind of case is that where there are technological developments which,
even when they did derive from scientific theories, turned out to be independent
of the truth of the theory. A particularly striking example is Hertz’s production of
radio waves on the basis of Maxwell’s electromagnetic theory, a central element in
which was its assumption of an all-pervading aether. Hertz’s account of radio
waves was premised on the existence of an aether, but his subsequent develop-
ment of the radio showed clearly that the production of radio waves was inde-
pendent of this assumption, no matter how fundamental it may have been to the
theory on which Hertz’s discovery was based.28 Moreover, even in cases where the
successful technological development was explicitly based on a successful scientific
one, the sense in which the latter can be said to have given rise to the former is
questionable. Take the case of the steam turbine, a crucial turn-of-the-century
innovation which displaced the wasteful steam engine. Its inventor Charles
Parsons noted ‘that the work was initially commenced because calculation showed
that, from the known data, a successful steam turbine ought to be capable of
construction. The practical development of this engine was thus commenced
chiefly on the basis of the data of physics.’29 The physics here doesn’t tell us how
we might go about constructing a machine of this kind, only that such a
machine—which might or might not be able to be constructed—would not
violate the laws of physics.
In short, technological developments do not just follow on from scientific ones.
At the same time, scientific ones may be dependent on technological ones. In his
1950 autobiography, the physicist Robert Millikan suggests that results in the
nineteenth- and twentieth-century physical sciences derived largely from devel-
opments in engineering, providing some revealing examples:
Historically, the thesis can be maintained that more fundamental advances have been
made as a by-product of instrumental (i.e. engineering) improvement than in the direct
and conscious search for new laws. Witness: (1) relativity and the Michelson-Morley
experiment, the Michelson interferometer came first, not the reverse; (2) the spectroscope,
a new instrument which created spectroscopy; (3) the three-electrode vacuum tube, the
invention of which created a dozen new sciences; (4) the cyclotron, a gadget which with
Lauritsen’s linear accelerator spawned nuclear physics; (5) the Wilson cloud chamber,
the parent of most of our knowledge of cosmic rays; (6) the Rowland work with
gratings, which suggested the Bohr atom; (7) the magnetron, the progenitor of radar;
(8) the counter-tube, the most fertile of all gadgets; (9) the spectroheliograph, the
creator of astrophysics; (10) the relations of Carnot’s reversible engine to the whole of
thermodynamics.30
28 See Jed Z. Buchwald, The Creation of Scientific Effects: Heinrich Hertz and Electric Waves
(Chicago, 1994).
29 Charles Parsons, The Steam Turbine: The Rede Lecture 1911 (Cambridge, 1911), 1.
30 Robert Millikan, The Autobiography of Robert A. Millikan (New York, 1950), 219.
It is clear that in these cases the idea of technology as ‘applied science’ simply gets
things the wrong way round, and there are well-attested earlier examples of this
phenomenon.31
The dependence of science on technology becomes clearer and more all-
pervasive as we move into the era of post-First World War ‘technoscience’,32
but there is no shortage of earlier examples. We saw in Chapter 4 how problems in
machine design could be translated into forms of theoretical enquiry: in compar-
ing Joule’s original practical formulation of problems in thermodynamics with his
later ‘scientific’ formulation, and in comparing Carnot’s practical statement of the
problem of heat engines with the later formulations of Clapeyron and William
Thomson. The Joule and Thomson reformulations are cases in which science
does not generate its own problems but starts from practical engineering and
technological problems and reworks them, giving them a different kind of
standing. This raises the question of the extent to which science generates its
own subject matter. We need to come to terms with the fact that, particularly
from the nineteenth century onwards, there are numerous cases where it is
practical engineering and technological concerns that provide the subject matter
for the physical sciences.33
The key question is not whether technology shapes the tasks of science, for the
answer to that is clear; rather, it is how it shapes these tasks. The problem is not a
new one, and as early as the middle of the nineteenth century there was already
emerging a sophisticated grasp of the ways in which engineering and scientific
considerations could merge. Consider Rankine’s solution to problems in steam
engine design in mid-century, a development that lay at the origins of what came
to be designated engineering science.34 One of the most pressing problems in the
design of high-pressure steam engines centred on the question of the expansion of
31 See, for example, Peter F. Drucker, ‘The Technological Revolution: Notes on the Relationship
of Science, Technology, and Culture’, Technology and Culture 2 (1961), 342–51.
32 The term is, I believe, due to Bruno Latour: see, for example, his Science in Action: How to
Follow Scientists and Engineers through Society (Cambridge, MA, 1987). Latour’s approach to
technoscience, which construes preference for one theory over another in terms of authority and
power in the institutions of science, is not plausible and is not an approach that I shall be following.
On the other hand, he is certainly right to reject the philosophers’ empty claim that ‘truth’ is the aim
of science (on the general question of the aims of science in the formative early modern period see, for
example, Gaukroger, The Emergence of a Scientific Culture, ch. 7). If the choice were simply between
epistemological and sociological conceptions of truth we would indeed be in trouble, and I am
offering something quite different from these here.
33 As Thomson noted in an 1883 talk on all-important electrical units of measurement, ‘resistance
coils and ohms, and standard condensers and microfarads, had been for ten years familiar to the
electricians of the submarine-cable factories and testing-stations, before anything that could be called
electric measurement had come to be regularly practised in almost any of the scientific laboratories of
the world.’ William Thomson, Popular Lectures and Addresses (3 vols, London, 1891–4), i. 82–3.
34 See David F. Channell, ‘The Harmony of Theory and Practice: The Engineering Science of
W. J. M. Rankine’, Technology and Culture 23 (1982), 39–52, whose account I follow in my
discussion here. See also Ben Marsden, ‘Ranking Rankine: W. J. M. Rankine (1820–72) and the
Making of “Engineering Science” Revisited’, History of Science 51 (2013), 434–56.
steam, and there were two opposing views. The ‘expansionists’ advocated cutting
off the supply of steam soon after it was introduced into the cylinder, letting the
expanding steam do work on the piston. In this way, there was an economization
of both steam and fuel. The ‘antiexpansionists’ by contrast questioned these
savings, and advocated introducing steam into the cylinder throughout most of
the stroke. What is of particular interest in the present context is the different
kinds of rationales and defences offered by the two sides. By the 1850s, the use of
a cut-off valve was generally accepted in Britain and in the United States, and the
procedure seemed to be securely based on a basic scientific law, due to Boyle and
Mariotte, that at constant temperature doubling the volume of a gas will be
accompanied by a halving in pressure. But tests first performed in the naval
dockyard in Brooklyn in 1859 cast doubt on the economy of using steam
expansively because the procedure led to condensation, resulting in lower pres-
sure, which neutralized the benefits of early cut-off. The expansionists argued that
if there was a problem, it could hardly be due to the science: the failure must be
owing to improperly constructed engines. It was a matter of making machines
conform to fundamental scientific principles. As one of its advocates wrote: ‘The
nearer we make our practice conform to theory, by providing against the inter-
ference of circumstances unrecognized by our theory, the nearer will our practical
results conform to our theoretical deductions.’35 The antiexpansionists by con-
trast questioned the application of the Boyle–Mariotte law to steam engines. As
the engineer in charge of the Brooklyn experiments put it: ‘The whole theory is
based on a pure and simple abstraction, that of the idea of perfect elasticity in
gases and vapours unaffected by any conditions of matter or of the steam-
engine.’36 For the antiexpansionists the solution lay in relying on the practical
experience of the builders of steam engines who, following rules of thumb and a
long tradition in the design and construction of steam engines, knew what worked
best and what did not work.
The issues underlying the dispute on the use of a cut-off valve ran deep, and
were reflected at an institutional level. When the Regius Chair in Civil Engineer-
ing and Mechanics was instituted at Glasgow in 1840, there was staunch oppos-
ition to the subject from the professors of natural philosophy and mathematics,
who argued that any theoretical questions were exclusively theirs, and the practical
skills associated with engineering and mechanics should be taught outside the
university through the apprenticeship system. The chemistry professor was par-
ticularly obstructive and managed, as a matter of principle, to prevent the teaching
of engineering in any of the university rooms for the whole of the first year. This
was the bifurcation of engineering that Rankine faced when he acceded to the
35 R. H. Thurston, ‘On the Economy Resulting from Expansion of Steam’, Journal of the Franklin
Institute 71 (1861), 193–5. Quoted in Channell, ‘The Harmony of Theory and Practice’, 41.
36 Benjamin Franklin Isherwood, Experimental Researches in Steam Engineering (2 vols,
Philadelphia, 1863), i. 140. Quoted in Channell, ‘The Harmony of Theory and Practice’, 42.
Chair in 1855. He combined serious interests in both science and technology,

having been a natural philosophy student at Edinburgh in the late 1830s, where
he was awarded the medal for his thesis on the relation between heat and light,
and he subsequently spent the 1840s on engineering works: railways, river
improvements, and harbours.37 His aim at Glasgow, as set out in his inaugural
address, ‘Preliminary Dissertation on the Harmony of Theory and Practice in
Mechanics’, was to develop a genuinely integrated form of engineering science,
something that cut across the institutional and professional distinctions.38 Four
years later, in the introduction to his treatise on steam engines and prime movers,
he distinguished between two kinds of technological progress: ‘the empirical and
the scientific. Not the practical and the theoretic, for that distinction is fallacious: all
real progress in mechanical art, whether theoretical or not, must be practical. The
true distinction is this: that the empirical mode of progress is purely and simply
practical; the scientific mode of progress is at once practical and theoretic.’39
‘Empirical’ progress is wholly practical and is characterized by gradual achieve-
ments in materials and workmanship, the result of ‘individual ingenuity in matter
of detail’. It is continuous and cumulative. ‘Scientific’ progress, by contrast, is
discontinuous, and of a different nature:
When the results of experience and observation on the properties of the materials which
are used, and on the laws of the actions which take place, in a class of mechanics, have been
reduced to a science, then the improvement of such machines is no longer confined to
amendments or enlargements in detail of previously existing examples; but from the
principles of science practical rules are deduced, showing not only how to bring the
machine to the condition of greatest efficiency consistent with the available materials
and workmanship, but also how to adapt it to any combination of circumstances, how
different soever from those which have previously occurred.40
Rankine’s framework for engineering science was to analyse a machine into the
essentials of materials and actions. The laws of action would rely more heavily on
formal theoretical conceptions while the properties of materials would rely more
heavily on practical experimental data, and in this way a route was open to
accommodating both theoretical and practical considerations.41 In terms of this
model, what happens in the steam engine is that the combustion of fuel creates
heat, which on being transmitted to the water transforms it into steam, which
creates a state of mechanical energy in the piston, enabling it to perform useful
work.42 As Channell sums it up, ‘the work of a steam engine depended on the
properties of steam and the creation and disappearance of a state of heat in that
37 Channell, ‘The Harmony of Theory and Practice’, 44.

38 William John Macquorn Rankine, A Manual of Applied Mechanics (London, 1858), 1–11.
39 William John Macquorn Rankine, A Manual of the Steam Engine and other Prime Movers
(London, 1859), xix.
40 Ibid., xx. 41 Channell, ‘The Harmony of Theory and Practice’, 46.
42 Rankine, A Manual of the Steam Engine, 223.
steam. Such a system combined theory and practice since the laws of heat relied
more heavily on formal theoretical concepts, while an understanding of the
properties of steam relied on experimental and practical data.’43
It was on this basis that Rankine tackled the controversy over the cut-off in the
steam engine. Using data from the Brooklyn and other trials, he focused on how
the laws of heat affected the properties of steam, showing that, although there is a
degree of condensation of the steam, the reduction in the effective pressure that
resulted was compensated for by the corresponding performance of work. Never-
theless, he argued, the data indicated that during the exhaust stroke the liquefied
steam began to extract heat from the walls of the cylinder, the re-evaporated steam
lowering the temperature of the cylinder, so that there was even more condensa-
tion in the next expansive stroke. Both expansionists and antiexpansionists, while
assigning the blame differently, had accepted that an actual engine does not
function as an ideal engine, but neither had identified the source of the lack of
efficiency. Rankine was able to isolate it and as a result he was able to suggest
remedies, such as showing how jacketing the cylinder would diminish condensa-
tion. Moreover, in his Manual of Applied Mechanics, he uses the resources of
mechanics and technology in a similarly inventive way on questions of structural
engineering, combining an account of the action of forces on a structure as a
whole with the strength of individual material components of the structure, with a
view to establishing the conditions of equilibrium of a structure such as a bridge.44
Rankine’s distinction between ‘empirical’ and ‘scientific’ forms of technological
progress is sharper than it needs to be for his establishment of engineering science
to go through. There are cases of successful inventors, who, like Edison, were
concerned with producing objects, such as an efficient incandescent light bulb,45
and who had very limited success in offering scientific understanding,46 but it is
unhelpful to draw a sharp line between ‘empirical’ and ‘scientific’ technology, as
this suggests a qualitative discontinuity, with everything clearly in one camp or the
other. It should rather, at least provisionally, be seen in terms of a gradual
transition from the one to the other, with inventions such as those of Edison,
while not wholly devoid of a scientific element, being effectively theory-free,
relying crucially on independent developments in devising sources of
43 Channell, ‘The Harmony of Theory and Practice’, 47. 44 Ibid., 49–50.

45 Efficiency was the distinctive factor in Edison’s light bulbs. As Ernest Freeberg notes: ‘Edison
could make no claim that he had invented the first working light bulb—the patent offices and
newspapers provided ample evidence that others had accomplished this feat months and even years
before. What he had done was create a complete lighting system that linked his powerful and more
efficient dynamo, through a central main, feeders, and switches, to an incandescent bulb of superior
design. . . . Alone among these rivals, his bulbs used a filament of high resistance, a crucial innovation
that saved money by using a relatively small amount of current for each lamp.’ The Age of Edison, 43.
46 See W. Bernard Carlson, Innovation as a Social Process: Eliuh Thomson and the Rise of General
Electric, 1870–1900 (Cambridge, 1991); Paul B. Israel, Edison: A Life of Invention (New York,
1998); Ian Wills, ‘Edison and Science: A Curious Result’, Studies in History and Philosophy of Science
40 (2009), 157–66.
electricity—in this case the replacement of batteries with the new powerful type of
dynamo developed in the 1870s—and teams of workers trying out various
possibilities on what was often just a hit-and-miss basis.47 As Thomas Hughes
points out, scientists unfamiliar with inventions and development ‘often deni-
grated this empirical approach, not realizing that to “hunt-and-try” was to
hypothesize and experiment in the absence of theory. Thomas Midgley, the
chemist and inventor responsible for the tetraethyl-lead additive for gasoline,
remarked that the trick was to change a wild-goose chase into a fox hunt.
Thomson, Sperry, and Edison, like other independent inventors, treasured their
model builders, their chemists, their scientists, and their laboratories, because
these facilitated experimentation, the life-blood of invention.’48 At the same time,
the experimental link between ‘science’ and ‘engineering science’ establishes a
continuum between these, a continuum along which technological resources and
highly theoretical ones can travel. It is this interaction of technology/engineering
and experiment that plays such an important role in shaping a scientific theory.
Nevertheless, there is no doubt that talk of a continuum here can only be a first
approximation. It suggests something linear, for example, but there is a case to be
made that a linear continuum fails to capture the complexity of the relations
between theoretical and non-theoretical forms of scientific and technological
activity. Donald Stokes, in a study of Pasteur, has identified the problem with
the linear model as lying in its implication that the closer some activity is to one
end of the model the further it must be from the other. But Pasteur’s work does
not fit this description at all: he aimed at both a new theoretical understanding of
the nature of diseases and at prevention of diseases such as rabies, and the spoilage
of milk and wine. They were explicitly both part of the same exercise.49 Moreover,
Walter Vincenti, in a study of aeronautical engineering in the 1930s, identified a
number of features of engineering practice which can be combined in different
ways. He distinguishes ‘descriptive’ knowledge, that is, factual knowledge; ‘pre-
scriptive knowledge’, knowledge of procedures or operations; and ‘tacit know-
ledge’, which is implicit, wordless, and pictureless. Different types of engineering
activity involve different aspects of engineering knowledge. So, for example, he
argues that learning to design requires descriptive and prescriptive knowledge, but
(somewhat controversially) very little tacit knowledge, whereas learning to prod-
uce requires prescriptive and tacit knowledge but less descriptive knowledge.50 It
47 Edison said that out of a hundred experiments he did not expect more than one to be
successful, and as to that one he was always suspicious until frequent repetition had verified the
original results. Quoted in Frank Lewis Dyer and Thomas C. Martin, Edison, His Life and Inventions
(2 vols, New York, 1910), ii. 612.
48 Hughes, American Genesis, 52.
49 Donald E. Stokes, Pasteur’s Quadrant: Basic Science and Technological Innovation (Washington,
DC, 1997), 12–18.
50 Walter Vincenti, What Engineers Know and How They Know it: Analytical Studies from
Aeronautical History (Baltimore, 1990), 197. For an attempt to identify the defining features of
technology more generally, see John Staudenmaier, Technology’s Storytellers: Reweaving the Human
is unlikely that there is a single model that fully captures the science–technology
nexus in general terms. For our purposes, however, it is not necessary to pursue
the complexities of this question here (we shall turn to some of these below when
we look at aeronautics), but to register the real problems that these complexities
pose, and to appreciate the inadquecy of the idea that science is, for example,
‘knowing that’ whereas technology is ‘knowing how’.
If there is one area that has traditionally been seen to combine ‘knowing that’
and ‘knowing how’, it is experimental practice. It is possible for science to take on
structural features from technology, above all its practical and piecemeal nature,
because experiment, which is where the link with technology lies, already operates
in a piecemeal way: and at the centre of scientific investigation, not at the
periphery. It is this that helps foster an ‘unruliness’ in science, and a pluralism
of explanation, something missed if one takes science to consist essentially in
theories alone.
Note that practical, technological questions have always played a role in
providing science with questions and goals. Ballistics, for example, provided the
problems for, and effectively generated, mechanics in the late sixteenth century.51
But there can be no doubt that the link between science and technology that
comes to the fore in the middle of the nineteenth century means that under-
standing technology begins to have consequences for understanding science in a
way, and on a scale, that it never did earlier. As Hughes has noted, for example,
after 1880 ‘the British Admirality began specifying the performance characteris-
tics it wanted in its engines, guns, ships, and other equipment, and challenged
inventors, engineers, and industrialists to develop those designs. The military also
began to pay at least a part of the costs for testing inventions. Invention of
armaments for the major industrial and military powers then became a “command
economy”, or the supportive direction of innovation by government.’52 And it is
also perhaps worth remembering that from 1905 onwards the Wright brothers
tried relentlessly to interest the US military in aeroplanes.53
These developments were consolidated with a wholly unprecedented govern-
ment investment in technology in the Great War. Although it is true that there
was a significant amount of catching up to do, and that the first three years of
Fabric (Cambridge, MA, 1985); Sven Ove Hansson, ‘What is Technological Science?’, Studies in
History and Philosophy of Science 38 (2007), 523–7; and John D. Anderson, Jr, ‘The Evolution of
Aerodynamics in the Twentieth Century: Engineering or Science?’, in Peter Galison and Alex
Roland, eds, Atmospheric Flight in the Twentieth Century (Dordrecht, 2000), 207–22. See also
Matthew Norton Wise, ‘Mediations: Enlightenment Balancing Acts, or the Technologies of
Rationalism’, in Paul Horwich, ed., World Changes, Thomas Kuhn and the Nature of Science
(Cambridge, MA, 1993), 207–56.
51 See Klein, Nützliches Wissen, who sees the association of science and technology/engineering as
a relatively continuous development since the sixteenth century. By contrast, I am arguing that,
despite its sixteenth-century origins, matters only come to a head in a transformative way in the mid-
nineteenth century.
52 Hughes, American Genesis, 96–7. 53 Ibid., 101–4.
the war were still fought with predominantly agrarian resources,54 there was a
determined move to mobilize science in everything from dyes for uniforms to
armaments and aeroplanes right from the beginning. As the October 1914
editorial in Nature announced: ‘This war, in contradistinction to all previous
wars, is a war in which pure and applied science plays a conspicuous role’.55
State-organized national cartels were established to manage innovation, institu-
tionalizing invention and making mass production the norm.56 As Agar notes,
‘in general, the First World War intensified the organizational revolution—a
gathering concern for scale, organization and efficiency that was discernible in
the mid- to late-nineteenth century’.57 This intensification did not come to an
end in 1918, but was continuously built upon.58 It shaped the subsequent
history of scientific practice, as science was transformed into ‘technoscience’,
characterizations of which differ, but the most appropriate for our purposes is
that offered by Krige and Pestre, in describing the activities of Ernst Lawrence,
founder of the Lawrence Radiation Laboratory at Berkeley: ‘a profound sym-
biosis previously unknown in basic science; a fusion of “pure” science, tech-
nology and engineering. It was the emergence of a new practice, a new way of
doing physics, the emergence of a new kind of researcher at once a physicist, in
touch with the evolution of the discipline and its key theoretical and experi-
mental issues, as conceiver of apparatus and engineer, knowledgeable and
innovative in the most advanced techniques (like electronics at the time) and
able to put them to good use, and entrepreneur, capable of raising large sums of
money, of getting people with different expertise together, of mobilizing
technical resources.’59
54 Compare this to World War II, where, for the first time in history, a substantial rearming of
combatants took place during the time-span of the conflict: see Alex Roland, ‘Science, Technology,
and War’, in Mary Jo Nye, ed., The Cambridge History of Science, volume 5: The Modern Physical and
Mathematical Sciences (Cambridge, 2003), 561–78: 565.
55 Quoted in Agar, Science in the Twentieth Century and Beyond, 93. There were nevertheless
precedents, e.g. in respect to the design and use of rifles, in the American Civil War and in the Boer
War.
56 See William H. McNeill, The Pursuit of Power: Technology, Armed Force, and Society Since A.
D. 1000 (Chicago, 1982), 318.
57 Agar, Science in the Twentieth Century and Beyond, 92.
58 Nevertheless, Vaclac Smil has argued in detail that ‘the fundamental means to realize nearly all
of the 20th-century accomplishments were put in place even before the [twentieth] century began,
mostly during the three closing decades of the 19th century and in the years preceding WWI. That
period ranks as history’s most remarkable discontinuity not only because of the extensive sweep of its
innovations but also because of the rapidity of fundamental advances that were achieved during that
time’: Creating the Twentieth Century, 5–6. Agar offers a similar assessment: Science in the Twentieth
Century and Beyond, 63.
59 John Krige and Dominique Pestre, ‘Some Thoughts on the Early History of CERN’, in Peter
Galison and Bruce Helvy, eds, Big Science: The Growth of Large-Scale Research (Stanford, CA, 1992),
78–99: 93.
The Standing of the Scientist 309

THE S TANDING OF T HE SCIENTIST
There is no more striking symbol of the scientific culture of the twentieth century
than aviation, in both science fiction (see Plate 4) and in the promotion of its
achievements (Plate 5). The development of aerodynamics in the early decades of
the twentieth century presents us with an opportunity to look in more detail at
why and how the different aims and procedures of mathematical physics and
engineering science have clashed. In this section, which focuses on the claims of
mathematical physics, we shall be looking at the ‘why’, in the next, which focuses
on the claims of engineering, on the question of ‘how’.
The ‘why’ is crucial, because we need to understand why technology has not
had the standing, as it were, to lay claim to various developments in which it
played a determining role. Conversely, what is it about the understanding of
science that has required that any such developments must ultimately be counted
as scientific ones? David Channell draws attention to a particularly egregious
example when he writes:
Much of the confusion over what is science and what is technology originated during
World War II. Vannevar Bush, an engineer who directed U.S. wartime research and
headed the Office of Scientific Research and Development, said that when he came to
discover that his British counterparts considered that ‘the engineer was a kind of second-
class citizen compared to the scientist,’ he decided to designate all wartime researchers
working in the Office of Scientific Research and Development as scientists. He noted that
even after World War II the public was led to believe that such an achievement as the
landing of the first astronauts on the moon was a great scientific achievement when in fact
‘it was a marvelously skillful engineering job.’60
The ‘why’ question is a question about the public standing of science, and our first
task must be to understand how this public standing arose and how it was
consolidated: how there arose the deeply entrenched idea of mathematical physics
as the ultimate model of understanding natural processes. The answer lies in
developments in university education in the nineteenth century, particularly as
regards mathematics teaching. The role of mathematics provides a good example
of the problems facing attempts to make science part of a general cultural
programme, and at the same time it throws light on the way in which this cultural
programme offers an image of the persona of the scientist as someone uniquely
equipped to provide a comprehensive understanding of the world.
60 Channell, A History of Technoscience, 1. Cf. a detailed study for the Pentagon of the role of basic
science in the study of weapons development in the 1960s, which concluded that basic science played
‘an inconsequential role’ in the development of major weapons systems: Glen R. Asner, ‘The Linear
Model, The U.S. Department of Defense, and the Golden Age of Industrial Research’, in Karl
Grandin, Nina Wormbs, and Sven Widmalm, eds, The Science–Industry Nexus: History, Policy,
Implications (Sagamore Beach, MA, 2004), 3–30: 3. See also the discussion of industrial science in
Steven Shapin, The Scientific Life: A Moral History of a Late Modern Vocation (Chicago, 2008), chs 4
and 5.
France provides a revealing starting point. In the immediate post-revolutionary

era in France, there was a strong commitment to the unity of science ultimately
founded on mathematics, the pursuit of which was considered to demonstrate a
capacity for reason that was indicative of Enlightenment egalitarian values. The
Écoles Centrales, set up in the wake of the Revolution for the teaching of
mathematics as a core discipline, boasted that ‘a few years will suffice for the
son of an artisan to be more educated [through the study of science and
mathematics] than the aristocrats once were after consuming their entire youth
in numerous [classics] classes’.61 But these institutions had effectively been
abandoned by 1800.62 While the importance of science and mathematics in the
curriculum continued into the empire, mathematical and scientific education
were gradually separated from the humanistic ideals of education.63 Moreover,
the approach to mathematics and the sciences was through the very formal path of
analysis. The exclusion of Republicans such as Carnot and Monge from the
Académie des Sciences meant that the practical programmes involving craftsmen
and practical engineers for example were ignored, while analytical mechanics was
unchallenged.64
Although there continued to be repeated calls for science education to replace
classics at universities,65 the attempt to establish mathematics as the basis for an
education matching that in the humanities encountered numerous problems.
There were limits to how much serious mathematics could be taught at a general
level, and the autonomy of mathematics came to be stressed as a means of
pursuing the discipline in a satisfactory manner. The promotion of their subject
by mathematicians and scientists engendered conflicting interests within the
disciplines, and the teaching of mathematics in universities in the early decades
61 Quoted in L. Pearce Williams, ‘Science, Education and the French Revolution’, Isis 44 (1953),
311–30: 302. See also Ken Alder, ‘French Engineers Become Professionals; or, How Meritocracy
Made Knowledge Objective’, in William Clark, Jan Golinski, and Simon Schaffer, eds, The Sciences
in Enlightened Europe (Chicago, 1999), 94–125. Cf. Huxley, seventy years later: ‘I weigh my words
well when I assert, that the man who should know the true history of the bit of chalk which every
carpenter carries about in his breeches-pocket, though ignorant of all other history, is likely, if he will
think his knowledge out to its ultimate results, to have a truer, and therefore a better, conception of
this wonderful universe, and man's relation to it, than the most learned student who is deep-read in
the records of humanity and ignorant of those of Nature.’ Thomas Huxley, Lay Sermons, Addresses
and Reviews (London, 1877), 176–7.
62 See L. Pearce Williams, ‘Science, Education and Napoleon I’, Isis 47 (1956), 369–82.
63 See John Hubbel Weiss, The Making of Technological Man: Social Origins of French Engineering
Education (Cambridge, MA, 1982).
64 See Lorraine Daston, ‘The Physicalist Tradition in Early Nineteenth Century French
Geometry’, Studies in History and Philosophy of Science 17 (1986), 269–95; and Ivor Grattan-
Guiness, ‘Work for the Workers: Advances in Engineering Mechanics and Instruction in France,
1800–1830’, Annals of Science 41 (1984), 1–33. Note however that Carnot and Monge were key
figures in the establishment of the École Polytechnique, which has remained hugely influential to the
present day.
65 See, for example, in the case of Britain, J. Graham Kerr, ‘Biology and the Training of the
Citizen’, Nature 118 (1926), 102–12.
of the nineteenth century was caught between two different aims. In the case of
Britain, for example, between the 1830s and the 1850s there was a fierce dispute
between Whewell and the mathematician William Hopkins over whether math-
ematics had a purely pedagogical value or whether it was a practically useful and a
morally worthwhile end in itself, revealing the mathematical principles of God’s
creation.66 In the case of France, in the 1820s the mathematical community
struggled hard with the opposing forces of ‘exemplary’ mathematics, where it was
treated as the core of a humanistic culture, and ‘separatist’ mathematics—an
abstract, analytical, autonomous form of advanced mathematics—and in the long
run it was the separatists who prevailed.67 In Germany ‘exemplary’ mathematics
was pursued almost exclusively before the 1820s.68 Mathematics was studied
within the philosophy faculties which, as Joan Richards notes, ‘laid the educa-
tional ground for students interested in going on to the more prestigious faculties
of law, theology or medicine. Interest in the subject reflected this hierarchy
of value. In 1800, the only major German mathematician was Carl Friedrich
Gauss . . . who was a solitary researcher.’69 This changed in the wake of the
teaching and research reforms of the 1820s, and with the institution, at the
University of Königsberg in 1834, of a joint mathematics-physics ‘seminar’ by
Carl Jacobi and Franz Neumann, exemplary mathematics was decisively replaced
by separatist mathematics.70
It would, however, be a mistake to see the shift from exemplary to separatist
mathematics as an abandonment of a cultural role for mathematics. Mathematics
continued to play a significant cultural role: it just shifted from a general cultural
role to an elite one. This is particularly marked in Britain. Richards has drawn
66 Andrew Warwick, Masters of Theory: Cambridge and the Rise of Mathematical Physics (Chicago,
2003), 94. Matters were complicated by other aspects of the question of the point of the exercise:
while Cambridge was able to produce students with a very high level of technical competence, for
example, this was largely a matter of technical skills for their own sake, and there was no discernible
‘research school’ in mathematics in Cambridge between the 1830s and the 1870s, when Maxwell’s
Treatise became the focus of collective research: see ibid., ch. 6.
67 Joan L. Richards, ‘Rigor and Clarity: Foundations of Mathematics in France and England,
1800–1840’, Science in Context 4 (1991), 297–319.
68 See, for example, the defence of the moralizing effect of mathematics in Christian Gottlieb
Ferdinand Engel, Welchen Einfluss äussert das Studium der Mathematischen Wissenschaften auf das
Gemüth? (Berlin, 1820).
69 Joan L. Richards, ‘The Geometrical Tradition: Mathematics, Space, and Reason in the
Nineteenth Century’, in Mary Jo Nye, ed., The Cambridge History of Nineteenth and Twentieth
Century Science (Cambridge, 2003), 449–67: 458.
70 See Gert Schubring, ‘The Conception of Pure Mathematics as an Instrument in the
Professionalisation of Mathematics’, in H. Mehrtens, H. Bos, and T. Schneider, eds, Social History
of Nineteenth Century Mathematics (Basel, 1981), 111–34; Gert Schubring, ‘Die deutsche
mathematische Gemeinde’, in J. Fauvel, R. Flood, and R. Wilson, eds, Möbius und sein Band. Der
Aufstieg von Mathematik und Astronomie im Deutschland des 19. Jahrhunderts (Basle, 1994), 394–406;
Tom Archibald, ‘Images of Mathematics in the German Mathematical Community’, in
U. Bottazzini and A. D. Dalmedico, eds, Changing Images in Mathematics from the French
Revolution to the New Millennium (London, 2001), 49–67.
attention to the way in which mathematics began to replace classics as the vehicle
for the cultivation of minds in Cambridge in the early decades of the century:
In a period of rising enrollments, examinations began to play an increasingly important
role at both Cambridge and Oxford. At Oxford, the examinations were in classics, which
focus was justified as a way of broadening young minds rather than as imparting
specialized knowledge. The same kind of rationale was applied at Cambridge, where,
however, the central examination was in mathematics. Until the middle of the century one
needed to pass on this examination [the ‘Tripos’] in order even to take the parallel
examination in the classics. Even though the Tripos became more and more mathemat-
ically demanding, the justification for requiring that the students study for it continued to
be broadly humanistic rather than specific or professional. Throughout the century the
center of England’s mathematical education pursued the subject as a way to help students
become fully formed human beings.71
The mathematics Tripos had two parts, and it was the first, less advanced part,
that featured as the general curriculum. But what was the ‘mathematics’ of the
first part? The Report of the Examination Board for 1849 sets out the course of
study, which begins with basic geometry, algebra, plane trigonometry, and the
geometry of conic sections. But as the list continues, we are in the realm of
‘applied’ mathematics: ‘the elementary parts of Statics and Dynamics, treated
without the Differential Calculus; the first three sections of Newton, the Proposi-
tions to be proved in Newton’s manner; the elementary parts of Hydrostatics,
without the Differential Calculus; the simpler propositions of Optics, treated
geometrically; the parts of Astronomy required for the explanation of the more
simple phenomena, without calculation’.72 In other words, what is termed
mathematics here is effectively a form of theoretical physics, and as we shall see,
this is how its distinguished practitioners interpreted it.
The regime at Cambridge for those studying the second part of the Tripos—
after 1848 open to those who had performed sufficiently well in the first part—
was highly competitive, and advanced mathematics could not be learned from
formal lectures, oral debate, and reading alone. As Andrew Warwick has pointed
out, the vast majority of students ‘needed an ordered and progressive plan of
study, long periods of private rehearsal of problem solving, and regular face-to-
face interaction with a tutor, someone who was not the professor or the college
tutor, but a private tutor (usually a college fellow) who managed the student’s
studies for a fee’.73 Tutors were crucial, because, as Weintraub notes, it was the
Tripos examination, not those areas of mathematics—such as analysis—which
were important in the eyes of the best European mathematicians, that ‘defined the
concerns of the students and the program. Indeed, the very best Cambridge
71 Richards, ‘Rigor and Clarity’, 307–8.

72 Quoted in Joan L. Richards, Mathematical Visions: The Pursuit of Geometry in Victorian
England (San Diego, 1988), 40–1.
73 Andrew Warwick, Masters of Theory, 37, and, for more detail, 227–85.
mathematicians, men like J. J. Sylvester, and Arthur Cayley, gave lectures that
no students, or very few students, ever attended. Why should they have
attended since that material was never going to appear on any examination?’74
He continues:
Mathematics was thus defined, in England, by a set of tricks and details, based on Newton,
which were linked to applied physics and mechanics, and which could be tested in a time-
limited fashion. The function of the examination really was to provide a fixed ordering of
the degree candidates. The top performer all the way down to the last found his place in
the posted list of results. From Senior Wrangler . . . down to Wooden Spoon (last passing
grade), the order of finish of the Tripos defined one’s options in the world of scholarship at
least.75
Moreover it was a crucial part of the ethos that intellectual endeavour had to be
balanced against physical exercise, and the elite mathematics students were often
elite athletes as well. As Warwick notes, by the 1840s the combined pursuits of
mathematics and sport had become constitutive of the liberal education through
which good undergraduate character was formed,76 and an Indian student at the
time observed with astonishment on arrival at Cambridge that his fellow students
paid ‘as much attention to their bodily as to their mental development’.77 Here
it is not so much that science has been integrated into English culture, but
rather than it has begun to play a direct role in shaping that culture, through
the educational formation of the country’s elite: the names of the top
mathematics students, the ‘wranglers’, for example, were published in the news-
papers, particularly The Times, from the mid-1840s, occasionally accompanied by
short biographies.78
On the question of the scientific shaping of university students, there was a link
between the formation of a well-rounded character for the scientist and the idea
that the scientist was participating in a unified understanding of the world. The
dignity of science and the dignity of its practitioners were intimately tied. Taken
in broad terms, this was not a new requirement. From the last decade of the
74 E. Roy Weintraub, How Economics Became a Mathematical Science (Durham, NC, 2002), 13.
75 Ibid., 14
76 Warwick, Masters of Theory, 179. Note that exercise was largely collective and competitive:
individual exercise was discouraged, except for long walks. There is a contrast with Germany here,
where gymnastics was seen as a complement to intellectual endeavour: see Finkelstein, Emile du Bois-
Reymond, 180–2. Du Bois-Reymond wrote a pamphlet defending gymnastics: Über das Barrenturnen
und über die sogenannte rationelle Gymnastik (Berlin, 1862).
77 Quoted in Warwick, Masters of Theory, 198. The exercise regime was also followed by
mathematics coaches such as Routh, although here it was individual exercise (a two-hour walk
every fine day at exactly the same time in Routh’s case): ibid., 200. Routh was a brilliant applied
mathematician, having pushed Maxwell into second place in the Tripos examination in 1854. His
1877 Adams’ Prize essay, published as A Treatise on the Stability of a Given State of Motion,
Particularly Steady Motion (London, 1877), became fundamental to the understanding in Britain
of aerodynamic stability.
78 Warwick, Masters of Theory, 202.
sixteenth century, a crucial part of Bacon’s project for the reform of natural
philosophy, for example, was a reform of its practitioners. One ingredient in
this was the elaboration of a new image of the natural philosopher, an image that
conveyed the fact that the natural philosopher was no longer an individual seeker
after the arcane mysteries of the natural world, employing an esoteric language
and protecting his discoveries from others, but a public figure in the service of the
public good.79 As we have seen, the need to establish the dignity of the scientific
calling was far from secure in the seventeenth and early eighteenth centuries. By
the nineteenth century, however, in the university environment science was a
beneficiary of a more general rise of the standing of professorships.80 The German
university, the most admired institution of higher education in the Western world
in the nineteenth century,81 provides an instructive example. Leonore O’Boyle,
writing on the idea of ‘knowledge for its own sake’ in German universities, notes
the acceptance of the new academic profession, sparked by a change in the
standing of philology, the field which, due to its radical reform of teaching
methods from lectures to ‘seminars’, was in the vanguard of the changes, particu-
larly affecting science teaching as it was transformed from passive attendance at
lectures to hands-on laboratory training.82 Philology itself was transformed from a
remedial discipline to one that stood at the core of the Philosophy Faculty,
something ‘made evident in the public’s willingness to confer money, autonomy,
and status’.83 How, O’Boyle asks, was this process served by the ideal of know-
ledge for its own sake?:
79 See Gaukroger, Francis Bacon, ch. 2.

80 See, for example, Avraham Zloczower, Career Opportunities and the Growth of Scientific
Discovery in Nineteenth Century Germany, with Special Reference to the Development of Physiology
(New York, 1981).
81 See, for example, Edward Shils and John Roberts, ‘The Diffusion of European Models Outside
Europe’, in Walter Rüegg, ed., A History of the University in Europe, Volume III: Universities in the
Nineteenth and Early Twentieth Centuries (1800–1945) (Cambridge, 2004), 163–230; McClelland,
State, Society and University in Germany; George Weisz, The Emergence of Modern Universities in
France (Princeton, NJ, 1983); George Haines IV, Essays on German Influence upon English Education
and Science, 1850–1919 (Hamden, CT, 1969); Ronald L. Gougher, ‘Comparison of English and
American Views of the German University, 1840-1865: A Bibliography’, History of Education
Quarterly 9 (1969), 477–91.
82 See Bas van Bommel, Classical Humanism and the Challenge of Modernity: Debates on Classical
Education in Nineteenth-Century Germany (Berlin, 2015). Philology seminars were modelled on that
instituted by August Böckh in Berlin in 1812. The statutes declared that the aim of the seminar was
to ‘sustain, propagate, and expand’ the knowledge of antiquity. The natural science institutes usually
started as service facilities for university ‘cabinets’ etc., but became part of the philosophy faculty’s
research objectives when they were transformed into Praktika—internships—where students started
to be actively engaged in enquiry-based learning. The earliest example is Liebig’s laboratory at
Giessen. See Frederick Lawrence Holmes, ‘The Complementarity of Teaching and Research in
Leibig’s Laboratory’, Osiris (1989), 121–64.
83 Leonore O’Boyle, ‘Learning for its Own Sake: The German University as Nineteenth-Century
Model’, Comparative Studies in Society and History 25 (1983), 3–25: 8.

A summary answer is that the ideal helped those who professed it to move from a lowly
class of origin to a position of alliance with a governing class of aristocrats and patricians.
First, all suggestion of association with manual labor and the kinds of material reward
satisfying to lesser men was eliminated. Second, the focus on abstract principles and forms
in the philological disciplines was viewed as a concern with the general rather than the
specific and concrete; all things were presumably seen by the scholars as parts of a whole
rather than as isolated facts of the sort that distracted most men and indeed absorbed their
life.84
Note two things here. The first is the importance of a shift in social status, crucial
for the standing of the professional scientists who had almost completely replaced
disinterested amateur practitioners by this stage. This shift is part of a long-term
development in the persona of the natural philosopher/scientist which begins at
the end of the sixteenth century with the removal of natural philosophy from the
responsibility of clerical Aristotle commentators. There were significant national
variations: in England, although Bacon had set out to establish the dignity of the
collective research that he believed necessary for natural philosophy, he envisaged
a solely supervisory role for a gentleman such as himself in this respect; in the
Florentine court culture in which Galileo worked, the natural philosopher was a
client of a patron and had to transform his social standing through demonstration
of knowledge of the liberal arts for example; for Descartes, the kind of person best
suited to natural philosophy was not someone who had been trained in the
universities but the plain honnête homme, someone who embodied particular
social, intellectual, and moral qualities; and so on.85
Second, note that the criterion for the elevation of the status of the scientist is a
focus on seeing things as parts of a whole. This is reinforced by the idea that the
standing of the scientist lies not just in his pursuing science but also in the general
cultural standing that the persona of the scientist exhibits, and it in turn reinforces
the conception of the persona of the scientist as someone who, through a
commitment to the unity of science, manifests this quality in a paradigmatic
way. It is the unity of science, something which indicates a comprehensive
understanding of the world, that puts science in a controlling position. To this
extent, the unity of science is really not so much about the nature of science as
about the standing of the scientist: with the decline of religion, the scientist is the
only person who can lay claim to a comprehensive understanding of the world.
I want to draw attention to two aspects of these developments. The first is the
way in which the cultural standing of mathematics is able to meet the challenges
of the shift from an ‘exemplary’ to a ‘separatist’ role. Indeed, the transition means
that separatist mathematics has picked up a standing that mathematics did not
have until it first took on an exemplary role, for it then inherited this standing
from the latter. The second is the way in which, once its new separatist standing is
84 Ibid., 8–9.
85 Ch. 6 of Gaukroger, The Emergence of a Scientific Culture is devoted to these questions.
secure, it is immune to changes that would have compromised an exemplary form

of mathematics. In the case of Britain for example, there was a political shift in the
later decades of the nineteenth century that meant that the cultural value of
science was independent of any contest with the humanities. As the century
progressed, what civilization expected from science changed radically. As Frank
Turner notes:
After approximately 1875 the spokesmen for British science shifted their rhetoric and the
emphasis of their policy from the values of peace, cosmopolitanism, self-improvement,
material comfort, social mobility, and intellectual progress toward the values of collectiv-
ism, nationalism, military preparedness, patriotism, political elitism, and social imperial-
ism. Instead of being promoted as an instrument for improving the student morally and
bringing greater physical security or personal benefit to humankind, science came to be
portrayed as a means to create and educate better citizens for state service and stable
politics, and to ensure the military security and economic efficiency of the nation.86
Did this mean that the image of mathematics, and the exact sciences more
generally, shifted to some form of pragmatism? This was the view of some in
engineering, such as Föppl. But the reaction among the Cambridge-trained
mathematicians and scientists, for example, was the exact opposite. What Kevles
has noted about the use of the term ‘pure science’ in the USA in the 1870s,
namely that it ‘was less purity of the subject as purity of motive’,87 captures
perfectly the approach of those trained in the Cambridge Tripos. Moreover, the
purity of the kind of mathematics pursued secured its absolute primacy over
empirical concerns. George Bryan, whose 1911 Stability in Aviation set out the
basic equations governing stability in aircraft,88 and who was one of the most
prominent representatives of the Tripos-educated mathematicians, made it clear
that hydrodynamics in his view consisted in the study of partial differential
equations and not ‘town water supply, resistance of ships, screw propellers and
aeroplanes’, and if the mathematician were to adapt his work to the needs of
engineers he might as well give up mathematics, relying instead ‘on the intro-
duction of constants and coefficients to save him from running his head against
insoluble differential equations’. The hypothetical conditions postulated in the
latter case do not reflect any empirical reality and, he writes, have ‘pretty well
done their duty when they have been made use of to write down differential
equations’.89 As David Bloor notes, for Bryan, ‘it was not the empirical status of
86 Frank M. Turner, ‘Public Science in Britain, 1880-1919’, Isis 71 (1980), 589–608: 592.
87 Daniel J. Kevles, ‘The Physics, Mathematics, and Chemistry Communities: A Comparative
Analysis’, in Alexandra Oleson and John Voss, eds, The Organization of Knowledge in America
(Baltimore, 1979), 139–72: 141.
88 George Bryan, Stability in Aviation: An Introduction to Dynamical Stability as Applied to the
Motion of Aeroplanes (London, 1911).
89 George Bryan, Review of R. de Villamil, ABC of Hydrodynamics, Mathematical Gazette 6
(1912) 379–80: 379.
Science and Machines 317
these conditions (that is, their falsity) that counted, but their power to help the
mathematician frame tractable equations’.90
S C I E N C E A N D MA C H IN E S
Every aeroplane is to be regarded as a collection of unsolved mathematical

problems.
George Henry Bryan, ‘Researches in Aeronautical Mathematics’ (1916)91
When the Wright brothers undertook the first sustained and controlled flights
between 1903 and 1905, they had worked largely by trial and error, although they
did occasionally experiment with rudimentary models. In the wake of their flights,
those constructing aircraft continued in a largely trial-and-error fashion, building
on the practical expertise of their predecessors. But at the same time there began
attempts to develop a theoretical understanding of the action of the air on wings.
The Wright brothers themselves, realizing that the lift data they had from the
German aviation pioneer Otto Lillienthal were unreliable, had experimented with
wind tunnels to produce theories of effective aerofoil shapes, although these were
very crude. The aim of the new theories was to understand lift (the force on the
wing that keeps it in the air), drag (the resistance of the air to motion), and
stability (the ability to correct for pressure producing turning moments that
would cause the wing to pitch).
The basic mathematical resources derived from hydrodynamics, the study of
bodies moving through fluids, were developed in the eighteenth century by
Daniel Bernoulli, d’Alembert, Euler, and Laplace among others. The area was
mathematically challenging, and in order to make it tractable numerous simpli-
fying assumptions had to be made. Euler’s and Laplace’s equations for the motion
of a body through a fluid invoked a ‘perfect fluid’, and they provided a rigorous
account of how a fluid would flow around a body if the body and the fluid were
moving with respect to one another (it doesn’t matter whether it is the body or the
fluid that is moving, the effects are demonstrably identical). But perfect fluids are
incompressible and inviscid (they have a viscosity of zero) and the flow exercises
no resultant force on the body. The problem was that, despite the fact that it was
well-developed mathematically, ideal-fluid theory was considered useless for the
study of real fluids such as air: what was needed instead was an account of
viscosity. Such an account had been provided by the Cambridge mathematician
George Stokes mid-century in a series of equations called the Navier–Stokes
90 David Bloor, The Enigma of the Aerofoil: Rival Theories on Aerodynamics, 1909–1930 (Chicago,
2011), 95.
91 George Henry Bryan, ‘Researches in Aeronautical Mathematics’, Nature 96 (1916), 509–11:
510.
equations (named after their co-discoverers). They extend Euler’s equations into
the area of viscous fluids, by applying Newton’s second law to fluid motion. But
the Navier–Stokes equations could only be solved in a few simple cases (it is not
clear even now that, outside a handful of cases, solutions in three dimensions
exist), so their application to the forces acting on wings was very limited.
In light of this, the task was to find a way of making ideal-fluid theory more
realistic, and there were two basic approaches to this. The problem with perfect
fluids arises from the fact that they are continuous and irrotational (they do not
rotate around the body immersed in them). But as Bryan pointed out: ‘As soon as
you give up the equations of a medium derived from the assumed definitions of an
ideal fluid, you obtain equations of which integrals cannot be found; at least
mathematicians have tried over and over in vain to find them.’92 Two different
approaches attempted to solve the problem by introducing discontinuities in the
one case, and circulating vortices around the moving body in the other. The
approaches were associated with very different conceptions of what the under-
standing of physical phenomena consisted in. The first one maintained that any
account must be anchored in—and ultimately be deducible from—mathematical
physics, particularly as conceived in the Cambridge Tripos tradition, for as
Groenewegen notes, the appreciation of mathematical knowledge as necessary
and inevitable truth, derived axiomatically, ‘was an aspect of Cambridge math-
ematical training which justified its pre-eminence in the university honours
syllabus, combined as it was with methods by which such truths could be
mastered. This was a point stressed by Whewell in his defence of the value of
mathematical specialization. A high wrangler in particular would have been
heavily imbued by this specialized feature of mathematical knowledge.’93
The alternative approach, which was in the tradition of ‘practical mechanics’,
particularly as conceived in the tradition of the Technische Hochschule, rejected
such foundational aspirations, and manipulated mathematical and theoretical
resources in such a way as to achieve a particular engineering result. The situation
becomes especially interesting when we consider the resistance that advocates of
the first approach demonstrated to the success of the second. Such resistance has
complex motivations, but I want to argue that one crucial motivation derived
from the association of those in the Cambridge Tripos tradition with a model of
science as something comprehensive and certain, which is accompanied by a
conception of the persona of the scientist—in the form of the mathematical
physicist—as the ultimate arbiter of what counts as a satisfactory outcome of
scientific enquiry. To a large extent, the resistance arises from the fear that this
conception would be compromised by abandoning the idea that the physical
nature of the world can ultimately be derived from a unified set of fundamental,
mathematically formulated laws.
92 Bryan, Review of R. de Villamil, 379.

93 Peter Groenewegen, A Soaring Eagle: Alfred Marshall 1842–1924 (Aldershot, 1995), 116.
Consider first the theory of discontinuous flow. This was originally formulated
by Helmholtz, and subsequently developed by Kirchhoff and Rayleigh.94 Bloor
summarizes the theory well:
Helmholtz argued that the result of zero resistance arose because an ideal fluid could wrap
itself around an object and exert pressure from all sides in a way that cancelled out any
resultant force. The discontinuous flow approach exploited the possibility that there could
be discontinuities in the velocity of different bodies of ideal fluid that were in direct
contact with one another. The flow was assumed to break away from the edges of an
obstacle and create a wake behind it. The wake would be ‘dead water’ or ‘dead air’, and the
main body of ideal fluid would flow past it. Such a flow pattern in an ideal fluid, with a
wake of dead fluid, turned out to be compatible with the Euler equations. Furthermore, it
could be established that, given such a discontinuous flow, the pressure on the front face of
an object would be greater than the pressure of the dead fluid on the rear. . . . If the
resultant force proved large enough, here was a theory that could, in principle, explain the
lift of a wing as well as the resistance to motion, the ‘drag’.95
However, by the second decade of the twentieth century, the prospects of the
discontinuity theory were looking dire. Aerodynamic researchers identified a
number of fundamental problems. It was a crucial assumption, for example,
that the pressure in the dead-water region was the same as that of the undisturbed
flow, but measurements showed it to be less than the latter. The young Cam-
bridge mathematician Geoffrey Ingram Taylor noted in his 1914 PhD thesis that
the ability to generate rigorous mathematical solutions in hydrodynamics was in
inverse proportion to their experimental support, singling out the discontinuity
theory, which he argued must now be relegated to the realm of mathematical
curiosities.96 At stake here was an issue that those involved in the practical aspects
of aeroplane design felt strongly about. In a short 1916 piece, J. H. Ledeboer, the
editor of Aeronautics, wrote:
I don’t deny the infinitely valuable role of pure science, still less that of theory, but science
should have some relation to practice, since it is its foster-mother. There is more than one
aeroplane designer who knows just enough mathematics to make twice two work out at
four, but he will turn out machines equal in performance to the best. We in this country
94 Hermann von Helmholtz, ‘Über discontinuirliche Flüssigkeits-Bewegungen’, Monatsbericht

der königlich preussischen Akademie des Wissenschaften zu Berlin (1868), 215–28. Gustav Kirchhoff,
‘Zur Theorie freier Flüssigkeitsstrahlen’, Crelle’s Journal für reine und angewandte Mathematik 70
(1869), 289–98. Lord Rayleigh, ‘On the Resistance of Fluids’, Philosophical Magazine 2 (1876),
430–41; Lord Rayleigh, ‘On the Irregular Flight of a Tennis Ball’, Messenger of Mathematics 7 (1877),
14–16.
95 Bloor, The Enigma of the Aerofoil, 77–8. I am indebted to Bloor’s exemplary account in what
follows.
96 See ibid., 97. Taylor’s thesis remained unpublished but he did draw on it for two subsequent
1916 reports: ‘Pressure Distribution Around a Cylinder’, and ‘Pressure Distribution over the Wings
of an Aeroplane in Flight’, published in Reports and Memoranda of the Advisory Committee for
Aeronautics 191 and 287 respectively.

know, as they do in the United States, of eminent designers who see a new type of machine
rather than design it.97
The point is made in a less diplomatic way a year later by the founding editor of
The Aeroplane, Charles Grey, who writes that much harm had been done ‘both to
the development of aeroplanes and to the good repute of genuine aeroplane
designers by people who pose as “aeronautical experts” on the strength of being
able to turn out strings of incomprehensible calculations resulting from empirical
formulae based on debatable figures acquired from inconclusive experiments
carried out by persons of doubtful reliability on instruments of problematic
accuracy’.98 Grey had no doubt that we should trust not the scientist but ‘the
man who guesses, and guesses right’.99 In a 1922 short article on the development
of aviation, W. H. Sayers, author of work on stability in aeroplane design, noted
that developments before the Great War were the result of individual adventure,
and ‘individual designers worked, as artists worked, by a sort of inspiration as to
what an aeroplane ought to be like, and built as nearly to their inspiration as the
limited means, appliances and increasing knowledge they possessed would allow
them’.100 Brooke Hindle, in a detailed account of how invention typically
proceeds by emulating established mechanical devices for new purposes, has
described how visual imagination is central to invention, noting that many
inventors were also artists.101 As he remarks: ‘Designing a machine requires
good visual or spatial thinking. It requires mental arrangement, rearrangement,
and manipulation of projected components and devices. It usually requires a trial
construction of the machine, or at least a model of it, and then more mental
manipulation of possible changes in order to bring it to an effective working
condition.’102
The alternative to the discontinuity theory, which had now effectively become
defunct, was circulation theory, which was more in tune with the concerns of
engineers. The basic idea behind circulation theory is this.103 It was known from
Bernoulli onwards that, if the air behaves like an ideal fluid, then the faster the air
flowing over a wing the less the pressure it exerts, and the slower the flow the
greater the pressure. So for lift, the pressure of the moving air on the upper surface
of the wing, which pushes downward, must be less than the pressure on the lower
surface, which pushes upwards. It is this excess pressure that constitutes lift. If the
97 John Henry Ledeboer, ‘The Function of Literature’, Aeronautics 11 (1916), 33.

98 Charles Grey, Preface to Frank Barnwell, Aeroplane Design (London, 1917), 4.
99 Charles Grey, ‘Editorial Comment’, Aeroplane 12 (1917), 1284.
100 W. H. Sayers, ‘The Arrest of Aerodynamic Development’, Aeroplane 22 (1922) 138.
101 Brooke Hindle, Emulation and Invention (New York, 1981).
102 Brooke Hindle and Steven Lubar, Engineers of Change: The American Industrial Revolution
1790–1860 (Washington, DC, 1986), 75.
103 I draw here extensively on Bloor, Enigma of the Aerofoil, ch. 4. See also the collection of
seminal papers collected by R. T. Jones, Classical Aerodynamic Theory, NASA Reference Publication
1050, 1979.
airflow around the wing follows the surface of the wing—i.e. is not
discontinuous—then what needs to be explained is why the air immediately
above the wing of an airborne aeroplane is moving more quickly than that
immediately below the wing.
The circulation theory proceeds by distinguishing two distinct types of flow: a
steady flow of air with constant speed and direction, and a swirling vortex that
rotates around a central point, and whose speed drops off uniformly with distance
from the centre. These are superimposed on one another (by a simple process of
vector addition). Imagine—in two dimensions—the steady flow moving from left
to right, and the vortex positioned to the right of it, the parts of the vortex moving
in a clockwise direction. The motion of those parts of the vortex moving in the
opposite direction to the flow will, on encountering the flow, be retarded because
we will be subtracting one motion from the other, whereas the speed of those
moving in the same direction will be increased, because we will be adding the
motions. In the case envisaged, interaction of the airflow with the clockwise
rotation would cause an increase in its speed over the wing and a decrease
under the wing. Accordingly, if the wing generates a vortex, this would explain
its lift.
The challenge for circulatory theories was to account for the origins of the
assumed vortical motion. The first version of the theory was set out in Frederick
Lanchester’s 1907 Aerodynamics.104 The argument starts with the observation that
birds’ wings, which have evolved into a shape that conforms to the pattern of
airflow necessary for lift, have an arched profile with a slight downward inclin-
ation at the front edge. What must happen, Lanchester argues, is that the air must
be moving upward as it approaches the leading edge of the wing and downward as
it leaves the trailing edge. There is an exchange of momentum: the initial upward
vertical component of the motion must be reduced to zero as the air passes over
the wing, and then replaced with a downward vertical component.
Critics of the theory such as Taylor argued that it was lacking in physical
content: it provided no mechanism by which circulation around the body could
be created. Moreover, Taylor argued, Lanchester was assuming a perfect fluid, but
in a stationary perfect fluid setting a body in motion could not create such a flow.
As Bloor points out, Taylor’s reaction to Lanchester assimilated Lanchester’s
analysis to the classical framework of perfect fluid theory, that is, to the equations
of Euler and Laplace’s equation. But Lanchester was well aware that rotation in an
ideal fluid can neither be created nor destroyed: ‘As an engineer working with real
fluids, such as air and water, he hardly expected mathematical idealizations to be
accurate. The important thing was to learn what one could from the idealized case
but not be imposed upon by it.’105
104 Frederick Lanchester, Aerodynamics, constituting the First Volume of a Complete Work on Aerial
Flight (London, 1907).
105 Bloor, Enigma of the Aerofoil, 147.
One of the main complaints by critics of Lanchester was the narrow scope of his
account. He focused on small angles of incidence, and defended the limited scope
of his investigation, whereas in the Tripos tradition which his critics assumed, one
might start with a single case but the point was then to generalize to all others.
This was what genuine knowledge enabled one to do, and indeed this was the
whole point of scientific investigation: one sought general explanations, not
limited particular ones. There is nothing new in this, and the criticism largely
mirrored criticisms of those working in experimental natural philosophy (in optics
and pneumatics) in the seventeenth century. The general mathematics of flow
accorded no special significance to the case of small angles of incidence that
Lanchester examined. Accordingly, one objection that was made, for example by
Cowley and Levy in their Aeronautics in Theory and Experiment,106 was that
Lanchester’s choice was arbitrary, where the criterion whether something is
arbitrary or not is a whether it is deducible from the basic equations of fluid
dynamics. But Lanchester is not working to the criteria of fluid dynamics, which
would have required a sharp separation between inviscid fluids (i.e. the perfect
fluids of basic fluid mechanics) and viscous fluids. He includes both viscous and
inviscid fluids in his account. This is not a lack of awareness of the standard
mathematical procedures of aerodynamics on his part. Rather, it is the result of a
flexibility in the handling of the mathematics.
In a memorandum written in 1936, well after the circulation theory had been
accepted as the correct account of lift and drag, Lanchester wrote that the
National Physical Laboratory had not taken his work seriously twenty years earlier
because ‘my methods did not appeal to them in view of their training. They
mostly belonged to the Cambridge School, whereas I was a product of the Royal
College of Science.’107 What lies behind this difference is a deep divide between
the view that the behaviour of aircraft wings is fundamentally a matter of a
mathematically elaborated aerodynamics, and, by contrast, the bringing together
of several different kinds of approach which may retain some degree of autonomy,
but which can be manipulated with a view to providing a solution that makes
sense in engineering terms.108 The issues are best brought out by considering the
way in which the German scientists/engineers of the Technische Hochschule dealt
with the questions, and by focusing on the question of what the relation is
between what the textbooks issuing from the Cambridge school had distinguished
as ideal and real fluids.109
106 William Lewis Cowley and Hyman Levy, Aeronautics in Theory and Experiment (London,
1918).
107 Unpublished memorandum, quoted in Bloor, Enigma of the Aerofoil, 195. The Royal College
subsequently became part of Imperial College.
108 In this respect respect there are parallels with the establishment of atomism, which we looked
at earlier, in Chapter 5.
109 See, for example, Horace Lamb, A Treatise on the Mathematical Theory of the Motion of Fluids
(Cambridge, 1879), and Horace Lamb, Hydrodynamics (5th edition, Cambridge, 1924).
Lanchester’s circulation theory had been received enthusiastically right from

the beginning by Ludwig Prandtl, who had been appointed professor of fluid
mechanics in the Hannover Technische Hochschule in 1901. The Technische
Hochschulen had been founded as what were in effect technical or engineering
universities, and their future as research institutions was secured with the granting
of the right to issue doctoral degrees in 1899.110 Prandtl entered the Technische
Hochschule Munich in 1894 and graduated with a doctorate six years later. He held
seminars on Lanchester’s Aerodynamics as soon as it appeared, immediately over-
seeing its German translation, and it was he who developed circulation theory in a
particularly compelling way. Prandtl’s student, Georg Fuhrmann, had carried out
a number of experiments with model airships in a wind tunnel in 1910 to
examine the predictions of ideal-fluid theory.111 He distinguished two forms of
drag: pressure drag, which acts normal to the surface, and is the kind of thing that
can be caused by a perfect fluid; and friction drag, which acts at a tangent to the
surface, and is caused by viscosity. He showed that the graph of the observed
pressure distribution of the air flow was very close to that predicted by ideal-fluid
theory except at the very tail of the airship models. Second, models with a
rounded nose, slender body, and long tapered tail had extremely low resistance.
Third, nearly all of the small residual drag could be accounted for by the frictional
drag of the air on the surface. Even given the slight deviation at the tail and the
effect of the air in immediate contact with the surface of the airship, what was
surprising was the extent to which the air behaved like an ideal fluid.
In a report that he wrote up in 1923, Prandtl set out a survey of ideal-fluid
theory followed by details of Fuhrmann’s experiments, concluding that: ‘The
theoretical theorem that in the ideal fluid the resistance is zero, receives in this a
brilliant confirmation by experiment.’112 But as Bloor points out, for Prandtl’s
British counterparts, a ‘theoretical theorem’ could only be the result of deduction
from the premises of the theory, and as such it would be something to be judged
by logical, not experimental, criteria: it did not describe a real fluid but only an
ideal one. Accordingly, Prandtl was claiming that the experiment provided
‘brilliant confirmation’ of what was essentially a mathematical theorem, which
his critics considered ridiculous. By contrast, as Bloor notes, Prandtl and Fuhr-
mann ‘found they could use the theory of ideal fluids to design airships that were
very close approximations to the zero resistance entailed by the theory of perfect
fluids. It helped them to identify the places where smooth flow was breaking
down so that they could reduce it further. Their efforts were informed by an ideal
110 On the development of technical education in Germany in the second half of the nineteenth
century, see Kees Gispen, New Profession, Old Order: Engineers and German Society, 1815–1914
(Cambridge, 1989), chs 3 and 4.
111 Georg Fuhrmann, ‘Theoretische und experimentelle Untersuchungen an Ballonmodellen’,
Jahrbuch der Motorluftschiff-Studiengesellechaft 5 (1911–12), 64–123.
112 Ludwig Prandtl, ‘Applications of Modern Hydrodynamics to Aeronautics’, National Advisory
Committee for Aeronautics, Report No. 116 (1923), 174.
that they were striving to attain. The ideal was not kept distinct from practice, or
set in opposition to it, but was integral to it and gave practice its direction and
purpose.’113
One thing that distinguishes the Tripos tradition of aerodynamics from that of
the German engineers was set out clearly by Prandtl’s thesis supervisor, August
Föppl. The engineer, Föppl argued, needs to get answers, and this may mean
using ideas that cannot be justified in terms of established physical knowledge.
This is not unlike the difference I alluded to earlier, between the inventor and the
scientist. For the inventor, it is an all-or-nothing exercise: one either produces the
working object or one does not. Similarly the constraints on the engineer and the
mathematician differ significantly. On Föppl’s account, engineers need to develop
their own concepts, which do not meet the logical demands of those forged in
established theory. While the scientist can wait for inspiration, he writes, ‘the
engineer, by contrast, is subject to the force of necessity. He must, without delay,
deal with the matter when some phenomenon interferes and interposes itself in
his path. He must, in some way or another, arrive at a theoretical understanding of
it as best he can.’114 Accordingly, the engineer has a freedom to create new
concepts that the scientist or mathematician does not have. Föppl doesn’t deny
that such concepts and procedures may be absorbed back into the main body of
knowledge at some stage, but, by contrast with scientific concepts, this is not what
their legitimacy consists in. None of the engineers directly asserted the literal truth
of ideal-fluid theory, but they weren’t worried about the fact that it was literally
false because that didn’t mean that one couldn’t work with it.
It is in a 1923 paper by Hermann Glauert that we find the clearest statement of
a rationale for the attempts of Prandtl and others to combine theoretical hydro-
dynamics and aeronautical engineering.115 Glauert identifies three steps in the
solution of physical problems in aerodynamics. The first is the setting out of
assumptions about what quantities—such as gravity, compressibility, and
viscosity—can be neglected in dealing with the questions towards which the
enquiry is directed. The second is the expression of the physical system in a
mathematical form, that is, primarily, in terms of differential equations and
boundary conditions. Third, the equations and the terms they contain must be
manipulated until they yield numerical results that can be tested experimentally or
used for some practical purpose. This may sound comparatively anodyne, but

114 Föppl, Vorlesungen über technische Mechanik, 11–12. See the discussion in Bloor, Enigma of the
Aerofoil, 187–94.
115 Hermann Glauert, ‘Theoretical Relationships for the Lift and Drag of an Aerofoil Structure’,
Journal of the Royal Aeronautical Society 27 (1923), 512–18. This absolutely central paper has
ramifications for all design professions, including architecture for example. Glauert came from an
English germanophone family, and had gone through the Cambridge Tripos examinations (where he
took a first class honours with distinction), but always had extensive connections with German
engineers.
when Glauert comes to discuss the relationship between the first and the third
stages, he captures something distinctive about the engineering approach. Because
the mathematical problems may be insurmountable in the third stage, the
engineer needs to return to the initial physical assumptions, simplifying them
further or restricting attention to a smaller range of cases. There are, he notes, no
absolutely rigid assumptions: any assumptions can be reworked to enable a
solution to be yielded. In this way, the first and third steps become non-
sequential, and this is the distinctive feature of the engineering approach, by
contrast with the mathematical hydrodynamics approach. Problems encountered
in the third step constantly feed back into the first step in an effort to make the
problem tractable, and there is a constant movement between the first and third
steps in a process of repeated problem redefinition. Applying this approach to the
circulation theory, Glauert argues that while viscosity cannot be wholly neglected,
this does not preclude the use of perfect fluid theory. We cannot ignore two facts
about viscosity: the ‘no-slip’ condition, that is, at a solid boundary, a viscous fluid
will have zero velocity relative to the boundary, by contrast with the behaviour of
a perfect fluid, which has a finite slip; and second, the fact that viscous forces,
which are proportional to the rate of change of velocity, are significant when close
to the body but negligible at large distance. These are physical facts for which
approximations must be found, but this cannot be done at Stage 1: the approxi-
mation is introduced after the fact, as it were, in Stage 3.116
For Glauert, under the right conditions, the equations of inviscid flow are
legitimate approximations to the viscous equations. The mathematical hydro-
dynamics approach had separated out the two completely, arguing that ideal-flow
theory was false and therefore could not be deployed in aerodynamics. But
Glauert is arguing that there is a relationship between the two, because inviscid
flow must be understood as a limiting case of viscous flow:
It is known that the solution obtained by ignoring the viscosity is unsatisfactory, but it is
by no means obvious that the limiting solution obtained as the viscosity tends to zero is the
same as the solution for zero viscosity. In particular, in the case of a body with a sharp edge,
there is a region where the velocity gradient tends to infinity, and where the viscous forces
will be of the same order of magnitude as the dynamic forces, however small the viscosity.
On the other hand, the layer around the body in which viscosity is of importance can be
conceived of as zero thickness in the limit, and this conception is equivalent to allowing
slip on the surface of the body. It appears, therefore, that the non-viscous equations will be
the same as the limit of the viscous equations, except in the region of sharp edges.117
In this context, consider a 1927 paper by Prandtl in which he asks how circulation
arises (i.e. how an aircraft gets off the ground), and why perfect fluid theory,
116 As noted in Bloor, Enigma of the Aerofoil, 329.

117 Glauert, ‘Theoretical Relationships’, 514.
despite being strictly false, provides the solution.118 On the first question, he
argued that both his proposed circulatory flow and the irrotational flow of the
mathematicians were consistent with the classical theorems of hydrodynamics.
But only flow with circulation and the smooth confluence at the trailing edge is
physically realizable. As Bloor notes, even if the process by which circulation was
generated remained obscure, it was the logical possibility of circulation, and the
logical right to postulate it, that really mattered for the perfect fluid approach.119
On the second question, Prandtl deploys his theory of the ‘boundary layer’,
first developed in a paper published in 1905.120 This paper is a classic instance of
modelling achieving something that theorizing could not have done.121 The
problem to which it was directed was a discrepancy between the mathematical
theory of fluid flow and experimental results: the former could give no account of
why the very small frictional forces present in the flow of air or water around a
body were able to create a no-slip condition at the solid boundary. To investigate
this question, Prandtl built a small water tunnel in which fluid flowed past
different kinds of bodies (Fig. 10.2). By these means, he was able to highlight
different kinds of flows in regions of the fluid close to the body and in those more
remote from it. He showed that in certain areas, frictional forces were insignifi-
cant, and in those where they were significant, approximations to the Navier–
Stokes equations meant that they could be applied to the boundary layer. In this
way he was able to provide a bridge that connected the behaviour of real viscous
fluids and the unreal inviscid fluid of ideal-fluid theory. He showed how the field
of flow could be divided into two areas separated by a boundary: a viscous flow
inside the boundary layer, which was responsible for the drag experienced by the
body; and one outside the boundary where viscosity was negligible, so that its
behaviour was very close to that of an ideal fluid.
In a 1927 paper, he applies the boundary layer theory directly to the question
of the success of perfect fluid theory in explaining lift. He notes that the boundary
layer is the cause of the formation of vortices. The equations of classical hydro-
dynamics only apply to viscid flows, with the result that, in motion starting from
rest, circulation will be zero (or close enough to zero) in fluid circuits that do not
pass through the boundary layer. It is because of this that real fluids behave like
perfect fluids at a distance from the solid body. But the boundary layer can be
manipulated, as Prandtl demonstrated with photographs and films of experiments
where it had been removed by suction, and this had a dramatic effect on the flow.
This is Prandtl’s justification for using perfect fluid theory. As he puts it: ‘We thus
get the unique characteristic that it is precisely these turbulent flows of low
118 Ludwig Prandtl, ‘The Generation of Vortices in Fluids of Small Viscosity’, Journal of the Royal
Aeronautical Society 31 (1927), 720–41.
120 Ludwig Prandtl, ‘Über Flüssigkeitbewegung bei sehr kleiner Reibung, in A. Krazer, ed.,
Verhandlungen des dritten Internationalen Mathematiker-Kongresses (Leipzig, 1905).
121 See the discussion in Morrison, ‘Models as Autonomous Agents’.
Fig. 10.2. Prandtl’s hydraulic tunnel (1904)
resistance around bodies which can be so closely represented by the theory of the
perfect liquid’.122
What we have in these debates in aerodynamics is a conflict between the
procedures of mathematical physics, which in the Cambridge Tripos system for
example are treated as models of physical explanation, and those of engineers
whose procedures were taken by the Cambridge mathematicians as being at best
pragmatic, but which I hope will now be clear should be seen rather as a different
way of working, one which involves a degree of freedom or flexibility that is
inappropriate in the mathematical tradition. The flexibility can involve mixing
mathematical and experimental results, and regularly, and as a matter of course,
going back to revisit premises when arguments based on these premises fail to
deliver. Such constant revision and reworking is not compatible with the idea of
deductively watertight inference that regulated the approach of the Cambridge
mathematicians.
The issues that arise in such a stark form in the disputes over aerodynamics are
not new ones. They mirror those in debates between mechanists and proponents
of experimental natural philosophy in the seventeenth and eighteenth centuries.
122 Prandtl, ‘The Generation of Vortices’, 739.

In considering the seventeenth-century cases in the first volume in this series,123

I drew attention to the difference in how the two parties constructed the domain
of investigation of physical enquiry. This difference was premised on two very
different understandings of ‘science’ (or natural philosophy as it was then). For the
mechanists, explanation was a matter of accounting for macroscopic behaviour in
terms of the mechanically described behaviour of the corpuscules/atoms making
up all material things. But in the 1660s, Boyle and Newton, though committed to
this as a general model, found that in order to account for certain phenomena
(pneumatic in Boyle’s case, optical in Newton’s) in a satisfactory way they had to
suspend their commitment to corpuscularianism as the sole reference point for
explaining physical behaviour. Since they also rejected what was in effect the only
alternative comprehensive form of foundational explanation, namely an appeal to
Aristotelian essences, they needed some way of organizing the phenomena under
investigation other than in terms of underlying structure. Now underlying
structure plays a crucial role in suggesting how the phenomena might be organ-
ized, not least in virtue of the clues it provides as to the connections between the
phenomena. But, particularly in the case of the optical spectrum, these clues may
turn out to be misguided, as Newton realized. One therefore has to proceed
without the benefit of any supposed clues, but there is one possible source of
guidance, the experimental apparatus itself. This produces a certain range of
phenomena which defy explanation in ‘fundamental’ terms, but which cannot
be dismissed because the results themselves cannot be faulted. The only way to
proceed is to take the results at face value and start from them, but from a
foundationalist perspective these results show no internal coherence, because in
corpuscularian terms they are anomalous. The way in which they are generated is
therefore crucial, not just because this is what legitimates them but also because, if
they have any coherence at all, it has something to do with the way in which they
are generated. It is the way in which they are generated, for example, that holds
them together as connected phenomena, and that excludes what might otherwise
seem—for example, on corpuscularian grounds—to be related phenomena. The
way in which the results are generated is a function of the experimental apparatus,
the way in which this apparatus is manipulated, and what one is able to do with it.
The experiment or instruments bring a domain of investigation into focus,
replacing the underlying structure that would traditionally have occupied this
role. What occurs is a form of tailoring of the explanation to results produced by
particular experiments or instruments, in short, a tailoring of explanans to explan-
andum, thereby reversing the normal direction of enquiry. But note that it is not a
question of one side alone having to tailor things. The seventeenth-century
mechanists always and as a matter of routine had to tailor the explanandum to
123 Gaukroger, The Emergence of a Scientific Culture, ch. 10. I summarize material from this
chapter here.
The Non-Discursive Products of Science 329
the explanans, and this tailoring—by means of the doctrine of primary and
secondary qualities, for example—was in fact as problematic as it was radical.
The rejection—in Boyle and Newton in the seventeenth century, in Gray in
the eighteenth, and in Faraday in the nineteenth—of the idea that a natural-
philosophical system determines in advance what is to be explained and what is
not, has obvious parallels with the engineers’ rejection of the mathematicians’
attempt to specify in advance the field of aeronautics and what resources one is
and is not entitled to use in exploring it. In other words, science in the modern era
has always been structurally complex. There is no doubt that the addition of
engineering to experiment as a crucial part of the scientific enterprise has been
very significant, but it has highlighted rather than transformed what might be
termed the problem of the non-discursive products of science. The construction
of ships in the eighteenth century, for example, has a number of parallels with the
construction of aeroplanes in the twentieth. As Larry Ferreiro notes: ‘sailing ships
were the most complex engineering structures of the day. They combined heavy
wooden construction of hull and masts with a dizzying array of standing rigging to
support the masts, hundreds of lines and blocks to control the yardarms and sails,
capstans for hauling up the anchors, tillers and wheels to turn the rudder, bilge
pumps, and such, for which the constructors had overall responsibility to integrate
into the ship.’124 Even more striking is the fact that there were two opposing
approaches to ship-building in the early decades of the eighteenth century: that of
the French, who were trying to rationalize the various trades along the lines of a
scientific method, and that of the (far more successful) British, whose constructors
picked up their skills in shipyards through apprenticeships, and were often
familiar with arithmetic and geometry in as far as it bore on their craft, but
were otherwise illiterate.125
THE NON-DISCURSIVE P RODUCTS OF S CIENCE
These issues have a direct bearing on how we conceive of the non-discursive

products of the science/engineering complex. In general terms, the question is:
once we have abandoned the idea of technology as ‘applied science’—that is to
say, once we have abandoned the idea that science is essentially something purely
theoretical, which may or may not then be applied—what are the consequences
for our understanding of what science is and what it does? In most respects, asking
this question is simply in the nature of an alert, for it goes beyond anything that
we can attempt to answer here. But there are two issues that do require our
immediate attention, however cursory. These are the nature of the products of
124 Larrie D. Ferreiro, Ships and Science: The Birth of Naval Architecture in the Scientific Revolution,
1600–1800 (Cambridge, MA, 2006), 24.
125 Ibid., 24–6.
science/engineering, and the nature of the conditions under which they are
produced and used. Moreover, we need to distinguish between the period
between the middle of the nineteenth century and the First World War, when
the products were predominantly machines, and the interwar period, when they
became predominantly complex systems.126
Consider the period before the Great War first. What is at issue is not just the
construction of these products, but their successful operation. From our discus-
sion of aeronautics, one might be led to think in terms of the reconciliation of
scientific knowledge with engineering knowledge. But that is misleading. Engin-
eering produces a very different kind of product from science as traditionally
conceived (i.e. knowledge). And if we are to understand the products of the
integration of science and engineering, then, crudely put, it is a question of
moving from a knowledge model to a working machine model. We might
think of machines (devices that apply a power source other than human or
animal) as embodied knowledge, but there is a complicating factor: the knowledge
that they embody is not just that of the construction of the machine, but also that
of its successful operation.
In asking for the conditions under which particular machines can actually be
used, we can start by noting that there has often been a significant delay between
the invention of a particular kind of machine and the use of the machine. Steam
engines are a case in point. Newcomen’s steam engine, first installed at a mine in
1712, has sometimes been lauded as heralding the Industrial Revolution in
Britain, but steam engines were marginal in production in eighteenth-century
Britain and, it would seem, from public awareness: there is no entry for them in
Johnson’s Dictionary, for example, and there is only one mention, in passing, of a
‘fire engine’ in Smith’s Wealth of Nations. Early railways had been horse-drawn,
working with an inclined plane over hilly ground, and as late as 1829, in the wake
of the introduction of the Stephenson engine, there were still those who thought
trucks pulled by a cable or by horses might be more efficient than steam engines.
One issue was safety. If heavy machinery exploded the results were disastrous.
British canon manufacturers, for example, were well aware that hidden hollows
and impurities in the metal used in French canons in the Seven Years War had
caused the canon to blow up at the point of firing with deadly consequences, and
the pressure built up in steam engines was a significant cause of concern in this
respect. James Watt was particularly opposed to high-pressure steam engines on
safety grounds, regarding their use as irresponsible. Richard Trevithick, inventor
of one of the most advanced forms of steam engine, wrote to a friend: ‘I have been
branded with folly and madness for attempting what the world calls impossibil-
ities, and even from the great engineer, the late Mr James Watt, who said to an
eminent scientific character still living, that I deserved hanging for bringing into
126 Here the category of ‘technoscience’ is potentially problematic, for it tends to collapse these
two into one another.
use the high-pressure engine.’127 Another issue was the weight of the engine.
Horses put no pressure on the rails at all, but steam engines were heavy and could
crack and distort the rails, which meant that engines with full loads running on a
regular basis could not run without wholly new, stronger sets of rails, but what
were these to be made of?
Rails and roads provide a good example of the difficulties in getting technology
to work in the requisite (economical and safe) way. The problem with rails was
that wooden rails were too soft and perishable whereas iron ones were too brittle.
The solution did eventually come, with the manufacture of wrought iron rails.
The method of turning scrap iron into wrought iron, by melting it down with raw
coal, in a process known as ‘puddling and rolling’, had been patented in 1783, but
the crucial developments came only when the process was improved radically with
the introduction of super-heated blast furnaces in 1828. The greater malleability
and tensile strength of the wrought iron tracks, which from 1828 were rolled out
in appropriate lengths, were as crucial for the sudden take off of interest in
railways in the 1830s as were the engines. At the same time as the transformation
of railway tracks, Britain witnessed a transformation of its roads. Significant
attention had been paid to alleviating problems with road transport by improving
the suspension of carriages in the eighteenth and early nineteenth centuries, to
little avail. But in 1816 John Loudon McAdam, then sixty years old and with no
scientific background or engineering experience, realized that the problem lay not
in the suspension of the vehicle but in the surfaces of the roads on which it
travelled. The solution was to provide a top layer of small stones which, when
compressed by carriage wheels, formed a smooth well-drained surface.128
Steam engines, no matter how sophisticated, would not have been able to have
anything like the impact they had without the wholly independent discovery of an
economic process for the production of wrought iron. Nor could rail travel have
been successful without the wholly independent invention of the electrical
telegraph, which provided the basis for the long-distance signalling technology
that allowed for the systematization and coordination of railway travel. And the
same time, motor vehicles would have proved useless without prior innovations of
the kind provided by McAdam for roads. Technological developments are dead
ends without the right conditions for use. But it is not just a matter of material
conditions here. Levels of expertise also play a necessary role in establishing the
right conditions for use.
The question of machine expertise was paramount. The kind of large-scale
machinery deployed in cotton and woollen mills, for example, was such that the
manufacturer needed to dismantle the machinery and ship it in parts. This meant
that a skilled engineer was needed at its destination to reassemble it. Consider the
case of Benjamin Phillips’ transfer of machines, including a version of Compton’s
127 Quoted in Weightman, The Industrial Revolutionaries, 49. 128 Ibid., 123–5.
mule, to America in 1783. To avoid the customs restrictions caused by the

Revolutionary Wars, he packed the machines in pottery casks, loading them on
a ship at Liverpool and sending his son on to Philadelphia to receive them. Phillips
himself was to accompany the machine parts, reconstructing them when he
arrived in America, but he fell ill on the voyage and had to be sent home,
where he died shortly afterwards, so it was left to his son to reassemble them on
their arrival. But this was a formidable task, one the son was unable to accomplish,
and ultimately he was forced to sell the parts on. The new purchaser found that he
also was unable to assemble them, and ended up putting them in storage,
eventually selling them on himself, to a new purchaser who realized he had no
option but to send them back to England, where, now successfully reassembled,
they stayed.129
Discussion of technological expertise has tended to focus on the role of
expertise in the design of machines, but the role of expertise in the working of
machines is equally important. Technology does not consist simply of machines,
but is rather a union of machines and specific skills. This was perhaps most
evident in the popular electrical machine exhibitions of the mid-nineteenth
century. As Morus has pointed out, such exhibitions ‘provided spaces where
the machine-maker as showman could demonstrate his control over the machine
and over the audience for his display. Those who worked and displayed the
results of their labors at such places were skilled mechanics who needed a space
where by demonstrating their control over their machines and experiments, they
could also demonstrate their ownership of these technologies and the phenom-
ena they produced. Invention was very much about controlling the products of
labor.’130
Similarly in the more mundane context of commerce. Exporting machines
typically meant sending an engineer who could set up the machine and keep it
working. In some cases the skills required were not unlike those of the experi-
mental scientist. This was particularly clear in the seventeenth century, when the
two roles were performed by the one person. Hooke and Huygens noted that they
were the only two people who were able to get their air pumps, particularly
mechanically complex machines, to work satisfactorily. Here designing the ma-
chine and working it successfully are part of the same thing, and this, I suggest,
should be our model for understanding how machines function. Machines and
their successful manipulation go together. And just as the roles of invention of the
machine and enabling it to work are part of the same enterprise, the physical
object that constitutes the machine cannot be separated from the process by
which it functions and the conditions under which this functioning is successful.
But the non-discursive products of the science–engineering complex are not
129 Ibid., 38–9. More generally see David Jeremy, Artisans, Entrepreneurs and Machines: Essays on
the Early Anglo-American Textile Industries (Aldershot, 1998).
130 Morus, Frankenstein’s Children, 71.
confined to objects such as machines, and it is important that we register the

transformations of the non-discursive products of the science–engineering com-
plex that are typical of the interwar period.
Hughes draws attention to the problems in simply extrapolating from the pre-
war machine technology to that of the interwar period:
To associate modern technology solely with individual machines and devices is to overlook
deeper currents of modern technology that gathered strength and direction during the
half-century after Thomas Edison established his invention factory at Menlo Park. Today
machines such as the automobile and the airplane are omnipresent. Because they are
mechanical and physical, they are not too difficult to comprehend. Machines like these,
however, are usually merely components in highly organized and controlled technological
systems. Such systems are difficult to comprehend, because they also include complex
components, such as people and organizations, and because they often consist of physical
components, such as the chemical and electrical, other than the mechanical. Large
systems—energy, production, communication, and transportation—compose the essence
of modern technology.131
As representative of the early developers of electrical technology, Hughes takes the
electrical light and power systems of Samuel Insull, noting how different these
were from the mechanical products of Ford and others. In particular, it was
concepts of electrical circuitry rather than machinery that shaped the ways in
which builders of electrical systems thought and acted: ‘they manipulated inter-
actions, not the simpler linear relations of cause and effect. The builders of
electric-power and chemical-process plants also envisaged flow rather than the
movement of batches of materials and mechanical parts. Instead of being the age
of the machine, the interwar years emerged as the apogee of the age of electric
power and chemical processes. The machine as a symbol of an age applies better to
the British Industrial Revolution of more than a century earlier.’132
What is at issue here is not confined to the production of commercial products,
but runs through the whole of the science–engineering complex, impacting on
131 Hughes, American Genesis, 184–5. Hughes defines a system as being ‘constituted of related
parts or components. These components are connected by a network, or structure, which for the
student of systems may be of more interest than the components. The interconnected components of
technical systems are often centrally controlled, and usually the limits of the system are established by
the extent of this control. Controls are exercised in order to optimise the system’s performance and to
direct the system toward the achievement of goals. The goal of an electric production system, for
example, is to transform available energy supply, or input, into desired output, or demand. Because
the components are related by the network of interconnections, the state, or activity, of one
component influences the state, or activity, of other components in the system. The network
provides a distinctive configuration for the system.’ Thomas P. Hughes, Networks of Power:
Electrification in Western Society 1880–1930 (Baltimore, 1983), 5.
132 Hughes, American Genesis, 186. Cf. Jon Agar’s account, in The Government Machine:
A Revolutionary History of the Computer (Cambridge, MA, 2003), of the uptake of technology in
the British civil service, and the way in which the civil service, ‘cast as a general-purpose universal
machine, framed the language of what a computer was and could do.’ (3) See also Agar, Science in the
Twentieth Century and Beyond, ch.16, on information systems.
such ‘basic’ questions as the nature of experimental enquiry. To understand how

the results of technologically complex experiments in nuclear physics are gener-
ated for example, it is often no longer a question of constructing the apparatus for
oneself and checking that the results are as claimed. It is rather one of learning
how the experimental set-up works in the course of prolonged visits to the
laboratory where the experiment has been performed, learning the skills involved
and the protocols used.133 Laboratories have become large-scale enterprises, with
all the organizational complexity this brings with it, and recognition of this has
sometimes been slow. As Daniel Todes has noted, for example, in each of the four
consecutive years that Pavlov was considered for the Nobel Prize, the committee
were unable to decide how to credit the experimental work done in his large
laboratory complex:
Guided by the image of the heroic lone investigator, the Nobel Prize Committee here
confronted a different form of scientific production. Just as the workshop was yielding
pride of place to the factory in goods production, so, as the nineteenth century wore on,
were leading laboratory scientists increasingly likely to be the managers of large-scale
enterprises, Justus von Liebig and Felix Hoppe-Seyler in chemistry, Karl Ludwig and
Michael Foster in physiology, Robert Koch and Louis Pasteur in bacteriology, and Paul
Ehrlich in immunology all presided over distinctively social enterprises involving substan-
tial capital investment, a relatively large workforce, a division of labour, and a productive
process that involved managerial decisions.134
The general lesson from the developments that we have examined in this chapter
is that taking technological and engineering developments seriously—both in the
production of explanations, and also in the generation of non-discursive products
such as aeroplanes and electrical systems—means rethinking the nature of science.
Instead of converging on an increasing unity, the modularity and pluralism of
science become not simply reinforced but wholly transformed through the
intimate connections between science and technology. As if that isn’t enough,
once we abandon the attempt to see science as driving modernity and focus
instead on technology, we find that the idea that technology drives modernity is
open to different kinds of objections, as we saw in Chapter 2. But if neither
science nor technology, individually or collectively, can be considered to drive
modernity, what then is their relation to it? To pose this question fruitfully we
need to ask how science was promoted as carrying the values of progress and
modernity in the nineteenth and twentieth centuries. The answer lies in what has
been termed ‘popular science’, and I now want to show how popularized and
133 See Collins, Changing Order; and Galison, How Experiments End.
134 Daniel P. Todes, ‘Pavlov’s Physiology Factory’, Isis 88 (1997), 205–46: 206. Cf. the remark of
Edison’s chief engineer, Francis Jehl: ‘Edison is in reality a collective noun and means the work of
many men’: quoted in Richard Munson, From Edison to Enron: The Business of Power and what it
Means for the Future of Electricity (Westport, CT, 2005), 14.
fictionalized forms of science played a formative role in establishing the standing

of science. These have shaped the values that have been associated with science in
the nineteenth and twentieth centuries to such an extent that it would not have
been possible for it to have the standing it has had without them. In the end, it
was not science as a model of truth that placed it at the centre of modern culture,
but science as a model for the future.

Oso 9780198849070 Chapter 11

Uploaded by

Copyright:

Available Formats

Oso 9780198849070 Chapter 11

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Oso 9780198849070 Chapter 11

Uploaded by

Copyright:

Available Formats

OUP CORRECTED PROOF – FINAL, 11/12/2019, SPi

292 Technology and the Limits of Scientiﬁc Theorizing

Technology and the Limits of Scientiﬁc Theorizing 293

5 Sarton, The History of Science and the New Humanism, 188.

294 Technology and the Limits of Scientiﬁc Theorizing

Technology and the ‘Unruliness’ of Science 295

on ‘science’ a disuniﬁed, unruly, modular character. Second, I want to look at the

TECHN OLO GY AND THE ‘ UN RU LIN ES S’ OF SCIENCE

I referred earlier to an unruliness about scientiﬁc practice, something that is at

296 Technology and the Limits of Scientiﬁc Theorizing

Technology and the ‘Unruliness’ of Science 297

looked at the role of autonomous models, in the form of paper-tools, in chemistry.

298 Technology and the Limits of Scientiﬁc Theorizing

The category of ‘applied science’ is a nineteenth-century invention, and the

Technology and the ‘Unruliness’ of Science 299

300 Technology and the Limits of Scientiﬁc Theorizing

A striking nineteenth-century example is the Giffard steam injector (Fig. 10.1),

Technology and the ‘Unruliness’ of Science 301

302 Technology and the Limits of Scientiﬁc Theorizing

Technology and the ‘Unruliness’ of Science 303

304 Technology and the Limits of Scientiﬁc Theorizing

Chair in 1855. He combined serious interests in both science and technology,

37 Channell, ‘The Harmony of Theory and Practice’, 44.

Technology and the ‘Unruliness’ of Science 305

43 Channell, ‘The Harmony of Theory and Practice’, 47. 44 Ibid., 49–50.

306 Technology and the Limits of Scientiﬁc Theorizing

Technology and the ‘Unruliness’ of Science 307

308 Technology and the Limits of Scientiﬁc Theorizing

The Standing of the Scientist 309

310 Technology and the Limits of Scientiﬁc Theorizing

France provides a revealing starting point. In the immediate post-revolutionary

The Standing of the Scientist 311

312 Technology and the Limits of Scientiﬁc Theorizing

71 Richards, ‘Rigor and Clarity’, 307–8.

The Standing of the Scientist 313

314 Technology and the Limits of Scientiﬁc Theorizing

79 See Gaukroger, Francis Bacon, ch. 2.

The Standing of the Scientist 315

316 Technology and the Limits of Scientiﬁc Theorizing

secure, it is immune to changes that would have compromised an exemplary form

Science and Machines 317

Every aeroplane is to be regarded as a collection of unsolved mathematical

318 Technology and the Limits of Scientiﬁc Theorizing

92 Bryan, Review of R. de Villamil, 379.

Science and Machines 319

94 Hermann von Helmholtz, ‘Über discontinuirliche Flüssigkeits-Bewegungen’, Monatsbericht

320 Technology and the Limits of Scientiﬁc Theorizing

97 John Henry Ledeboer, ‘The Function of Literature’, Aeronautics 11 (1916), 33.

Science and Machines 321

322 Technology and the Limits of Scientiﬁc Theorizing

Science and Machines 323

Lanchester’s circulation theory had been received enthusiastically right from

324 Technology and the Limits of Scientiﬁc Theorizing

113 Bloor, Enigma of the Aerofoil, 178.

Science and Machines 325

116 As noted in Bloor, Enigma of the Aerofoil, 329.

326 Technology and the Limits of Scientiﬁc Theorizing

Science and Machines 327

122 Prandtl, ‘The Generation of Vortices’, 739.