Download as pdf or txt
Download as pdf or txt
You are on page 1of 12

Studies in History and Philosophy of Science 89 (2021) 283–294

Contents lists available at ScienceDirect

Studies in History and Philosophy of Science


journal homepage: www.elsevier.com/locate/shpsa

English engineer John Smeaton's experimental method(s): Optimisation,


hypothesis testing and exploratory experimentation
Andrew M.A. Morris
Centre for Logic and Philosophy of Science, Vrije Universiteit Brussel, Pleinlaan 2, Room 5B425, B-1050, Brussels, Belgium

A R T I C L E I N F O A B S T R A C T

Keywords: In this paper I provide a detailed account of eighteenth-century engineer John Smeaton's experimental methods,
John Smeaton with the aim of bringing our understanding of his work into line with recent research in the history and phi-
Science-technology relation losophy of science. Starting from his use of the technique of parameter variation, I identify three distinct
Parameter variation
methodological aims in the research he carried out on waterwheels, windmills and hydraulic mortars. These aims
Exploratory experimentation
Hypothesis testing
are: optimisation, hypothesis testing and maxim generation. The main claim of this paper is that Smeaton did
Optimisation more than merely improve engineering methods by systematising earlier artisanal approaches, which is the classic
view of Smeaton's method developed by historians of technology in the 1990s. I argue instead that his approach
bridged the divide between science and technology, by integrating both hypothesis testing and exploratory
experimentation. This is borne out, in particular, by the way that Smeaton emphasised the exploratory side of the
work he published in the Philosophical Transactions, in contrast to his account of the construction of the Eddystone
lighthouse, which was aimed at a broader, non-specialist public. I contribute to recent research on exploratory
experimentation by showing – in line with other work on this topic – that exploratory experimentation is not
incompatible with hypothesis testing. This new perspective on Smeaton's method will hopefully lead to further
research and new insights into the relation between science and technology at the start of the Industrial
Revolution.

1. Introduction positions Smeaton as an ingenious practical tinkerer who systematised


and improved ad hoc artisanal methods. This I refer to as the ‘classic view’
John Smeaton (1724–1792) was an English engineer active in the in what follows. In the last few decades however, parameter variation has
second half of the eighteenth century. His main achievements include the mainly been associated with the experimental approach called explor-
Eddystone lighthouse off the coast of Plymouth, ground-breaking ex- atory experimentation.3
periments on waterwheel efficiency, and improvements to a wide variety In this paper I show that Smeaton used parameter variation much
of scientific instruments and machines.1 He is also known as an early more broadly, as part of an experimental methodology with three
proponent of the method of parameter variation.2 different objectives: optimisation, hypothesis testing and maxim gener-
Although the historiography of eighteenth-century science has been ation via exploratory experimentation. By looking more closely at his
revolutionised over the past few decades, not much research has been methodology in the research into waterwheels, windmills and hydraulic
carried out on Smeaton's role in the history of science and technology, mortars, we will see that Smeaton's approach went beyond merely sys-
our understanding of which has changed little since the early 1990s, tematising earlier artisanal and practical methodologies focusing on the
when Walter Vincenti (1990) and Donald Cardwell (1994) sketched the design considerations, such as optimisation, that are central to engi-
outline of a Smeatonian method based on a widely applicable method of neering and technology. Smeaton also tested common hypotheses and
systematic parameter variation and the use of scale models, which extracted empirical regularities from his experiments, and such activities

E-mail addresses: amorris@vub.be, andrewmamorris@gmail.com.


1
For an overview of Smeaton's achievements, see Turner and Skempton (1981).
2
This is a key feature of the analysis in Vincenti (1990).
3
For an extended treatment of exploratory experimentation, see Steinle (2016).

https://doi.org/10.1016/j.shpsa.2021.07.004
Received 8 December 2020; Received in revised form 5 July 2021;
Available online 20 September 2021
0039-3681/© 2021 Elsevier Ltd. All rights reserved.
A.M.A. Morris Studies in History and Philosophy of Science 89 (2021) 283–294

were the preserve of more strictly scientific research, which was driven cites both Vincenti and Cardwell, Smeaton's method was ‘a method of
less by directly practical outcomes, and more by the relation to theory, parameter variation through experimentation, which is a systematic way
either testing an existing hypothesis or generating a new hypothesis.4 of making gradual improvements in λ [prescriptive knowledge] in the
By giving a detailed account of Smeaton's experimental practices, I absence of a wide epistemic base’ (p. 42).
aim to bring our understanding of his methodology into line with recent Channell (2009) points out that, ‘[w]ith the development of steam
research on scientific practice during the Industrial Enlightenment.5 engines, railways, ocean-going iron-hulled ships, and large scale iron
Further, Smeaton was a pioneering civil engineer, so gaining a deeper bridges, it became impractical and uneconomical for engineers to use
understanding of his methodology will also provide insight into the traditional rule-of-thumb or trial-and-error techniques’, so engineers like
development of the relation between science and technology in the early Smeaton developed more sophisticated methods that could replace such
Industrial Revolution.6 techniques (p. 126). Channell views the relation between science and
technology during the eighteenth century as led by engineers, on the one
2. What is at stake? hand, who were trying to systematise their approach, and scientists, on
the other hand, who were trying to use their scientific research in prac-
Before turning to look at Smeaton's method, it is worth getting an tical applications.
understanding of what is at stake in the analysis that follows. The classic So the classic view, whether implicitly or explicitly, has tended to
view of Smeaton's method not only provides an interpretation of Smea- classify Smeaton as a systematic but fairly unscientific practical man,
ton's work, it also implies a certain understanding of the relation between who improved engineering methods with little use of any scientific
science and technology, which I spell out in this section. By reinter- knowledge or methods. Although this classic view has been undermined
preting Smeaton's method in light of more recent research into scientific by recent trends in the history and philosophy of science, as we will see in
methodology, I also call into question the vision of the relation between the next section, Smeaton's methodology has not received the full reap-
science and technology that is implicit in the classic view. praisal that it deserves.
Our understanding of the relation between science and technology
during the Industrial Revolution has been influenced by two somewhat 3. Recent historiographical trends
outdated strands of historical research from the twentieth century, which
are represented in the debate between A.E. Musson and Eric Robinson on A number of developments in recent history and philosophy of sci-
the one hand, and A. Rupert Hall on the other. Musson and Robinson ence have changed the way that science itself is understood. With the
claim that science was a central part of the early Industrial Revolution.7 increased focus on practices and experiments (trends that are sometimes
Hall (1974), on the other hand, argues that the technological de- labelled as the ‘practical turn’ or ‘practice turn’ and the ‘new experi-
velopments of the late eighteenth century actually owed very little to the mentalism’), researchers began to focus more on the details of the
science of the period. experimental work carried out by scientists throughout history.8 In
This division corresponded to disciplinary distinctions between his- addition to the philosophical focus on practice, the turn towards scientific
torians of science and historians of technology. Vincenti, an historian of practices has a long history in the sociology of science and technology,
engineering, rejects the idea of a close relationship between science and beginning in the 1970s, which is often associated with the work of re-
technology because this has often reduced technology to applied science. searchers such as Harry Collins, David Gooding, Steven Shapin or Simon
Instead, he proposes to study engineering as a distinct intellectual Schaffer.9
domain, with its own methods. By focusing on Smeaton, Vincenti (1990) And this new focus revealed that much experimental science was
articulates what he takes to be the quintessential engineering approach, highly technological, involving the development of complex experi-
forming ‘the basis of an autonomous tradition in engineering research’, mental devices and the deft manipulation of those devices.10 Much of this
which was later applied successfully to the design of devices such as attention given to experiment, not only in the philosophy of science, but
aircraft propellers (p. 138). also in the related fields of history of science and sociology of science and
Despite believing that science and technology go hand-in-hand, technology, has emphasised the importance of instrumentation in
Cardwell (1994) also argues that ‘[w]ith Smeaton, engineering in En- experimental scientific practice.11 This increased focus on experimental
gland became autonomous, with at least one common method for all practices brought into relief the multiplicity of experimental strategies
problems’ (p. 195). He claims that Smeaton's method did not ‘involve used throughout the history of science. As a result of this research,
scientific research into the basic laws of the subject or system under experiment was no longer assumed to merely test hypotheses – as it was
investigation’ but instead what it did was ‘formally banish from the realm more closely studied, so a wide variety of experimental aims and methods
of techniques the element of craft mystery, passed down from master to were identified (Hacking, 1983). Another consequence was the increased
apprentice’ (Cardwell, 1994, p. 195). So both Vincenti and Cardwell see focus on the technological nature of scientific practices.
Smeaton as systematising practical, artisanal methods, without actually It could easily be argued that talk of a science-technology distinction
doing anything specifically scientific. in the eighteenth century is anachronistic. It would perhaps be more
This classic view of Smeaton's method was adopted by more recent accurate to refer to the relation between natural philosophy and early
researchers such as Joel Mokyr or David F. Channell, despite the fact that engineering. This has been the topic of much valuable secondary litera-
our understanding of the distinction between science and technology has ture, in particular Larry Stewart's and Margaret Jacob's research, showing
been evolving over the last two or three decades. For Mokyr (2002), who how the Royal Society was central to the application of ‘Newtonian’
natural philosophy to practical mechanics.12

4
There have been many proposals about ways to understand the science-
technology distinction. See, for example, Radder (2009), Radder (2003) and
8
Hansson (2015). Some of the historically oriented literature on this topic will be For some reading on the ‘practice turn’, see Soler (2014). The term ‘new
discussed in the following section. experimentalism’ was coined by Ackermann (1989).
5 9
For more on the Industrial Enlightenment, see Jones (2008) and Mokyr See, for example, Collins (1992), Gooding, Pinch, and Schaffer (1989) or
(2002). Shapin and Schaffer (1985).
6 10
The terms ‘science’ and ‘technology’ are anachronistic. For more on the It is unsurprising that Hall (1974) determines that little theoretical scientific
historiographical issues related to the distinction between knowing and doing, knowledge contributed directly to the inventions of the Industrial Revolution.
11
see Roberts, Schaffer, and Dear (2007). See, for example, Gooding, Pinch, and Schaffer (1989), Baird (2004) or
7
See Musson and Robinson (1969, pp. 27–30) and Musson (1975, pp. Bertoloni Meli (2006).
12
633–637) for the critique of Hall. In particular, see Stewart (1992) and Jacob and Stewart (2004).

284
A.M.A. Morris Studies in History and Philosophy of Science 89 (2021) 283–294

To this research can be added various studies on the different spaces 4. Overview of Smeaton's methods
of knowledge transmission and exchange: sites where different knowers
came into contact, such as workshops, laboratories and factories.13 At the Smeaton's overarching method can be divided into three phases: 1)
same time there was a proliferation of hybrid figures who, in different optimisation, 2) hypothesis testing and 3) maxim generation. Smeaton's
ways, moved between these different ‘milieux’ and connected different primary aim as a civil engineer was to determine the most efficient,
practices of knowing and making from the worlds of natural philosophy effective waterwheel, windmill or hydraulic mortar, which he did by
and engineering.14 varying parameters and choosing the most optimal configuration. The
Danilo Capecchi's recent monograph on epistemology and natural parameters he varied often corresponded to workmen's maxims or
philosophy in the eighteenth century draws attention to Smeaton's role contemporary construction practices, which were also put to the test in
linking science and technology, urging us to see the relation as more than this optimisation process. Finally, Smeaton generated his own maxims
science influencing engineering, as defined by earlier scholars like Hall based on the results of his experiments.
(Capecchi, 2021). He also emphasises the importance of ‘less formalized This section will look at how Smeaton put these methods into prac-
theoretical considerations’ produced by Antoine Parent (1666–1716) tice. As noted by Cardwell and Vincenti, Smeaton carried out parameter
and Jean-Charles de Borda (1733–1799) as well as the crucial influence variation on scale models. In the case of his research into waterwheels, he
of natural philosophical experimental methods (Capecchi, 2021, p. 505). used a miniature waterwheel, and for his experiments on windmills, he
The role of experimental practice in Smeaton's work has also been devised an apparatus that replicated the action of a windmill. In the
emphasised by Betty Jo Teeter Dobbs and Jacob (1995), who note that experiments on hydraulic mortars, Smeaton carried out tests on two-
‘[w]ith these disciplined methods of verification and replication drawn inch-wide balls of mortar. As we will see, Smeaton selected which pa-
from scientific practices, British engineers believed themselves to be rameters to vary based on either natural constraints (such as the quantity
scientists, or at the least their imitators’ (p. 113). of water or wind that would power the waterwheels or windmills) or
The current paper builds on this point about experimental practices, contemporary construction practices and methods.
by attempting to provide a methodological framework for our under- By varying these parameters, Smeaton was able to create a table
standing of Smeaton's experimental method, that might subsequently be synthesising his results, which allowed him to select the best kind of
used to trace the influences at work on Smeaton. A first suggestion might waterwheel, windmill or hydraulic mortar. This was the primary aim of
be to compare this method with Newton's own method, following the his research: to optimise waterwheels, windmills and hydraulic mortar
suggestion of Dobbs and Jacob (1995), although this goes beyond the recipes to provide the most efficient power output and the strongest
scope of the present paper. mortar. In so doing, Smeaton challenged some common conceptions
This could help understand how the engineering approach based on about the functioning and construction of waterwheels, windmills and
trial-and-error and tinkering became increasingly ‘scientific’, as David hydraulic mortars. His final aim was to provide principles or ‘maxims’
Philip Miller (2019) calls it (p. 156). It is for this reason that comparison that could be used by other engineers to help guide their practice.
with cases of exploratory experimentation are instructive – exploratory The following brief methodological remark from Smeaton's paper on
experimentation provides a framework for understanding the devel- waterwheels provides a helpful introduction to our discussion of these
opment of a scientific discipline in the absence of a fully-developed methods:
theoretical background. This goes further than the classic view
Many obvious and considerable improvements upon the common
because Smeaton's methods are themselves seen as being properly
practice naturally offer themselves from a due consideration of the
scientific.
principles here established, as well as many popular errors show
My aim in this respect is modest. I hope to go into some detail
themselves in view: but as my present purpose extends, no farther
about the methods Smeaton used in his work on waterwheels, wind-
than the laying down such general rules as will be found to answer in
mills and hydraulic mortars, in order to show that Smeaton's status as
practice, I leave the particular application to the intelligent artist, and
a hybrid knower extended to his methodology (cf. Morris, 2021).
to the curious in these matters (Smeaton, 1759, p. 138).
Smeaton not only moved in social circles connected with both the
world of natural philosophy, such as the Royal Society Club, and the Smeaton emphasises here how his development of general principles
world of engineering, as a founding member of the Institution of Civil (‘laying down such general rules’) can serve both to optimise waterwheel
Engineers, but he also carried out research on natural philosophical design (‘improvements upon the common practice’) and refute existing
topics such as vis viva, publishing two papers on the subject in the hypotheses about waterwheel functioning (‘popular errors’).
Philosophical Transactions, in addition to papers on instrument making However, the way in which he combined these different methods or
and engineering.15 techniques was flexible. In some cases – particularly the more scholarly
This hybridity is also reflected in the way that Smeaton presented work published in the Philosophical Transactions – Smeaton placed a
his engineering methods – in work published in the Philosophical greater emphasis on the search for general rules or maxims, whereas in
Transactions, Smeaton gave greater emphasis to the search for general other cases – such as the account of research into mortars for the Eddy-
principles or rules, whereas in his Narrative on the construction of the stone lighthouse – he emphasised optimisation and testing practical hy-
Eddystone lighthouse, the emphasis was placed on solving problems potheses over the generation of maxims. In this way we can see how the
by evaluating workmen's maxims and optimising the object of study.16 distinction between natural philosophy and early engineering can be
This difference in emphasis tells us that certain kinds of early engi- mapped onto the more recent distinction between science and technol-
neering research were already of interest to fellows of the Royal ogy. When Smeaton was working in the eighteenth century, exploratory
Society. research was considered of greater interest to the natural philosophy-
oriented readers of the Philosophical Transactions, as opposed to the
work of testing workmen's hypotheses and optimisation.
13 Smeaton carried out a broad variety of different kinds of research
See Jacob (2007) and Stewart (2007).
14 during his career, writing many different reports and accounts of that
See Klein (2017) on the figure of the ‘hybrid expert’. Jones (2008, pp.
116–129) discusses the related figure of the ‘savant fabricant’. research. Interestingly, the two sources of information about Smeaton's
15
Smeaton's research on vis viva can be found in Smeaton (1776) and Smeaton methods in his engineering experiments span almost the entirety of his
(1782). On this research, see Morris (2018). On Smeaton's hybrid status and his career. The waterwheel and windmill experiments were carried out in the
social circle, see Morris (2021). 1750s and the results were published in 1759, and the Eddystone light-
16
I thank the anonymous reviewer who suggested I give this issue more house was built between 1756 and 1759, although Smeaton only
attention.

285
A.M.A. Morris Studies in History and Philosophy of Science 89 (2021) 283–294

published his Narrative of the construction of the lighthouse in 1791, Smeaton varied the number of paddles on the wheel, which was a
shortly before he died.17 Importantly, the Narrative included an account question that had begun to interest researchers when Henri Pitot
of the mortar research he carried out after the lighthouse was finished, (1695–1771) made the case for a certain distribution of waterwheel
suggesting that the core components of the methods that he used paddles in a paper presented to the French Academie des sciences in 1729,
remained more or less stable (Morris, 2021, p. 13). which was then repeated by Bernard Forest de Belidor (1698–1761) and
John Theophilus Desaguliers (1683–1744), both of whose work was very
5. Optimisation via hypothesis testing: waterwheels and familiar to Smeaton.18 Varying the weight lifted by the turning of the
windmills wheel corresponded to the different uses to which the waterwheel would
be put (such as grinding corn or raising water), represented in the
Smeaton's background in instrument-making prepared him well for experiment by the weight raised by the action of the wheel, shown at (R)
working on experimental models (Turner and Skempton, 1981, p. 8). In in Fig. 1.
order to carry out sufficiently broad parameter variation, Smeaton Perhaps the most important parameter to be varied was the type of
needed to work on a small scale, which he did by building model wa- force turning the wheel – impulse or gravity – in order to compare the
terwheels and windmills, and by studying small samples of mortar. efficiency of overshot and undershot waterwheels. Since the beginning of
However, he still needed to establish which parameters he could the eighteenth century, practical mechanics and engineers had begun to
reasonably vary in a limited amount of time. Scale models and small suspect that undershot and overshot wheels, powered by impulse and
samples made his work easier, but he still could not vary an unlimited gravity respectively, were not equivalent. Smeaton (1759) wrote that: ‘[i]
number of parameters. This was where hypothesis testing became useful: n reasoning without experiment, one might be led to imagine’ that overshot
to determine which parameters could be varied most fruitfully, Smeaton and undershot waterwheels were equally efficient (p. 124). In order to
looked to contemporary construction practices and ideas, which he convert his apparatus to an overshot design, Smeaton added a race –
would be able to put to the test (such as ‘salt water is not suitable for represented by the dashed lines extending from g to f in Fig. 2 – which
mortars’) or compare with other hypotheses (such as comparing overshot directed the water over the top of the wheel, rather than under it. Thanks
and undershot waterwheels, or comparing different ratios of lime to to Smeaton's instrument making ingenuity, the difference between two
tarras in mortars, or different angles of attack in windmill sails). waterwheel models – undershot and overshot – could be reduced to the
This was how Smeaton selected the parameters to vary. His results variation of a parameter in a single model. Smeaton demonstrated that
would combine both the testing of relevant hypotheses as well as the the overshot wheel was more efficient than the undershot, and in so
optimisation of the process under examination. Optimisation is a design doing became embroiled in the vis viva controversy between Leibnizian
feature – it does not directly tell us anything about the world. The en- and Newtonian conceptions of motive force (see Morris, 2018).
gineer outlines an objective, such as waterwheel efficiency, defined as The same approach was continued in his research on windmills.
the relationship between input and output, or material strength, defined Smeaton designed a device (Fig. 3) with a model windmill at the end of a
as the ability to retain hardness under specific conditions. They will then revolving arm, freeing him from the reliance on natural wind, because
be able to use the data produced by the variation of parameters to select the faster he turned the device, the stronger was the air resistance acting
the best solution, ie. the waterwheel with the greatest output, or the most on the windmill. Once again, he varied five different parameters.
resistant mortar. Smeaton compared windmill sails weathered according to the research of
In addition to designing the most suitable materials and devices, this Parent, who determined that each sail should be angled at 35 , to ‘the
approach to optimisation meant that Smeaton directly tested the notions angle most commonly made use of by practitioners,’ which was the much
of the workmen and artisans. The parameters that had already been shallower 12 –15 . Smeaton also varied the twisting of the sails, which
subject to debate were often the most suitable parameters to vary, was important when the sails were understood to be in constant motion
because disagreement indicated that improvements might still be made. rather than accelerating from rest – this because the outer edges of the
Smeaton designed a simple yet precise apparatus for testing water- sail turned faster, so a shallower angle was needed at the edges to
wheel efficiency (Fig. 1). The device was a scale version of a waterwheel minimise air resistance to the rotary motion. Smeaton (1759) compared
with an additional pulley system for raising weights, and a hand-pump the angles suggested by the research of Scottish mathematician, Colin
mechanism for circulating the water. Smeaton tried to vary as many Maclaurin (1698–1746), to the sails weathered ‘in the Dutch manner,’
parameters as possible in order to find both the most efficient way to run that he likely observed during a tour of the low countries four years
the machine, as well as taking note of the maximum output of the ma- earlier, in 1755 (pp. 146–148).19
chine. He varied five main parameters by adjusting the device – most of Another parameter to be varied was the wind speed, in order to
these adjustments had been integrated into the design of the apparatus. It ensure that the windmill would be efficient in different weather condi-
is worth indicating here that variation of parameters need not be un- tions. This Smeaton did by varying the force with which he pulled the
derstood as implying that the parameter under consideration can be cord, Z, which turned the arm. He was able to regulate the rate at which
varied continuously – many parameters varied by Smeaton were the arm turned by timing the motion with the built-in pendulum (V)
discontinuous, such as materials used in mortars, number of paddles on a attached to the device. This did away with any reliance on natural wind
waterwheel, or whether the waterwheel was undershot or overshot. In (Smeaton, 1759, pp. 138–139). The results of Smeaton's experiments can
this paper, I take parameter variation to refer to any experimental feature be seen represented in Fig. 4.
that can be changed whilst leaving the rest of the experiment the same. Here it is perhaps worth correcting a misconception about the func-
Smeaton varied the size of the aperture and the height of the water in tioning of this apparatus that has crept into the secondary literature. John
the reservoir, which was equivalent to varying the flow rate of the water D. Anderson (1997), in a broad survey of the history of aerodynamics,
provided by the river or stream. In Fig. 2, we see that Smeaton raised or claims that ‘[t]he windmill blades at the end of the whirling arm were
lowered the sluice (I) by fixing it with the pin (K). The height of the water spun by a cable-and-pulley mechanism activated by a falling weight’ (p.
in the reservoir was measured by the float (G) and maintained by steady 58). Channell (2019), following Anderson, writes that Smeaton was
pumping. The pump mechanism was designed to reduce the rising and ‘using a falling weight to cause the windmill blades to spin’ (p. 93).
falling of the water with each action of the pump, with features such as It appears, however, that this was not how the apparatus functioned.
the tapered piston at (V) (Smeaton, 1791, p. 109). Instead, if we look at Fig. 3, we see that Smeaton pulled the cord (Z)

17 18
For more information about the publication of the Narrative, see Morris See Reynolds (1983, p. 210). For Pitot's original study, see Pitot (1729).
19
(2021). For the account of the trip, see Smeaton ([1755] 1938).

286
A.M.A. Morris Studies in History and Philosophy of Science 89 (2021) 283–294

Fig. 1. A sketch of Smeaton's waterwheel apparatus from Smeaton (1759, p. 101).

causing it to unwind the bobbin (H), which rotated the arm (FG) to which also burnt and slaked samples of the lime himself, in order to establish
the windmill sails were attached at (I). As the sails moved through the air the best method for carrying out this process (Smeaton, 1791, p. 103). He
on the rotating arm, they also began to turn, winding in the cord leading varied eight different parameters in his hydraulic lime experiments.
to the pulley mechanism (ONM), and this raised the weight held in the Smeaton (1791) varied whether the limestone was burnt completely
scales at (P). The more efficient the sails, the more weight would be lifted or not, as he ‘had heard much complaint from the workmen, of limes not
in a given amount of time. being well burnt’ (p. 103). He tested the common mixture of lime and
Another parameter to be varied was the weight lifted by the device, sand, which was ‘denominated excellent mortar for the common pur-
which was equated to the quantity of work done by the windmill. A poses of building’ (Smeaton, 1791, p. 103). He also tested the ‘common
weight placed in the scale (P) could be raised without being affected by composition of Tarras mortar’ as well as varying the quantities of sand
the motion of the arm (Smeaton, 1759, p. 140). Smeaton (1759) also and tarras in the composition. The hardness of the limestone was also
varied the area of sail because, as he noted, ‘[m]any have imagined, that varied:
the more sail, the greater the advantage, and have therefore proposed to
I found it indeed commonly asserted by Masons, that the harder or
fill up the whole area’– this rule he also associated with Parent (p. 149).
stronger the Lime-stone was, the stronger would be the Lime, but
whether this maxim chiefly regarded the usual composition of lime
6. Optimisation via hypothesis testing: hydraulic lime
and sand in common buildings (with which they were chiefly
acquainted) or whether it held good also in Tarras Mortar, did not
Smeaton (1791) relied on the opinions of others even more in his
appear (Smeaton, 1791, p. 103).
research into hydraulic limes, where most of the hypotheses he tested
were taken from ‘the workmen’ and construction practices of the period Smeaton (1791) explored whether saltwater or fresh water should be
(p. 104). Here, Smeaton did not have recourse to instruments. Instead, his used in the mortar: ‘[i]t was also generally agreed by masons, that
experiments were carried out by mixing different ingredients together in mortar, if mix'ed up with salt water, would never harden in so great a
varying proportions, adding water, then rolling the mortar paste into degree, as the same kind of composition would do if made with fresh
balls of about two inches in diameter. These balls were then placed in water’ (p. 103). The geographical sources of the limestone were varied,
water for up to three months and tested for hardness manually (Smeaton, with Smeaton (1791) having visited an area and seen a specific type of
1791, p. 104). Smeaton (1791) also highlighted the fact that he carried limestone, or ‘having heard of a lime produced from a stone found at
out the experiments many times to be sure of his results: ‘To avoid pro- Aberthaw,’ for example (p. 105). The degree to which the mortar was
lixity, I have mentioned these experiments as single trials; but they were beaten was varied, because Smeaton wanted ‘to examine the validity of
all repeated, and some of the principal ones several times’ (p. 110). He the notion entertained by workmen, respecting Tarras Mortar, that the

287
A.M.A. Morris Studies in History and Philosophy of Science 89 (2021) 283–294

Fig. 2. A technical drawing of the workings of Smeaton's waterwheel apparatus from Smeaton (1759, p. 102).

longer it was kept and the oftener it was beaten over, the stronger it working windmills were generally in continuous motion throughout the
would set’ (Smeaton, 1791, p. 105). Smeaton (1791) also varied the type day.
of material used, testing shell lime and plaster of Paris, because he ‘had In his experiments on hydraulic mortars, Smeaton followed the same
heard that Shell Lime […] made an excellent mortar’ (p. 106) and he was approach, establishing which mortar recipes would best resist certain
‘aware that the plasterers use some compositions of this kind [ie. plaster]’ conditions, such as full submersion, alternating wet and dry or
(p. 110). completely dry. He also provided a table of mortar mixtures (Fig. 5)
Smeaton's methodological approach for these experiments was quite ranging from the most expensive kind (to be used in the most weather-
clear: starting from current construction practices, he aimed to find a worn parts of the lighthouse) to cheaper, less resilient mortars for less
mortar that would best withstand the wet conditions on the Eddystone exposed areas (Smeaton, 1791, p. 122).
rock. The workmen's maxims served as hypotheses to be tested, but they The hypothesis-testing part of the experiments adhered to the most
also helped guide Smeaton in choosing which parameters to vary, in basic, traditional view of scientific practice: evaluating a theory by
order to make the experiments feasible. An experiment with an unlimited testing the consequences of that theory. Either the experiment will
number of parameters to vary, or an unlimited number of variations for confirm the hypothesis, or it will disprove it. This kind of theory-testing
each parameter, would take an unlimited amount of time to complete, so vision of scientific activity has a long history in the philosophy of science,
it was necessary to reduce the number of variables to a manageable and has generated pushback from more recent philosophers of science
amount. who wish to counter such a theory-driven vision of science with an
The hypotheses that served to narrow the different parameters experiment-driven vision grounded in actual scientific practices rather
needing to be varied were directly challenged in the course of gathering than rational reconstructions.20
the experimental data. Once a number of parameters had been varied, it The continued focus on theory testing could arguably be attributed
was possible to choose the optimal configuration for the objective in to the distinction between the contexts of discovery and justification,
question. For Vincenti (1990), systematic, experimental parameter vari- according to which the philosophical significance of science was only to
ation is the characteristic methodology that is at the heart of optimisation be found in the way that science justified knowledge, not in the way that
in engineering, although this kind of optimisation has not received much it generated knowledge.21 However, since the last decades of the
attention in the secondary literature (p. 168). For example, although the twentieth century, historically-informed philosophers of science have
undershot waterwheel often had a higher output than the overshot begun to devote more attention to the broader practices, methods and
wheel, the overshot wheel was actually more efficient. So it was up to the aims of science, beyond justification. Despite no longer being consid-
engineer to define optimisation conditions that best suited the project ered to be the quintessential feature of scientific method, hypothesis
that they were working on. Similarly, the straight-sailed windmill was
better gaining motion from rest than other sails, but twisted sails were
more efficient in motion (as we have seen, the outer edges of straight sails 20
This is a common theme in the literature. See Hacking (1983, p. 154),
generated too much lateral air resistance to their circular motion). Early O'Malley (2007), Elliott (2007) and Steinle (2016, p. 317).
experiments would have tested sails accelerating from rest, whereas 21
See Hacking (1983, p. 5) and Schickore and Steinle (2010).

288
A.M.A. Morris Studies in History and Philosophy of Science 89 (2021) 283–294

Fig. 3. A sketch of Smeaton's windmill-testing apparatus from Smeaton (1759, p. 139).

testing is still an important aspect of scientific activity.22 However, by What is interesting about these maxims is that they are action-
looking at both discovery and justification, we gain a much clearer oriented rules of thumb that will be most useful to someone who is
understanding of Smeaton's methodology. Discovery is the topic of the actually constructing the machine in question, such as a waterwheel. In
next section. his paper on waterwheels, Smeaton also made reference to Leonhard
Euler's paper entitled ‘Maxims for arranging most advantageously ma-
7. Generating maxims chines intended to raise water by means of pumps’ (1754) which used
maxims in a similar way.
The third aim of Smeaton's experiments was establishing maxims Maxims are important because they show that Smeaton went beyond
(rules of thumb or empirical rules) that would help guide engineers in merely optimising machines and techniques; he also derived more general
their work. These maxims were generated from the data collected during empirical rules from his experiments. Importantly, however, establishing
the course of his experiments. The term ‘maxim’ gained in popularity maxims was not always simply a matter of reading off the relevant results –
around the middle of the eighteenth century, generally meaning a rule in his paper on waterwheels and windmills, Smeaton used the sub-title
for guiding conduct. It is perhaps best known to English-speaking phi- ‘Observations and deductions from the foregoing experiments’ to intro-
losophers in virtue of Immanuel Kant's usage of the term in his moral duce the sections containing maxims and other more general conclusions
philosophy, although it was also used widely by the French moralists in drawn from his experiments. This sub-title indicates to us that Smeaton was
the previous century.23 aware of the need to carry out supplementary work to provide these
Its use in scientific contexts is most clearly illustrated by the pio- maxims and deductions, either by determining a mathematical relationship
neering French engineer from the first half of the eighteenth century, between two features of the experiment, such as the input and output of a
Bernard Forest de Belidor, whose work was carefully studied by Smeaton. waterwheel, or by carrying out supplementary research to generalise the
According to de Belidor (1737), maxims are rules derived from technical, relationship, which he did in his work on mortars. In other words, it was not
mathematical research that can be understood and used by the layperson possible to simply ‘read off’ the relevant maxims from the data tables.
and/or artisan in their practical projects. These maxims are not only a Thus, we see that the four maxims derived from his experiments on
distillation of the current scientific research, they are also a simplification waterwheels are generalised, mathematical relationships approximately
of that research, aimed at readers who might be unable to understand derived from his results. They are simple rules that will aid an engineer
complex mathematics (de Belidor, 1737, p. 71). who is attempting to construct a waterwheel:

Maxim I. That the virtual or effective head [the virtual height through
22 which the water falls – see footnote] being the same, the effect will be
O'Malley (2007, p. 354) notes that hypothesis-driven research is still fav-
oured when it comes to attracting funding.
nearly as the quantity of water expended.
23 Maxim II. That the expence of water being the same, the effect will be
For a survey of the term in Kant's work, see: Gressis (2010). La Rochefou-
cauld (1665) is a good example of a seventeenth-century moralist's work con- nearly as the height of the virtual or effective head.
taining maxims.

289
A.M.A. Morris Studies in History and Philosophy of Science 89 (2021) 283–294

Fig. 4. The table of results for Smeaton's experiments on windmills from Smeaton (1759, p. 144).

Maxim III. That the quantity of water expended being the same, the disconnected collection of those, but often a coherent system’ (Steinle,
effect is nearly as the square of its velocity. 2002, p. 421).
Maxim 4th. The aperture being the same, the effect will be nearly as Maxims such as these could serve as handy rules of thumb – Smeaton
the cube of the velocity of the water (Smeaton, 1759, pp. 116–120). is explicit about how these are approximations – that would help other
engineers who were tasked with building their own waterwheel or
The following are the first three maxims derived from the windmill windmill.
experiments (out of a total of nine), and they have much the same Smeaton placed much greater emphasis on maxim generation in his
structure as the maxims derived from Smeaton's waterwheel papers published in the Philosophical Transactions, because he was con-
experiments: cerned with providing general, widely applicable rules for the con-
struction of waterwheels and windmills, rather than giving an historical
Maxim I. The velocity of windmill sails, whether unloaded, or loaded account of the process of developing the specific mortar used in the
so as to produce a maximum, is nearly as the velocity of the wind, construction of the Eddystone Lighthouse. As pointed out earlier in the
their shape and position being the fame. paper, this difference in emphasis demonstrates that the exploratory
Maxim 2. The load at the maximum is nearly, but somewhat less than, research carried out by Smeaton was viewed as ‘scientific’ in ways that
testing workmen's maxims and optimisation weren't. For this reason,
as the square of the velocity of the wind, the shape and position of the
sails being the same. Smeaton gave much less attention to maxim generation in his Narrative
on the construction of the Eddystone lighthouse. He did develop one
Maxim 3d. The effects of the same sails at a maximum are nearly, but
maxim, however.
somewhat less than, as the cubes of the velocity of the wind (Smeaton,
By his method of parameter variation, Smeaton was able to determine
1759, pp. 154–155).
that the Aberthaw ‘Blue Lyas’ limestone was a good limestone for his
purposes. However, this conclusion was very limited – in order to
Further, with his maxims relating to windmill sails, Smeaton was able
formulate a more general maxim, Smeaton needed to generalise the
to show that certain maxims actually follow from the combination of
principle behind the suitability of Aberthaw limestone. The workmen's
other maxims. This resonates with Friedrich Steinle's claim that ‘[i]t is
maxim that had been refuted was ‘the harder the limestone, the harder
not just a single regularity that is the result of exploratory work, or a

290
A.M.A. Morris Studies in History and Philosophy of Science 89 (2021) 283–294

Fig. 5. The table of results for Smeaton's experiments on mortars from Smeaton (1791, p. 122).

the lime’, which would have allowed the engineer to select the appro- needing to repeat Smeaton's lengthy experiments, future engineers would
priate limestone based on its hardness. However, since this rule did not only need to dissolve their limestone samples in aqua fortis to determine
apply to hydraulic mortars, Smeaton sought to formulate another, similar whether they would be suitable for hydraulic mortars or not.
rule. By chemically analysing different types of limestone using aqua
fortis (nitric acid), Smeaton established that it was a clay-like residue in 8. Did Smeaton carry out exploratory experiments?
the rock sample that was correlated with the suitability of that kind of
limestone for hydraulic mortars. Smeaton was not a skilled chemist, so he Although Smeaton was working as an engineer, his approach shared
had asked his friend William Cookworthy (1705–1780) to help him many similarities with the early phases of scientific research in new
analyse different limestone types. Cookworthy was a local apothecary fields. This because, in both artisanal engineering practices and pio-
who had trained in London. He is best known today as the first person in neering scientific research, a mature theoretical framework was
Britain to succeed in replicating the Chinese porcelain-making process, conspicuously absent, hence the need for preliminary, exploratory
after discovering that deposits of the necessary ingredients – kaolin and research. In the eighteenth century, engineering had not yet become the
petuntse – could be found in Cornwall (Madge, 1993, p. 3). So Cook- formalized, scientifically informed discipline of the nineteenth century
worthy already had experience in analysing rock samples. and later. Much early engineering practice was based on rules of thumb,
The presence of this clay meant that Smeaton (1791) could replace trial-and-error, and approximation. This gave the discipline a structural
the maxim of ‘the harder the limestone the harder the lime’ with a better similarity with other emerging scientific fields, such as research into
rule: ‘This suggested to me the idea, that an admixture of Clay in the electricity. As we will see in the next section, more recent research has
composition of limestone, when treated as above specified, might be the shown that even established fields of research can have exploratory
most certain index of the validity of a limestone for Aquatic Buildings’ (p. aspects.
108). As the focus in the philosophy of science has shifted away from a
In this way Smeaton could establish rules for selecting the correct theory-based picture of science towards an approach focused on actual
limestone by correlating the results of the chemical analysis with the scientific practices instead of idealised reconstructions, so discussion of
results of his mortar tests. This further step was necessary to provide a broad terms like ‘inductive method’ have given way to more precise
framework for extending his results. So instead of just saying ‘Aberthaw accounts of exploratory practices, which have been grouped under the
limestone is better than chalk for the purpose of hydraulic mortars’, name of exploratory experimentation. Although first associated with the
Smeaton could say that limestone containing clay was better than lime- work of Andre-Marie Ampere (1775–1836) and Michael Faraday
stone without clay, for the purpose of hydraulic mortars. Instead of (1791–1867), exploratory experimentation was not intended to be

291
A.M.A. Morris Studies in History and Philosophy of Science 89 (2021) 283–294

limited to specific historical periods or scientific disciplines.24 Indeed, 315). This can be seen in Smeaton's experiments on hydraulic lime
more recent research has found evidence of exploratory methods in a mortars, whose exploratory nature depended on the engineer's experi-
wide variety of scientific fields and historical periods. mental practice, rather than any specific kind of instrument.
A related feature that connects discussion of Smeaton's method and Steinle (2016) emphasises the role of exploratory experiment in
exploratory experimentation, and that also distinguishes them from the establishing empirical regularities (p. 314). Although Smeaton's experi-
broader focus on inductive method, is the emphasis on the role played by ments were resolutely practical, he wanted to arrive at conclusions with
parameter variation. Parameter variation, which was heavily used by greater generality than merely concluding that this waterwheel is more
Smeaton, is also an important feature of exploratory experimentation. efficient than that one, or this type of limestone is better suited to this
Why is parameter variation so central to exploratory experimentation? particular hydraulic lime. The practical nature of Smeaton's research
Parameter variation is, in essence, a systematic method for tinkering with meant that he focused more on formulating regularities rather than on
an entity, phenomenon or experimental setup. This approach, which the formation of novel concepts (which was another key feature of
bears a close relationship with trial-and-error problem-solving, involves exploratory experimentation underlined by Steinle), although Smeaton's
altering one variable whilst keeping everything else the same. When participation in the debate over vis viva does suggest some conceptual
done systematically, this allows the experimenter to establish empirical, instability was at play in the research into waterwheels (Steinle, 2002, p.
causal relationships, by identifying which parameters influence which 421).
experimental outcomes. To generate empirical rules, Smeaton needed to transform his
A simple example of this kind of variation can be illustrated by early experimental observations into general maxims, which he did using
research into electricity: the experimenter would vary the material either mathematical relationships or chemical properties. By establishing
through which the electricity passed, in order to determine which ma- mathematical relationships between the elements of his maxims, Smea-
terials were most conductive (Steinle, 2002, p. 411). This method was ton made those maxims applicable to many different possible waterwheel
particularly effective when other potentially relevant parameters, such as configurations. And finding a more general way of identifying and
temperature or the size and shape of the wire, were controlled. selecting the best kind of limestone for hydraulic mortars would be useful
Steinle and Elliot have been the most explicit about what might count for other engineers who might not have access to exactly the same kind of
as an exploratory experiment, although, as we will see in the next section, limestone as Smeaton.
this vision has been considerably refined since its first articulation. There Can we conclude that Smeaton's experiments were exploratory?
are two key areas that distinguish exploratory experimentation from Steinle in particular has pointed out that the philosophical focus has
other kinds of experiment: tended to fall on hypothesis testing, because the context of justification
tended to be treated separately from the context of discovery (see
1) The relation to theory Schickore and Steinle, 2010). Steinle (2016) sees his work on exploratory
2) The experimental process experimentation as breaking down this distinction by paying greater
attention to the generation of hypotheses, rather than their evaluation (p.
Concerning the relation to theory, exploratory experimentation is 4).26 This distinction helps us see how Smeaton's experimental work was
often opposed to theory or hypothesis testing.25 This is the most impor- twofold, both evaluating existing maxims and generating new maxims.
tant distinguishing feature of this kind of experimentation. In more Significantly, it is the more exploratory research that was underlined in
recent literature on this topic, there is debate about whether exploratory the paper on waterwheels and windmills published in the Philosophical
experimentation and hypothesis testing are mutually exclusive or Transactions, which indicates that the natural philosophers of the Royal
whether they are merely distinct but complementary. What differentiates Society were interested in engineering research insofar as it was able to
these two kinds of experiment is the open-endedness of the experimental produce regularities or general rules applicable to a variety of cases.
procedure. Does the experiment aim to demonstrate one specific hy-
pothesis, or does it aim to help understand a phenomenon in a broader, 9. Compatibility of exploratory experimentation and hypothesis
less specific way? This aim is reflected in the experimental process. testing
The second major feature of exploratory experimentation is its
distinctive experimental process. In contrast to the highly focused ex- We have seen that, in Smeaton's research, maxim testing and maxim
periments used to test hypotheses, exploratory experimentation involves generation go hand in hand. Does this mean that Smeaton's experimental
a much looser experimental set-up that has the method of parameter approach was not ‘exploratory’? Not according to more recent research
variation at its core (Elliott, 2007, p. 323). The instruments used in on exploratory experimentation. I argue, in line with this research, that
exploratory experimentation are designed to allow for the manipulation although hypothesis testing and exploratory experimentation are meth-
of multiple variables, as opposed to the specificity of theory-testing odologically distinct, they are not mutually exclusive, and are often used
experimental set-ups, which generally aim to pick out only one experi- together. So the fact that Smeaton tested workmen's hypotheses did not pre-
mental outcome (Steinle, 2016, p. 316). vent him from also carrying out exploratory work.
So, a model waterwheel apparatus that was designed to only deter- In this section I provide a brief survey of the recent literature that
mine whether an overshot or undershot wheel was more efficient would makes a case for the compatibility of exploratory experimentation and
have had a much simpler design than Smeaton's waterwheel apparatus, theory testing, as well as comparing Smeaton's method with some of the
which, as we have seen, was designed to provide a wide selection of methods discussed in this literature. We will see that, despite the fact that
variable experimental parameters in order to optimise the waterwheel the rigid philosophical distinction between theory-driven and theory-free
design as much as possible. experiment does not stand up to scrutiny, the recent focus on exploratory
Steinle (2016) points out that not only are the instruments used in experimentation has nevertheless generated insights into experimental
exploratory experimentation unique, but the experimental practices practices with varying degrees of ‘theory-drivenness’ in different scien-
themselves also differ from those of hypothesis-testing experiments (p. tific fields and during different historical periods.27

24 26
Steinle (2016, p. 7) suggests that other episodes from earlier periods might Although Waters (2007, p. 3) warns us not to reduce the distinction between
also be used to illustrate the phenomenon of exploratory experimentation. It is exploratory experimentation and theory-driven experimentation to the distinc-
worth noting that Burian (1997) developed a closely related, but distinct, notion tion between discovery and justification.
27
of exploratory experimentation at around the same time. For a critical approach to the notion of exploratory experimentation and its
25
See Steinle (2016, p. 4) and Elliott (2007, p. 333). role in the philosophy of science, see Schickore (2016, p. 21).

292
A.M.A. Morris Studies in History and Philosophy of Science 89 (2021) 283–294

Maureen O'Malley (2007) suggests that philosophical concerns have help determine which causally relevant parameters to vary in a given
taken precedence in establishing the opposition between exploratory experiment.28
experimentation and hypothesis testing – such as challenging older ‘Why is all this relevant?’ a critical reader might ask. By pursuing the
conceptions of scientific practice or taking a position on the issue of the apparent similarities between exploratory experimentation and one facet
theory-ladenness of experiment (p. 351). O'Malley (2007) emphasises of Smeaton's method, we have found that the insights of more recent
that in practice, scientists use exploratory techniques and hypothesis research on exploratory experimentation by O'Malley, Colaço and
testing together: Karaca, according to which hypothesis testing and exploratory research
are intertwined in specific ways, are confirmed when looking at how
[A]s new concepts and classificatory frameworks are being developed
Smeaton proceeded. As our understanding of scientific practices becomes
to take proteorhodopsin into account, older theoretical frameworks
more nuanced, so too does our understanding of the relationship between
(e.g., about rhodopsin evolution) are being called into service and
science and technology.
examined for their adequacy in relation to this new knowledge. In
We have seen that Smeaton made use of different scientific methods and
other words, hypothesis testing is brought in as part of the exploratory
practices, and this in turn tells us about the specific ways that Smeaton made
approach (p. 350).
his technological research ‘scientific’. Not only did he integrate the tradi-
O'Malley (2007) proposes that providing a more nuanced account of tional scientific approach of hypothesis testing into his experimental prac-
the relationship between exploratory experimentation and hypothesis tice, he also carried out the more open-ended work of gathering data and
testing will be a potential avenue of future research (p. 355). One could finding regularities in the results of his experiments – this with the aim of
argue that it is obvious that scientists throughout history used experi- generating practical maxims. In methodological terms, by using hypothesis
ment to guide their research and not just to test their hypotheses. The aim testing and maxim generation in his practical experimentation, Smeaton
of the literature discussed in this section is to give an account of how these actually reduced the distinction between science and technology to a
different roles for experiment fit together. distinction between two different kinds of subject matter and two different
David Colaço (2018), using the notion of ‘auxiliary hypothesis,’ overall goals (action-oriented vs epistemic), both of which could be treated
shows how exploratory experiments can be directed by theories without using very similar methodological approaches.
necessarily testing those theories (p. 6). This muddies the supposedly Put simply, the fact that we can find the same methods used in sci-
clear distinction between exploratory and theory-driven experimenta- entific practices as diverse as nineteenth-century electrical research,
tion, and supports the claim already advanced by O'Malley, that actual twentieth-century particle accelerator experiments or Smeaton's wind-
scientific practice – in contrast to idealised philosophical reconstructions mill research, shows that, contrary to the classic view of Smeaton's
– usually involves a mix of exploratory and theory-driven methodology, his experimental research was distinctly scientific, being
experimentation. distinguishable from other scientific research only by the subject matter
But Koray Karaca (2017) comes closest to articulating the method and aims at optimisation. Although the intuition at the heart of the classic
employed by Smeaton, making the case for experiments that ‘can have view – that Smeaton systematised craft practices – is correct, we have
both exploratory and non-exploratory components that, respectively, seen that it is this very systematisation that makes Smeaton's method-
consist of exploratory and non-exploratory procedures and that ology scientific, because it leads him to propose and evaluate widely
contribute to the objectives of the experiment in different and comple- applicable, general maxims, instead of merely improving machines or
mentary ways’ (p. 340). Karaca (2017) wants to move away from the processes in an ad hoc manner.
view according to which entire research programs must be classed as
either exploratory or theory-driven, instead suggesting that experiments 10. Conclusions
can have both exploratory and non-exploratory elements.
As we have seen, the role of hypothesis testing is to give focus and The analysis of Smeaton's methodology in his research into water-
direction to Smeaton's experiments, which would risk becoming un- wheels, windmills and hydraulic mortars has revealed that he did more than
manageable otherwise. The selection of relevant parameters to vary merely systematise craft practices. Not only did Smeaton carry out what was
resulted in a narrowing of the possible outcomes of this research. How- once viewed as the key feature of scientific activity – hypothesis testing – he
ever, it is worth noting that this narrowing was only relative – on Karaca's also established empirical regularities via a procedure that has come to be
account, experimentation gains an exploratory aspect when there is a known as exploratory experimentation. As our understanding of the role of
broadening of the parameters varied, resulting in the potential results of experimentation in scientific practice has broadened over the past few de-
those experiments being much more open-ended, whereas ‘[t]he non- cades, this has opened up the possibility of re-evaluating the place in the
exploratory component of an experiment can involve procedures that history of science of engineers and technologists like Smeaton.
are aimed at narrowing down the scope of the experimental research to Although Smeaton was an excellent technician, I have made the case
the investigation of specific research questions’ (Karaca, 2017, p. 340). in this paper that it is no longer plausible to claim that he did little more
Smeaton's selection of parameters to vary, based on challenging the than systematised trial-and-error or organised, yet relatively mindless
maxims of the workmen, constituted the non-exploratory side of the tinkering, as the classic view of his methodology suggests. He carried out
experiments. Smeaton treated these maxims as hypotheses to be tested, a program of rigorous testing of theoretical claims, workmen's maxims
which fits Karaca's account of how non-exploratory components of an and contemporary construction practices concerning waterwheel and
experiment can be theory-driven (Karaca, 2017, p. 340). windmill functioning, and the mixing of hydraulic mortars. He also used
This is also where the issue of background knowledge comes into his results to formulate his own maxims, based on the empirical regu-
play. Although eighteenth-century engineers may not have been directly larities that were produced during his experiments. This is particularly
applying natural philosophy in their work, they did possess a wide evident in his celebrated paper on waterwheels and windmills published
background knowledge of natural philosophy which informed that work in the Royal Society's Philosophical Transactions, and directly contradicts
indirectly (see, for example, Miller, 2019, p. 157). This more nuanced Cardwell's claim, cited earlier, that Smeaton's work did not involve ‘sci-
position avoids the strong claims, either that technological developments entific research into the basic laws of the subject or system under
in the Industrial Revolution were not influenced by natural philosophy investigation’ (Cardwell, 1994, p. 195).
(Hall, 1974), or that the growth of eighteenth-century engineering was a
direct result of Newtonian principles being applied to practical mechanics
28
(Dobbs and Jacob, 1995). Instead, we see that familiarity with natural The role of background knowledge, in contrast to ‘local theories’, is debated
philosophy gave engineers the background knowledge they needed to in the literature on exploratory experimentation. I will not enter into the details
of this discussion here. See Colaço (2018, pp. 38–39).

293
A.M.A. Morris Studies in History and Philosophy of Science 89 (2021) 283–294

Having seen how Smeaton used methods drawn from both technology Jacob, M. C., & Stewart, L. (2004). Practical matter. Newton's science in the service of industry
and empire 1687-1851. Harvard University Press.
(optimisation) and science (hypothesis testing, exploratory experimen-
Jones, P. M. (2008). Industrial enlightenment. Science, technology and culture in Birmingham
tation), my hope is that this study can not only contribute to viewing and the West Midlands 1760-1820. Manchester University Press.
Smeaton's methodology in a new light, but that it might also constitute a Karaca, K. (2017). A case study in experimental exploration: Exploratory data selection at
step towards a broader reappraisal of the relation between science and the Large Hadron Collider. Synthese, 194(2), 333–354. https://doi.org/10.1007/
s11229-016-1206-x
technology at the beginning of the Industrial Revolution. Klein, U. (2017). Hybrid experts. In M. Valleriani (Ed.), The structures of practical
knowledge (pp. 287–306). Springer.
Madge, A. G. M. (1993). Cookworthy—chemist and potter. Pharmaceutical Historian,
Funding information 23(2), 3–4.
Miller, D. P. (2019). The life and legend of James Watt. Collaboration, natural philosophy, and
This work has been funded with a doctoral fellowship from the FWO – the improvement of the steam engine. University of Pittsburgh Press.
Mokyr, J. (2002). The gifts of Athena. Historical origins of the knowledge economy. Princeton
Research Foundation Flanders, grant number 148531. University Press.
Morris, A. M. A. (2018). John Smeaton and the vis viva controversy: Measuring
waterwheel efficiency and the influence of industry on practical mechanics in Britain
Acknowledgements 1759–1808. History of Science, 56(2), 196–223. https://doi.org/10.1177/
0073275317745455
I would like to thank my doctoral supervisor, Steffen Ducheyne, for Morris, A. M. A. (2021). The joint labours of ingenious men. John Smeaton’s Royal
Society network and the Eddystone lighthouse. Centaurus, 63(3), 513–531. https://
his help during the writing of this paper. Various versions have been
doi.org/10.1111/1600-0498.12398
presented at the Centre for Logic and Philosophy of Science at the Vrije Musson, A. E. (1975). Review: Science and technology in the Industrial Revolution.
Universiteit Brussel, and I would like to thank all those present for their Reviewed work(s): Historical perspectives: Studies in English thought and society by
Neil McKendrick. Minerva, 13(4), 633–637.
valuable comments. I have regularly solicited the advice of my colleagues
Musson, A. E., & Robinson, E. (1969). Science and technology in the Industrial Revolution.
Jip van Besouw, Pieter Present and Frederik Dhondt, and I am very Manchester University Press.
grateful for the time and effort they put in. I would also like to thank the O'Malley, M. A. (2007). Exploratory experimentation and scientific practice:
anonymous referees for their comments and suggestions. Metagenomics and the proteorhodopsin case. History & Philosophy of the Life Sciences,
29(3), 337–360.
Pitot, H. (1729). Remarques sur les aubes ou pallettes des moulins, & autres machines
References mûes par le courant des rivieres. In Histoire de l'Academie Royale des Sciences (pp.
359–366). Pierre Mortier.
Radder, H. (2003). The philosophy of scientific experimentation. University of Pittsburgh
Ackermann, R. (1989). The new experimentalism. The British Journal for the Philosophy of Press.
Science, 40(2), 185–190. https://doi.org/10.1093/bjps/40.2.185 Radder, H. (2009). Science, technology and the science-technology relationship. In
Anderson, J. D. (1997). A history of aerodynamics. Cambridge University Press. A. Meijers (Ed.), Philosophy of technology and engineering sciences (pp. 65–91). Elsevier
Baird, D. (2004). Thing knowledge. A philosophy of scientific instruments. University of North Holland.
California Press. Reynolds, T. S. (1983). Stronger than a hundred men. A history of the vertical waterwheel. The
Bertoloni Meli, D. (2006). Thinking with objects: The transformation of mechanics in the Johns Hopkins University Press.
seventeenth century. The Johns Hopkins University Press. Roberts, L., Schaffer, S., & Dear, P. (Eds.). (2007). The mindful hand. Inquiry and invention
Burian, R. M. (1997). Exploratory experimentation and the role of histochemical from the late Renaissance to early industrialisation. Royal Netherlands Academy of Arts
techniques in the work of Jean Brachet, 1938-1952. History & Philosophy of the Life and Sciences.
Sciences, 19(1), 27–45. Schickore, J. (2016). ‘Exploratory experimentation’ as a probe into the relation between
Capecchi, D. (2021). Epistemology and natural philosophy in the 18th century: The roots of historiography and philosophy of science. Studies in History and Philosophy of Science,
modern physics. Springer International Publishing. https://doi.org/10.1007/978-3- 55, 20–26. https://doi.org/10.1016/j.shpsa.2015.08.007
030-52852-2 Schickore, J., & Steinle, F. (Eds.). (2010). Revisiting discovery and justification. Springer.
Cardwell, D. (1994). The Fontana history of technology. Fontana Press. Shapin, S., & Schaffer, S. (1985). Leviathan and the air-pump. Hobbes, Boyle and the
Channell, D. F. (2009). The emergence of engineering science: An historical analysis. In experimental life. Princeton University Press.
A. Meijers (Ed.), Philosophy of technology and engineering sciences (pp. 117–154). Smeaton, J. (1759). An experimental enquiry concerning the natural powers of water and
Elsevier. wind to turn mills, and other machines, depending on a circular motion. Philosophical
Channell, D. F. (2019). The emergence of engineering science. In D. F. Channell (Ed.), The Transactions of the Royal Society, 51, 100–174. https://doi.org/10.1098/
rise of engineering science (Vol. 35, pp. 75–109). Springer International Publishing. rstl.1759.0019
https://doi.org/10.1007/978-3-319-95606-0_5. Smeaton, J. (1776). An experimental examination of the quantity and proportion of
Colaço, D. (2018). Rethinking the role of theory in exploratory experimentation. Biology mechanic power necessary to be employed in giving different degrees of velocity to
and Philosophy, 33(5–6). https://doi.org/10.1007/s10539-018-9648-9 heavy bodies from a state of rest. Philosophical Transactions of the Royal Society, 66,
Collins, H. (1992). Changing order. Replication and induction in scientific practice. University 450–475. https://doi.org/10.1098/rstl.1776.0029
of Chicago Press. Smeaton, J. (1782). New fundamental experiments upon the collision of bodies.
de Belidor, B. F. (1737). Architecture hydraulique, ou L’art de conduire, d’elever et de menager Philosophical Transactions of the Royal Society, 72, 337–354. https://doi.org/10.1098/
les eaux pour les differens besoins de la vie. Premiere partie. Tome premier. Charles- rstl.1782.0023
Antoine-Jombert. Smeaton, J. (1791). A narrative of the building and a description of the construction of the
de La Rochefoucauld, F. (1665). Reflections ou sentences et maximes morales. Claude Barbin. Edystone Lighthouse with stone. H. Hughs.
Dobbs, B. J. T., & Jacob, M. C. (1995). Newton and the culture of Newtonianism. Humanity Smeaton, J. (1938). John Smeaton's diary of his journey to the low countries 1755. The
Books. Newcomen Society.
Elliott, K. C. (2007). Varieties of exploratory experimentation in nanotoxicology. History Soler, L. (Ed.). (2014). Science after the practice turn in the philosophy, history, and social
& Philosophy of the Life Sciences, 29(3), 313–336. studies of science. Routledge.
Euler, L. (1754). Maximes pour arranger le plus avantageusement les machines destinees Steinle, F. (2002). Experiments in history and philosophy of science. Perspectives on

a elever de l’eau par le moyen des pompes. In Memoires de l’academie des sciences de Science, 10(4), 408–432. https://doi.org/10.1162/106361402322288048
Berlin (pp. 185–232). L'Imprimerie Royale. Steinle, F. (2016). Exploratory experiments. Ampere, Faraday, and the origins of
Gooding, D., Pinch, T., & Schaffer, S. (Eds.). (1989). The uses of experiment. Studies in the electrodynamics (A. Levine, trans.). University of Pittsburgh Press.
natural sciences. Cambridge University Press. Stewart, L. (1992). The rise of public science: Rhetoric, technology, and natural philosophy in
Gressis, R. (2010). Recent work on Kantian maxims I: Established approaches. Philosophy Newtonian Britain (pp. 1660–1750). Cambridge University Press.
Compass, 5(3), 216–227. https://doi.org/10.1111/j.1747-9991.2009.00254.x Stewart, L. (2007). Experimental spaces and the knowledge economy. History of Science,
Hacking, I. (1983). Representing and intervening. Cambridge University Press. 45(2), 155–177. https://doi.org/10.1177/007327530704500203
Hall, A. R. (1974). What did the Industrial Revolution in Britain owe to science? In Turner, T., & Skempton, A. W. (1981). John Smeaton. In A. W. Skempton (Ed.), John
N. McKendrick (Ed.), Historical perspectives: Studies in English thought and society. Smeaton, FRS (pp. 7–34). Thomas Telford Limited.
Europa Publications. Vincenti, W. G. (1990). What engineers know and how they know it. In Analytical studies from
Hansson, S. O. (2015). The role of technology in science: Philosophical perspectives. Springer. aeronautical history. The Johns Hopkins University Press.
Jacob, M. C. (2007). Mechanical science on the factory floor: The early Industrial Waters, C. K. (2007). The nature and context of exploratory experimentation. Philosophy
Revolution in Leeds. History of Science, 45(2), 197–221. https://doi.org/10.1177/ of the Life Sciences, 9, 275–284.
007327530704500206

294

You might also like