Marius Stan • Christopher Smeenk
Editors
Theory, Evidence, Data:
Themes from George
E. Smith
Editors
Marius Stan
Boston College
Boston, MA, USA
Christopher Smeenk
Western University
Ontario, ON, Canada
ISSN 0068-0346
ISSN 2214-7942 (electronic)
Boston Studies in the Philosophy and History of Science
ISBN 978-3-031-41040-6
ISBN 978-3-031-41041-3 (eBook)
https://doi.org/10.1007/978-3-031-41041-3
© The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland
AG 2023
This work is subject to copyright. All rights are solely and exclusively licensed by the Publisher, whether
the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse
of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and
transmission or information storage and retrieval, electronic adaptation, computer software, or by similar
or dissimilar methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication
does not imply, even in the absence of a specific statement, that such names are exempt from the relevant
protective laws and regulations and therefore free for general use.
The publisher, the authors, and the editors are safe to assume that the advice and information in this book
are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or
the editors give a warranty, expressed or implied, with respect to the material contained herein or for any
errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional
claims in published maps and institutional affiliations.
This Springer imprint is published by the registered company Springer Nature Switzerland AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
Paper in this product is recyclable.
Introduction
This volume grew out of the conference On the Question of Evidence: A Celebration
of the Work of George E. Smith, which met at Tufts University in May 2018. It
was a venue for presentations by Jody Azzouni, Katherine Brading, Robert DiSalle,
Allan Franklin, Michael Friedman, Bill Harper, Teru Miyake, Eric Schliesser, and
Chris Smeenk. A number of those presentations evolved into chapters included here.
To them, we have added chapters by other scholars and thinkers who have been
influenced, directly or indirectly, by George’s work. Collectively, these pieces aim
to honor that work by engaging with prominent themes from it.
In what follows, we begin with a brief introduction to George’s philosophy of
science, and to his more prominent and distinctive contributions to doing it. Thereby,
we highlight some of the challenges this poses to long-held commitments within
philosophy, history, and sociology of sciences. Then we elaborate on key themes
in his philosophy as we summarize and introduce the chapters in this volume. We
conclude with some remarks on George as a teacher, and on the philosophical impact
of his teaching.1
George Smith on Evidence in Mature Science
The history of evidential reasoning, if viewed over extended periods rather than in
isolated episodes, and regarded as more than illustrations to support philosophical
positions, could produce a decisive transformation in our understanding of scientific
knowledge. Current views of the nature of scientific knowledge have been forged,
in large part, in response to the threat of radical theory change made prominent
first by Kuhn, then by subsequent historiography of scientific practice. To fend off
that threat, any claim to have achieved a distinctive form of permanent knowledge
1 This
volume began as a project jointly co-edited by Smeenk, Schliesser, and Stan. An encounter
with long Covid led to Eric Schliesser being unable to remain at the helm as co-editor.
v
vi
Introduction
must establish some form of continuity through transitions, such as from Newtonian
gravity to general relativity and from classical mechanics to quantum theory.
Many scholars have concluded that claims to establish secure knowledge should
be dismissed as a misconception, reflecting a false image perpetuated by science
pedagogy, for example, rather than genuine insights into what science has achieved.
In addition, the program to find a single logic of scientific method or scientific
inference seems to have failed—in virtue of accumulating evidence showing that
actual scientists are often opportunistic in their practices. In so far as such projects
still remain (e.g., Bayesianism or various kinds of causal modeling), they have a
highly normative character.
Yet such a dismissive response to claims about the epistemic status of many
sciences fails to account for evident progress in understanding the natural world. It is
hard to deny that various scientific fields have succeeded in establishing theoretical
claims that, in many respects, have also served as effective guides for action,
even for those skeptical about whether the theories themselves have any claim to
permanence.
George Smith has often described his work as inspired by asking how we came
to have high quality evidence in any field. In particular, how have scientists in
successful fields turned data from observations or experiments into strong evidence
for substantive claims? What is the nature of the knowledge claims that have the
best case for being firmly established? And in what sense, if at all, can we take
these claims as permanent, stable contributions to scientific knowledge? In pursuing
these questions he has developed a striking, distinctive account of the nature,
scope, and limits of scientific knowledge. He based his account on historical and
philosophical investigations of exemplary evidential reasoning spanning extended
lines of research (over decades or centuries) in physics.
George considers extended lines of research for several related reasons (see the
paper reprinted in this volume). He largely agrees with historians and sociologists
who have argued that the scientific community typically accepts substantive claims
based on surprisingly weak evidence. But he reframes the question of high quality
evidence as more appropriately posed with regard to the consequences of accepting
some claims as a first step in a line of inquiry. Rather than, for example, considering
only the case Newton could make in favor of gravity at the time of writing the
Principia, we should keep in mind the case Simon Newcomb could have made
two centuries later. George strikingly disagrees with historians who are tempted
to generalize from their apt criticisms of local case histories, often at the advent of a
line of inquiry, to a general claim that would apply to Newcomb as well as Newton.
His approach also contrasts with philosophers of science, who have too often
focused exclusively on the kinds of arguments that scientists make in motivating the
community to pursue new ideas—such as an emphasis on novel predictions. The
strongest evidence in favor of a well-established theory often differs from the kind
of case that could have been made for it when it was first introduced; or even when
it found widespread acceptance. More importantly, what is epistemically distinctive
about science, what enables it to make progress, will be more clearly discerned in
how a line of research unfolds subsequently rather than in the initial debates.
Introduction
vii
This position leads to a distinctive innovation in how George practices philosophy of science—an innovation that has led to a new genre of philosophical
scholarship. Many of his studies are working papers or case studies toward the
development of what one might call a longue durée review article. They can be
book-length pieces, as his famous “Closing the Loop” attests. In such a longue durée
review article, George surveys and evaluates how theory development and a field’s
evidential practices (which are constitutive of the field) interacted over an extended
period of time.
The shift to a long view is only useful in concert with Smith’s characteristically
thorough assessment of what is actually being put to the test through subsequent
research. This means that in practice George often redoes the calculations—and
pays careful attention to attempts to replicate earlier results, as well as to the way
that scientists assess the evidence in their own time as presented in textbooks and
ordinary review articles. This practice is manifest in both his case studies and
the more substantial, retrospective review articles. He redoes calculations not just
from meticulous care to understand how past scientists would have evaluated the
evidence, but, in particular, so as to enable his readers to assess which results
or measurements were constitutive or evidentially relevant for particular lines of
inquiry.
By contrast, philosophers of science too often rely on an abstract, hypotheticodeductive account of theory-testing: scientists test a theory T by deducing its
consequences, then checking these via observation or experiment. According to
this view, Newtonian gravity, for example, is tested by deducing predictions for
planetary positions given some information about the initial conditions. One of
the leitmotifs of Smith’s work is that an exclusive focus on successful predictions
neglects crucial aspects of scientific practice. Treating theories as a monolith
obscures the fact that evidence bears differently on specific aspects of the theory.
George has emphasized, for example, the stark contrast between ample evidence
for the inverse-square variation with distance of the gravitational force, as opposed
to the weak evidence for its dependence on the mass of both interacting bodies.
The hypothetico-deductive account also misses the role of a different type of claim:
in the case of planetary astronomy, calculations of planetary positions are based
on the substantive assumption that the masses and forces taken into account in the
derivation are complete. But if the comparison with observations requires a claim of
this kind, what can we say is actually being tested?
Paying attention to the role that theories play in guiding an extended line of
inquiry leads to a richer understanding of the logic of theory-testing, developed
in detail for the case of celestial mechanics in George’s magisterial “Closing the
Loop.” Based on surveying nearly 250 years of planetary astronomy, he argues
that the main question being pursued was not, as philosophers would have it,
whether Newtonian gravity “saves the phenomena.” Indeed, throughout the period,
he considers there was only a brief time during which the existing models fully
fit with available observations of planetary motion. At all other times, there were
systematic discrepancies between the calculated and observed motions. Celestial
viii
Introduction
mechanics used these discrepancies to discover further physical features of the solar
system.
The main aim of this line of research was to identify robust physical sources for
existing discrepancies—such as new planets, but also subtle physical effects such
as the changing rotational speed of the Earth—and then “close the loop,” by adding
this new feature to a more complete model. George once put it as follows:
Newton’s Principia forced the test question within orbital astronomy for his theory of
gravity to be not whether calculated locations of planets and their satellites agree with observations; but whether robust physical sources can be found for each systematic discrepancy
between those calculations and observation—with the further demand of achieving closer
and closer agreement with observation in a series of successive approximations in which
more and more details of our solar system that make a difference become identified, along
with the differences they make.2
Pursuit of this iterative approach, adding further details to develop ever more
sophisticated models and identifying even more subtle features of the solar system,
lead to an enormously rich picture of, as Smith puts it, the configurational details
of the solar system and the differences they make to observed motions. The
identification of what George calls “second order phenomena” depends on a contrast
between the motions, as described at a specific stage of inquiry, and increasingly
precise and carefully targeted observations. Success at each stage in discovering
more subtle effects that have clear physical sources indirectly confirms the earlier
steps in the process. A discrepancy at any given stage only has physical significance,
and helps to identify another detail that contributes to observed motions, if the
earlier calculations have incorporated the features with larger impacts accurately.
What makes “Closing the Loop” so compelling is George’s masterful history of
this field, combined with the philosophical and technical acuity needed to recognize
and articulate the logic of theory-testing that this history discloses. Part of his
insightful analysis pertains to the specifics of celestial mechanics. And George often
insists, in print and in person, that particular evidential strategies may not generalize
to other fields; that one first has to do the work in reconstructing the longue durée
evidential practices of them. This reluctance stems from George’s recognition that
each field faces often highly specific challenges to developing evidence, regarding—
to put it roughly—accessibility to the kinds of quantities that allow for fruitful
theorizing. This challenge is general, but giving a full account of how to respond
to it quickly becomes intricate and domain specific.
Before we articulate (despite George’s reticence) some general features of his
approach to evidential reasoning, it is worth noting that his detailed work also
corrects another tendency common among historians of philosophy. While we have
emphasized above that his work can be understood as incorporating and accepting
the findings of those focused on scientific practice at a given time, but by shifting
toward extended periods (also extended in space) undermines their nominalist
2 This is the first of three principal conclusions stated in the manuscript version of “Closing the
Loop,” which do not appear in the published text.
Introduction
ix
and skeptical tendencies, it is also worth noting that—while George reiterates
the emphasis on theories familiar from an older historiography, e.g., Koyré—he
undermines the older view that theories alone can explain important scientific
debates. His work shows that often attitudes toward particular lines of empirical
evidence shape how theories are adhered to.
While George is cautious about generalizing from the results of his life-time
of research, we think there are also general features of his approach to evidential
reasoning, revealed in other detailed case studies—such as his recent monograph
(with Seth) on early-twentieth century experimental investigations of molecular
reality. Here we will highlight two interrelated aspects of Smith’s account, before
considering their implications for questions of continuity.
First, physical theories typically re-describe phenomena by introducing novel
quantities, such as mass and force in Newtonian mechanics, and law-like regularities
that hold among them.3 This immediately raises a challenge: how do scientists
reliably gain access to the proposed fundamental quantities, and justify their use?
George’s work shows the value of taking these questions—rather than obtaining
successful predictions—as the central challenge. Scientists usually proceed by
exploiting functional relationships between the target quantity of interest and a
“proxy” quantity (or quantities) that can be measured (more) directly with high
precision. Experimental design focuses on finding measurable proxies with a particularly clear and reliable connection to the target quantities. The role of theory is to
establish such functional relationships, to enable theory-mediated measurements of
the target quantities—that is, to show how the theoretical quantities are manifest in,
and can be constrained by, observable phenomena, and can give rise to second-order
phenomena. But, crucially, success in theory-mediated measurements provides
evidence for the aspects of theory used to derive the functional relationship.
This success can take several forms: stability in the outcomes of repeated measurements of the same type; convergence in the values determined by measurements
of different types, based on different functional relationships; and amenability to
increasing precision. Smith and Seth (2020) clarify how each of these types of
success contribute to justifying the use of novel physical quantities and make
a persuasive case that success in all three senses can eliminate the possibility
that the target quantity is merely an artifact or useful fiction. On their account,
early twentieth century physicists established reliable access to the microphysical
realm (of atoms and molecules) through stable, convergent measurements of three
crucial parameters: Avogadro’s number, N0 ; the fine structure constant, α; and
the charge of the electron, e. Contemporary physicists now rely on collaborations
such as the CODATA group to determine consensus best values for dozens of
fundamental parameters. Rather than seeing these as merely values to be inserted
to facilitate comparisons with observations, on George’s view the stable, consistent
determinations of these parameters over time provides evidence for the array
3 A note on our terminology: by “phenomena” we mean roughly what Newton meant by it: a robust
empirical regularity accepted by a scientific or expert community.
x
Introduction
of physical theories that fix the relevant functional relationships used in these
measurements.
To describe the second aspect, we can turn to a feature of Newton’s methodology
that George has elucidated wonderfully: namely, his response to the challenge of
drawing conclusions from phenomena despite their complexity. Smith and Harper
(1995) aptly characterized this approach as a “new way of inquiry” in their
seminal paper, drawing a sharp contrast between Newton and seventeenth century
contemporaries such as Galileo and Huygens. Consider Newton’s realization, as he
was revising the De Motu drafts that would lead to the Principia, that the planets do
not travel along closed orbits. Due to the mutual gravitational interaction among the
planets and the sun, they would instead move around a common center of gravity
and follow trajectories too complex to admit of a simple geometric description.
Rather than give up on the project of determining the forces from observed motion in
light of this remarkable complexity, Newton developed the mathematical framework
needed to make inferences regarding the force that hold even if the orbits can be
described as only approximately elliptical. The first aspect of Newton’s response to
complexity is to establish functional relationships between observations and target
quantities that are robust, in the sense that an approximate description—thus not
necessarily exact—of the observed phenomena can still yield an approximate value
of the target quantity. Borrowing a Latin phrase that Newton used frequently, George
often describes this in terms of “if quam proxime (very nearly) then quam proxime’
reasoning.” In the context of reasoning from observed phenomena of motion to
forces responsible for them, Newton proves that, if the (observed) antecedent
holds very nearly (quam proxime), then the (inferred) consequent also holds quam
proxime.4 (We will discuss an example of this type of result momentarily.)
This is coupled with a recognition that a specific type of idealized representations
play an essential role in guiding further inquiry. Astronomers in the 1670s faced an
underdetermination problem: there were several distinct—and inequivalent—ways
of representing planetary motions. Newton recognized that the planets only follow
approximately Keplerian orbits for theoretical reasons, but in the case of the lunar
orbit the departure from Kepler’s area law was well-established observationally.
What then is the status of the Keplerian orbits? As George puts it:
The complexity of true motions was always going to leave room for competing theories
if only because the true motions were always going to be beyond precise description, and
hence there could always be multiple theories agreeing with observation to any given level
of approximation. On my reading, the Principia is one sustained response to this evidence
problem.
Newton recognized that the Keplerian orbits (unlike the alternatives) hold in
idealized, counterfactual circumstances: if the sun were at rest, and the interactions
4 However, we must note that Newton employs his phrase, quam proxime, in several semantic
regimes, not just one. As a matter of fact, a closer study of these regimes is part of George’s
current research on approximation in physical theory construction. Relatedly, see also the chapter
by Guicciardini in this volume.
Introduction
xi
among the planets neglected, then a planet would follow a Keplerian orbit. They thus
have a physical status that alternative descriptions lack. Clarity regarding exactly
what must be the case for Keplerian orbits to hold—counterfactually—also makes
it possible to treat departures from Keplerian motion as second-order phenomena: as
a means to identify further physical details relevant to orbital motions. In the case of
the moon, despite enormous challenges to developing a satisfactory theory, Newton
could make the case that the departures from Keplerian motion were due to the
perturbing effects of the sun. More generally, idealizations stating regularities that
would be exact in precisely specifiable circumstances support the identification of
discrepancies. In the best case, the character of these discrepancies further indicate
which assumptions of the idealization have to be modified or relaxed.
Although we have focused on George’s account of Newtonian methodology to
elucidate this approach to complexity, we expect an account along these lines holds
for much of modern physics, where it is constrained by robust background theory.
His own work highlights how similar considerations apply to the study of molecular
reality, and to the structure of molecules (see Smith and Seth 2020 and Smith and
Miyake 2021, respectively). More broadly, there is a striking similarity between this
account of methodology and the effective field theory approach in modern particle
physics, developed by Ken Wilson and others.
Taken as a whole, George’s account of methodology leads to a fundamental
reframing and response to concerns about permanence and continuity stemming
from Kuhn. Even profound transitions at the fundamental theoretical level do not
necessarily generate discontinuities in evidential reasoning.5 Following George’s
insights, the focus should be on whether the functional relationships presupposed
in evidential reasoning continue to hold in light of a new theory—for example,
as an approximation within a limited domain. This has a family resemblance to
a kind of structural realism that his friend, Howard Stein, has espoused, or the
real patterns that his long-time Tufts colleague, Dan Dennett, has defended. But,
in George’s work, the emphasis is on the challenges to sustaining a particular line
of evidential reasoning—not on ontology. In one sense, this is a much weaker
demand than what philosophers typically consider, and one that is often imposed by
scientists developing new theories, but in another sense it is much more ambitious: a
record of success in obtaining stable, convergent measurements, and of identifying
further details that make a difference in an iterative approach, in fact make a
compelling case that we should demand continuity of evidential reasoning, just as
Einstein took recovery of a Newtonian limit as a criteria of adequacy in formulating
general relativity. Finally, even though the evidential reasoning itself presupposes
and depends directly on (some aspects of) theory, the standards of success do not.
Whether a line of research has led to increasingly precise, convergent measurements
of fundamental parameters, or continued to identify robust physical sources that can
5 This is not to say that Smith rejects the possibility of Kuhnian discontinuities in evidence;
Buchwald and Smith (2002) consider the discontinuity in the treatment of polarization through
the transition from ray to wave optics.
xii
Introduction
be checked by a variety of independent means, can be evaluated in relatively theoryneutral terms.
We hope that this brief précis of one aspect of George’s thought is sufficient to
show that it has deep ramifications for central debates in history and philosophy of
science. But, we also want to highlight a different aspect of his work. As we noted,
he approaches historical material with a meticulous attention to detail—perhaps
inspired by his working practice as an engineer interested in failure of complex
systems (in particular, turbine engines)—often leading to profound reassessments
of canonical texts, or lines of argument, that one would expect to have few surprises
left to divulge. Several philosophers have, for example, taken Perrin’s case in favor
of atomism based on his own experimental measurements of Brownian motion, in
conjunction with an array of other results, as exemplifying a successful argument
in favor of unobservable entities (while disagreeing, of course, about the precise
nature of the argument or the correct conclusions to draw from it). Smith and Seth
(2020) show that many aspects of the philosopher’s lore about this case, even those
that are particularly relevant to the case for atomism, are simply incorrect. For
example, Perrin’s own measurements of N0 were rejected within a decade after the
publication of Atomes, due to a substantial and persistent discrepancy with other
precision measurements, threatening philosophical arguments that rely uncritically
on Perrin’s claim of sufficient agreement among the different measurements.
George’s most influential contributions along these lines have led to a dramatic
re-evaluation of Newton’s achievements and the reception of Newton’s work. Smith
and Schliesser (2000) give an account of the early response to Newtonian gravity
that corrects lore about the debate between Newton and his continental critics, in
particular Huygens and Leibniz. Those debates are not best understood as merely
a clash of metaphysical systems, but as driven, in large part, by the limitations
of available empirical evidence. In addition, George’s critical reading of Newton
has revealed aspects of the argumentative structure of the Principia that seem to
have escaped notice from Newton’s own time. Our favorite example concerns an
argument that Newton did not make, even though it is often attributed to him:
namely, that the inverse-square law of gravity follows from an observed Keplerian
ellipse. Even a relatively cursory reading of the Principia reveals that Newton did
not actually give this argument, but Smith gives an extremely insightful analysis
of the argument he actually gave and what it reveals about Newton’s methodology.
The incorrect argument requires that the antecedent holds exactly: if a body moves
on an ellipse with a force directed at one focus, then the force varies as the inverse
square of the distance. But Newton also proved that if a body moves on an ellipse
with the force directed at the center, then the force varies directly with distance. The
planets all move on nearly circular orbits, so it is extremely challenging to draw a
distinction between these two cases observationally. Any method that depends on
establishing exact results observationally is extraordinarily fragile. By contrast, the
argument Newton actually gives relies on a functional relationship that is robust.
The apsidal precession theorem establishes a functional relationship between the
motion of the apsides (the points on the orbit closest and farthest from the sun) with
each orbit and the exponent of the force law (for approximately circular orbits).
Introduction
xiii
For a closed orbit, with no motion of the apsides, the force varies with the inverse
square of distance. But, crucially, this result also implies—and this is an instance of
“if quam proxime then quam proxime” reasoning noted above—that for orbits with
nearly stable apsides the force law is nearly inverse-square.
Themes from George Smith: A Synopsis
George’s work has made an impact on epistemology, the history of philosophy
of science, and Newtonian scholarship. In casting new light on the Principia—
its argument structure, epistemic support, and long-term posterity—he has changed
research agendas and opened new vistas onto old problems and figures. The papers
in this volume attest to that broad influence.
Epistemology
In work on Newton and in collaborations on late-classical physics, George had
devised and relied on a tripartite division of the epistemic status a proposition
can have in the course of research. Specifically, he has distinguished between
a proposition counting as a hypothesis; being “taken to be true”; and being
established. What underpins this distinction? As we explained above, it is the quality
of the evidence for that proposition—at a given time; and over long stretches—and
the constitutive role assigned to it during ongoing research that involves it.
Knowledge In his paper, Jody Azzouni asks how Smith’s taxonomy of epistemic
stances dovetails with some central distinctions in traditional epistemology. He
argues that propositions taken to be true (by Smith’s criteria for that notion) really
count as knowledge. To make his case, Azzouni distinguishes two concepts of
knowledge, viz. ordinary and philosophical. The latter is an artifact of conceptual
engineering: it has an inbuilt constraint, namely, that knowledge be infallible.
However, Azzouni sides with the ordinary notion—the one behind vernacular uses
of “to know.” This notion, he explains, does not require infallibility (about the
particular proposition P at issue). And so, he concludes, if a community of inquiry
has enough evidence for taking P to be true, then that community knows that P.
Azzouni then finishes à rebours. From the vantage point of his dual picture, he
examines Descartes’ notion of knowledge. His verdict is that Descartes endorsed
the infallibility condition, notwithstanding the Cartesian distinction between scientia and cognitio. Moreover, Azzouni argues provocatively, Descartes’ above
endorsement appears to have contaminated Newton’s thought too, in two respects.
First, both Newton and Descartes take deduction to be—if suitably carried out—
a procedure that cannot fail. Second, both think that a proposition counts as
knowledge only if it has been “deduced.” There are, of course, key differences
xiv
Introduction
between their idea of deduction and our modern concept. Still, Azzouni concludes,
Newton shared with Descartes the commitment that “deduction” is the gold standard
for knowledge, and knowledge requires infallibility.
This line of argument suggests two directions for further research. It should
prompt us to take a deeper look at Descartes’ and Newton’s respective accounts of
deduction, and how they took specific mathematics—frameworks and approaches—
to underwrite the presumed infallibility of their “deductions.” On a broader note, it
encourages us to inquire into when and why the early modern tenet of infallibility
(of knowledge) gave way to the current situation in which fallibility is the default
position in epistemology.
Strong Evidence George first uncovered the very special role of systematic discrepancies in Newton’s work, and their legacy for gravitation theory in the centuries
after him. When they are systematic, such discrepancies become very valuable,
as second-order phenomena in the sense described above: they drive research
forward, by pointing to aspects of the theory that need refining or strengthening;
and they yield strong evidence for the theory, if it accounts for them from its
own resources. This particular insight suggests two lines of further research. Do
systematic discrepancies play this role in science more broadly, beyond research
in Newtonian gravitation, where Smith first uncovered it? And, did anyone else
between Newton and Smith grasp the special value that discrepant phenomena
have for research in the exact sciences? The paper by Teru Miyake takes up these
questions, and answers them compellingly. In his recounting, it was the natural
philosopher John Herschel who in the 1820s explained the evidential value of
“residual phenomena,” or observed discrepancies—between (first order) predictions
from theory and the observed behavior of target objects. Such phenomena further
confirm the theory, if it can account for them; or they require—and may even
suggest—revisions to theory, if its laws cannot handle those “residues.” Though he
saw the method as working at its best in physical astronomy (thus foreshadowing,
as it were, Smith’s magisterial paper of 2014), Herschel seems to have thought the
method was more general, potentially useful in many areas of exact science. His
lucid case for it would influence key figures of Victorian logic and philosophy of
science, as Miyake shows.
At the same time, against the backdrop of George’s complex picture of evidence
we presented above, we can see more clearly the limits of Herschel’s insight into
the confirmatory role of “residual phenomena.”
Empiricism and Metaphysics A recurring theme in Smith’s work is how, in the
process of theory building, Newton always took great care to anticipate how he
might go wrong, viz. how the evidence available falls short of supporting his
physical claims. Robert DiSalle’s paper takes up that theme and extends its reach.
In particular, he asks whether Newton took the same evidential care in regard to
the philosophical foundations of his theory, not just the physical details of the solar
system. DiSalle argues in the affirmative: he points out how much empirical import
attaches to some of the key foundational terms in the Principia—absolute velocity,
Introduction
xv
relative motion, true rest, uniform translation, absolute rotation, and the like. In his
recounting, Newton gradually found a way to demarcate which of these terms are
empirically well-grounded, and which ones are fated to remain ‘hypothetical.’ The
key, DiSalle explains, was Newton’s gradual refinement of the relativity of motion,
and the resulting insight into how to formulate a concept of inertia compatible with
Corollaries V and VI to the laws of motion.
Alongside his philosophical analysis, DiSalle offers a nuanced reconsideration of
historical figures like Berkeley and Mach, commonly known as critics of absolute
space and time. He suggests that their grasp of “Newtonian relativity” was really
deeper than we have appreciated so far. More broadly, we can see in his paper a plea,
inspired by George’s picture of Newtonian evidence, to reflect on the various ways
in which a metaphysics could be empiricist—by way of looking at the empirical
credentials of the metaphysical ingredients proffered as foundations for physical
theory.
Measurement and Evidence A very important theme from Smith is the evidential
value of measurements. In his reconstruction, when measured values of a key
parameter converge over time, they count as evidence (for the theory in which the
parameter is embedded).
That evidence is even stronger when that particular ingredient—the expression
that connects a theory-bound quantity with a measured parameter—can be further
refined. As we explained above, behind such expressions, or functional dependencies, are tacit idealizing assumptions. If these assumptions get relaxed (with the
functional expression refined accordingly); and if new, further measurements bear
out the newer, refined formulas linking theory to metric proxies—then that counts
as strong evidence for the theory.
Allan Franklin’s study brings this theme to the fore. His diachronic survey
highlights how attempts to measure a certain parameter (the mean density of the
earth) yielded more exact values in the long run: from Cavendish to the late
nineteenth century. At the same time, all such experimental attempts have to
grapple with the challenge of disaggregation. Namely, the need to identify factors—
temperature gradients, friction damping, material-specific properties—that might
lead to discrepant measurements; and to screen them off, or at least to estimate
their effects, so as to subtract them from discrepant phenomena.
In this respect, Franklin’s paper dovetails with the studies by Miyake, and Biener
and Domski in this volume. They too single out for reflection Smith’s emphasis on
discrepant phenomena, and to what extent we can marshal such discrepancies in
support for theory.
Evidence in the Sciences of the Past Certain key elements in Smith’s account
of strong evidence—for instance, the emphasis on systematic discrepancies, and
on converging values of a measured parameter—suggests that such evidence is
available just in advanced, strongly mathematized science. Which implies, inter
alia, that the sciences of the past—and disciplines that study one-off events,
in particular—are at an evidential disadvantage. That is because their specific
xvi
Introduction
domain exhibit neither repeatable patterns (from which we could discern systematic
discrepancies) nor serial values of measured parameters.
In recent years, however, Carol Cleland and others have argued that the historical
sciences are not at an evidential disadvantage, relative to the physical sciences
above. Craig Fox’s paper takes a critical look at the reasons for their epistemic
optimism. In their account, these sciences rely on a common pattern of confirmation.
Specifically, they infer from present traces of past events—artifacts, material remnants, physical leftovers, and byproducts of extinct processes—to some common
cause responsible for those traces. Underwriting this generic pattern of inference
is a key assumption, which these authors call the “Common Cause Principle.” Fox
inquires into the warrant for this assumption. That warrant, he explains, is a claim
further upstream, namely, that the past is overdetermined. That claim is really two
ideas. One is epistemological, and says that, for us to infer reliably to some past
event, we do not need all (causal) traces of that event as evidence for it—just some
traces are enough. The other regards ontology and says that past events always leave
at least some causal traces into the present.
These authors think their key claim—that the present overdetermines the past—
follows from David Lewis’ analysis of causation in terms of possible worlds.
However, Fox shows, the analysis does not support their metaphysical premise—
because in his account Lewis excluded certain sets of possible worlds by fiat.
In particular, Lewis’ analysis ignored “backtracking counterfactuals.” But that
move, as Fox points out, in effect excluded—illegitimately—the possibility that our
present is causally and nomically compatible with many, different pasts. And so,
their key premise (that the historical past is overdetermined) holds by stipulation,
not in virtue of metaphysical argument. Thus Fox’s incisive scrutiny undermines
(or challenges) one case for optimism about evidential reasoning in the historical
sciences—and leaves the important task of developing a more compelling case open
to further work.
Fox’s piece is on the cutting edge of recent studies that investigate the structure
of confirmation in science about the past. That enlarges and sharpens our picture of
the descriptive import and explanatory power of George’s account of evidence.
Newton Scholarship
George’s research has drawn renewed attention to Newton’s methodology—a
generous term that covers both heuristics, or guidance for “reasoning more securely”
in natural philosophy, with empirical results appropriately guiding inquiry; and
also logics of confirmation, viz. patterns of evidential reasoning, and constraints
on admissible inferences.
Newton on Methodology In regard to the former—heuristics—Monica Solomon in
her paper seeks to uncover a new aspect of Newton’s thought on that topic. She
argues that a key element in the famous Scholium on space, time, and motion is
Introduction
xvii
best understood as a piece of methodology. In particular, Solomon claims, Newton
meant his example of the two globes (connected by a cord, rotating around each
other in empty space) as a terse blueprint for setting up, and tackling, problems
in orbital dynamics. As she explains, the globes are an epitome of the type of
dynamical system that Newton studies in Principia. The example requires us to
think of the globes as having the quantitative properties Newton sets down in his
definitions—properties that covary in accordance with his laws of motion—and as
being sufficiently far away from perturbing factors that we can treat the globes as
a quasi-isolated system (a supposition that Newton at the end of his book, in the
General Scholium, seems to reaffirm).
Thereby, Solomon breaks with a long tradition (going back to Mach) that
saw Newton’s globes example as making a point about the metaphysics and
epistemology of true motion. In effect, her paper makes a case for further study
of Newton’s heuristic resources—a topic that harmonizes nicely with George’s
analyses, and which is sure to reward further scholarship on it.
Newton, Method, and Optics The paper by Howard Stein brings to the fore, as does
much of Smith’s work, the precision and discipline with which Newton evaluated
claims in natural philosophy. His richly historical account unfolds along two strands
of inquiry. One clarifies Newton’s distinctive treatment of the levels of certainty
that can be attained in scientific inquiry. Stein here sharply contrasts Newton’s
capacity for “multiple vision,” an ability to assess the differing roles and epistemic
status of components of a proposed theory, with the more one-dimensional views of
his contemporaries. Hooke, Huygens, and other critics often assumed that Newton
relied upon, but failed to acknowledge, some form of mechanical hypotheses or
metaphysical posits. The contrast reflects the novelty of Newtonian inquiry, a
theme that resonates with Smith’s work. But Stein further takes contemporary
commentators to task for misreading Newton’s methodology and reveals the
rewards of a more sophisticated reconstruction of Newtonian empiricism.
The second strand addresses the status of metaphysics, given Newton’s claims to
establish specific claims regarding the nature of light and gravity without needing
to “mingle conjectures with certainties.” By contrast with the dominant Cartesian
tradition, for Newton metaphysical notions ultimately have an a posteriori character.
Newton introduced or adopted those metaphysical elements in response to the
foundational and explanatory needs of physical theory, not before (and independently of) them. And in addition, Newton regarded his metaphysics as essentially
revisable—not at will, of course, but always in reaction to the changing evidential
fortunes of the empirical theories. Here Stein in particular reviews, briefly, the
transformative steps Newton took in formulating his optical theories, and later in
revising traditional metaphysical categories of space, body, and force in the course
of writing the Principia. Stein reads Newton’s assessment as again illustrating a
capacity for multiple vision, in this case treating the doctrines of space and time as
having a more secure status than the more conjectural account of the nature of body.
Furthermore, as Stein emphasizes, far from entirely eschewing hypotheses, such as
the corpuscular theory of light, Newton specifically acknowledged their value—not
xviii
Introduction
as something presupposed as part of his experimental reasoning but as a guide to
further inquiry.
The Principia and Hypotheses From his earliest work in optics and the disputes it
prompted, Newton always tried to demarcate results he had established from “mere
hypotheses.” Much of George’s work is a reconstruction of how Newton developed
a method to “reason more securely” in natural philosophy—by contrast with the
explicitly hypothetical methods of his contemporaries. Huygens, for example,
clearly endorsed a version of reasoning from hypotheses. The chapter by Zvi Biener
and Mary Domski revisits the broad idea that, at least in mathematical mechanics,
Newton did not reason from, nor did he endorse, hypothetical assertions.
For the case of the “mathematical” Book I of the Principia (and its application
in Book III to gravitational phenomena), Newton’s rejection of hypotheses appears
unimpeachable. However, when they turn to Book II—the mathematics of motion in
resisting media, plus Newton’s experimental basis for it—Biener and Mary Domski
think that his rejection of hypotheses looks shakier. For one, Newton there starts
from assumptions about the dynamics of a resisting medium: suppositions about
which of its properties are causally relevant; and about their kinematic effects (on a
solid moving in that medium). For another, it is hard, they argue, to find in Book II
the key elements that allowed Newton to rise above hypothetico-deductivism: robust
theorems (for inferring force laws from motion phenomena); convergent values of a
measured parameter; or a mathematical treatment that allows him to disaggregate
the respective contributions (of various physical causes and mechanisms) to a
complex motion effect, e.g., the decay of pendulum swings in water. But, if those
sources of evidential reasoning are lacking, Biener and Domski conclude, isn’t
Newton’s pattern of confirmation in Book II closer to Galileo’s and Huygens’
hypothetical approaches than we have thought so far?
The Architectonic of the Principia The argument structure of Newton’s treatise has
long been an elusive puzzle. Katherine Brading in her chapter sheds new light on this
difficult topic. The Principia unites two disciplines, or areas of inquiry, she argues.
Books I and II establish results in rational mechanics—the applied mathematics
of orbits induced by central forces of interaction. In contrast, Book III turns to
physics: a theory of a force (gravity) seated in bodies, its effects on them, and its
quantitative relations to other properties of material bodies, like mass, shape, and
volume. According to Brading, in the Principia, these two disciplines are not simply
juxtaposed; they are connected by a conceptual bridge, as it were. That bridge is the
definitions and “axioms, or laws of motion” that Newton placed at the outset, before
his three books. These elements serve a dual function in the treatise: they make up
the axiomatic basis of his mathematical mechanics; and they are the nomic basis of
his quantitative physics of gravity.
This framework allows Brading to elucidate the diachronic context of his theory,
not just its synchronic makeup. In particular, she explains, with the Principia
Newton continued, while transforming very drastically, a program that goes back
to Descartes. It was the program of combining rational mechanics with physics into
Introduction
xix
a “philosophical mechanics,” as Brading calls it (because physics then counted as
a branch of philosophy). He expanded the scope of rational mechanics well beyond
what anyone had done by 1700. And, he showed how the physics of gravity is
amenable to mathematization. Specifically, well-established empirical methods of
measurement give us a quantitative handle on the effects of gravity. From those
effects, Newton’s rational mechanics lets us infer the strength and direction of the
gravitational forces responsible for them; and to treat these forces as endowed with a
measure, or algebra. Thereby, Brading’s study dovetails with George’s well-known
emphasis on Newton treating forces as quantities, viz. actions endowed with a ratio
structure, or measure.
Mathematical Methods The Principia is famously geometric: geometric objects
(e.g., lines, plane curves, areas) stand for properties of motion and force; and
geometric methods (e.g., auxiliary constructions, diagrammatic reasoning) are
among its key vehicles for proof in rational mechanics. Newton’s reliance on
this geometric framework—its scope of representation, heuristic merits, inferential
limits, and equivalence to other frameworks then available—have been a topic for
much scrutiny. Debates began already in his time, and scholarship in the last halfcentury has greatly advanced our understanding of Newton’s formal methods. The
chapter by Niccolò Guicciardini makes a major contribution to that understanding,
by shedding light on a difficult, elusive, and little studied topic. Broadly described,
that topic is integration in Newton’s mathematical thought: its meaning, scope, key
techniques, and comparative virtues. Guicciardini establishes a number of novel
and important results. One is that, overall, Newton was a good deal more of an
“algebraist” than the Principia might lead us to expect. At least in matters of
integration, he clearly favored proof procedures that rely on the rule-governed
manipulation of algebraic formulas—rather than inferring from inspection of a
geometric object. Another major result is that Newton had an incipient notion of
differential equation and had worked out ways to solve some classes—primarily,
by expansion into power series; by numeric approximation, where feasible; and by
change of variable, or substitution.
This particular result is subtle, but matters greatly, in two respects. For one, as
Guicciardini notes, Newton’s approaches to “fluxional,” or differential, equations
differ from our modern approach (which goes back to Leibniz and his disciples)
that seeks analytic, closed-form solutions for them. We may be tempted to think
that difference counts as a weakness of Newton’s fundamental concept of “fluxion.”
But the reason lies elsewhere, as Guicciardini explains: many of Newton’s intended
applications—for the techniques above—are in the dynamics of perturbed systems.
In general, those systems are not integrable in closed form. His approaches (numeric
integration, and expansion into infinite series) are excellent approximations of
those generally unavailable exact solutions. And so, what prima facie looks like
a weakness is in fact a source of strength. For another, it helps us see that
the gap between Newton’s mathematical methods and modern formulations (of
gravitational dynamics) is not as great as a casual look at the Principia might
suggest. Thanks to Euler and Lagrange, mechanics settled into the form familiar to
xx
Introduction
us late-moderns, viz. of a connected set of differential equations (of motion) derived
from dynamical laws. In that regard, the Principia with its geometric language looks
separated by a gulf from our versions of mechanics. Guicciardini’s study reveals
that to be an exaggeration, ultimately. Though he lacked a function concept (and the
Leibnizian notation that eases their calculus so greatly), Newton after all did have
the key ingredient of modern mechanics: describing the motion (over an instant) by
way of a differential expression, which then must be integrated. Guicciardini’s paper
resonates with another theme from George Smith, among whose breakthroughs has
been the careful study of Newton’s mathematical methods at the more advanced
stages of theory construction in the Principia.
History of Philosophy of Science
George’s work has opened new lines of research on the diachronic aspects of
foundations for mechanics, and on how major philosophical figures responded to
Newton’s achievement. A number of chapters in this volume explore these new
lines.
The Status of Gravity A widespread view has it that Newton was averse to
metaphysical inquiry into the objects and results of the Principia, e.g., the ontology
of gravity; and that when competitors took issue with his physics on metaphysical
grounds, they were ultimately ill-advised to do so. In his paper, Andrew Janiak gives
good reasons to resist this picture. His starting point is Newton’s own attempts to
make sense of his key result, in Book III, proposition 7, that “gravity is in all bodies
universally.” Namely, to elucidate its real semantic content, underlying ontology,
and the broader metaphysical framework that supports them. As Janiak shows, this
was no easy task, and Newton grappled with it at length. Nor was the answer clear
and uncontroversial—not even to his fellow travelers like Roger Cotes, let alone
to antagonists like Leibniz. Newton’s struggles with this central question—what
does it really mean to assert that gravity is universal—required him to make forays
into the very metaphysics that he allegedly eschewed. Janiak’s study has another,
implicit benefit (in addition to turning the tables on the received view). In particular,
it breaks a new path to better situating other, important figures (such as Emilie du
Châtelet and Kant) in their own efforts to elucidate the meaning of gravity being
universal.
Kant and Newton George’s careful work on Newton’s approach to evidence has yet
another important benefit: it has opened a new vista on eighteenth-century philosophers’ dialogue with the Principia.6 Michael Friedman’s paper takes a further step
6 Studies of that dialogue are old, to be sure. However, before Smith they tended to be rather
one-sided, or narrowly focused on the inertial-kinematic basis of the Principia: its metaphysics of
space, time, and eighteenth-century reactions to it. Representative for that work are Earman 1989,
Friedman 1992, and Rynasiewicz 1995.
Introduction
xxi
on this new ground, by looking at Kant’s response to these aspects of Newton. He
thinks there are two significant aspects of this response. Kant strongly endorsed
Newton’s idea—which George was the first to highlight for us—that theorymediated measurement is a very significant source of confirmation in gravitational
dynamics. However, Friedman claims, Kant favored a notion of phenomena that is
thicker than Newton’s analogous concept. Specifically, Newtonian phenomena of
motion were exclusively kinematic: patterns of planet- or satellite motion over time,
equilibrium configurations, and the like. The notion Kant preferred, however, is that
of phenomena as “involving causal and dynamical information” as well, not just
kinematic content. Friedman argues in favor of Kant’s stronger notion, because he
thinks it does useful work in contemporary philosophy of science: it can help us
chart a middle path in the disputes between realism and instrumentalism (about the
relation of theory to its target objects).
The paper by Katherine Dunlop unfolds in the same register as Friedman’s
investigation above. Thereby, our image of Newton’s reception by the philosophers
after him becomes clearer. At a critical juncture in his argument for universal gravity,
Newton in Book III relies on his third law of motion. That key move won him few
friends, it seems. On the Continent, some objected that it asserts bodies to interact
without contact, which they dismissed as unintelligible. Others, like Roger Cotes—
and also Euler, later in the century—demurred that Newton lacked enough evidence
to claim the law applies to actions at a distance. In her study, Dunlop uses Kant so
as to cast a new, and more positive, light on Newton’s move.
Kant endorsed action-at-a-distance early in his career, and so he had none of the
Continentals’ qualms about it. Accordingly, his reaction to Newton’s Lex Tertia
was different. For one, Dunlop explains, he argued that Newton relies on that
law well before the master argument in Book III. In fact, he needs it already in
Book I; specifically, in section 11, where Newton shows how to reduce the twobody problem (of two particles interacting as they orbit around each other) to two
separate, more easily tractable one-body problems, viz. of a mass in Kepler motion
around a fixed center of force. Kant thinks that Newton needs to assume the third
law for his approach to go through, but that he did not acknowledge it as an explicit
premise.
For another, Dunlop adds, Kant’s natural philosophy helps elucidate some
important but otherwise baffling claims by Newton. Kant distinguished between
‘dynamical’ and ‘mechanical’ treatments of force. The former is quantitative, or
mathematical, but it has causal import—it regards forces as causes (of acceleration)
that a source could exert even while stationary. From that vantage point, Dunlop
argues, Newton’s treatment of force in section 11 counts as “dynamical,” hence
causally relevant. That would resolve the apparent tension between his well-known
claims that he treated gravity merely “mathematically,” and yet that his treatment
shows gravity to “really exist,” and “suffice for” celestial and terrestrial motions.
Foundations of Classical Mechanics The three laws of motion that Newton asserted
purport to be general: they apply to forced motion beyond the relatively narrow
class of gravitational effects. Many theorems in Book I—stating relations between
xxii
Introduction
accelerations and their corresponding orbits—are about forces other than inversesquare and placed at a focus (which gravity is). And, Newton in a famous passage
conjectured that many things led him “to have a suspicion that all phenomena
may depend on certain forces,” and hoped his theorems and methods will help
his posterity discover those further forces (Newton 2004, 60; emphasis added).
To be sure, many after him did continue his agenda. Still, a mere century after
the Principia, Lagrange created a genuine alternative to Newton’s foundation. His
book, Mécanique Analytique, unified statics and dynamics from a dual basis—a
principle of virtual work, and a postulate known as “D’Alembert’s Principle.” In
some respects, Lagrange’s framework is more powerful than Newton’s. The chapter
by Sandro Caparrini takes a closer look at Mécanique Analytique. It shines a light
on the layered structure of that key treatise, and on the early growth of the theory
it contains. In Lagrange’s lifetime, the book went through two editions; in the
interim, French mathematicians extended greatly the reach of mechanics, to novel
and difficult phenomena. Caparrini shows convincingly how Lagrange was able to
incorporate those new advances into his framework, turning it into an even more
formidable competitor for the tradition of theorizing that came out of the Principia.
Caparrini’s study resonates with an important theme from George Smith, though
not explicitly. In Katherine Brading’s terms above, the theory in Lagrange’s book
is a rational mechanics, not a physics. Warrant for its results cascades downstream:
from its basic, most general principles, by deductive reasoning alone. In Mécanique
Analytique, empirical facts are very scarce, adduced mostly as illustration, not
confirmation. That raises the weighty question, where does empirical evidence—
especially strong evidence, of the kind that Smith has so fruitfully explored so
far—enter Lagrange’s “analytic” mechanics? The question is far from easy, to be
sure, but it is to be hoped that scholars will take it up to grapple with it. Thereby,
another theme from Smith would come to the fore, namely, that confirmation in
advanced, strongly mathematized theories is diachronic: it is temporally extended,
and rests on a historical record of accumulating, and increasingly stronger evidence
for the theory.
Early Scientific Cosmology The mathematical astronomy in Ptolemy’s Almagest
rests inter alia on a small number of extra-mathematical assumptions—about what is
at rest, what moves, and how far the stars are from us. Ptolemy there merely gestures
at argument for these assumptions, or “hypotheses,” as he calls them. His proffered
support for them in that book is cursory and rash; which feeds the suspicion that
the Almagest is a collection of simulation software, as it were: mere algorithms for
predicting or retrodicting ephemerides and select orbital parameters—not a genuine
“system of the world.” The basis for that system would come from physics, we
may think. In particular, from Aristotle’s philosophical physics, which—thanks to
its doctrines of natural place, motion, and five elements—easily entails that the earth
is at rest in the world center while stars and planets revolve around it.
In his chapter, the late Noel Swerdlow subverts this received wisdom. He does
so by way of a synoptic study of Planetary Hypotheses, an important tract by
Introduction
xxiii
Ptolemy. That work, Swerdlow argues, contains a theory of cosmology, or system
of the world. But, it is not based on metaphysical premises from Aristotle. In
fact, it is properly scientific. Specifically, it is quantitative: it makes claims about
celestial distances and orbital parameters. Likewise, it is empirical: the inputs for
theory building are empirical givens, e.g., long-term observational data or patterns
of perception; and the evidence for the theory is empirical as well. And, Swerdlow
suggests, it is supported by physical assumptions, e.g., about the causal mechanisms
of planetary motion, and about the physical consequences of counterfactual setups.
Thus, Swerdlow concludes, Ptolemy’s Planetary Hypotheses has every right to
count as a scientific cosmology; indeed, it was the first one of its kind. To help
the reader, he ends with a synopsis of the complicated transmission and reception
of Ptolemy’s theory above.
There is a broader lesson here, and it lies at the confluence of two strands of
thought. For one, George’s painstaking work has shown inter alia not just how
subtle Newton’s methods for gathering evidence were—but also how easy it was
for many figures after him to miss those methods. For another, the papers above
show how unclear—and far from obvious or uncontroversial—the foundations of
Newton’s theory were, in the century after him: its ontology, basic semantics,
and generic methods. Together, these two strands suggest a revisionist conclusion
that challenges an interpretive consensus going back to Kuhn’s Structure. In that
influential book, Kuhn had argued that when paradigm-making work transforms its
field (into an arena for “normal” science), it ends previous controversies (about the
ontology and methods suited for that domain), and it produces consensus (about the
basic objects, acceptable methods, legitimate problems, and criteria for solutions).
Kuhn counted Newton’s Principia among the epitomes for his case (Kuhn 1996, 11,
17). But, the chief results in the papers by Friedman, Solomon, Janiak, and Dunlop
cast increasing doubt on Kuhn’s picture of the Principia’s role in the history of
exact science. Thereby, these results help clear the field for a new generation of
scholars to step in and determine what Newton’s book really did to physics and
its philosophy—in effect, how eighteenth-century philosophers and their successors
answered “Newton’s challenge to philosophy” (Schliesser 2011).
This volume ends, appropriately, with George E. Smiths’ reflections on the two
crafts—the philosophy and history of exact science—that he has cultivated and
fostered so admirably. The occasion for his reflections is the theme of revisiting
accepted science. More specifically, the diachronic process whereby pieces of
theory—once they become “accepted,” or used constitutively to carry out further
research—get tested over and over again, often with increased stringency; and the
long-term outcomes of such “revisiting.” He illustrates this process with examples,
discussed in exquisite detail, from gravitation theory and late-classical physics.
These examples, and others, support his concluding message—really, a dual lesson
for students of science. Philosophers who investigate the epistemic aspects of
science ought to pay close attention to the diachronic side of its credentials as
knowledge: for any given theory, they must study the history of the evidence
for it, in Smith’s memorable phrase. And, historians of science ought to avoid
dogmatic allegiance to the idea that social-group dynamics holds the master key
xxiv
Introduction
to understanding the birth and growth of scientific knowledge. Rather, they would
do well to pay attention to the extended record of testing and revisiting the epistemic
credentials of that knowledge as it grew.
To both communities above, in sum, George emphasizes the crucial importance
of longue durée, fine-grained study of the confirmation processes behind the
production of scientific knowledge. These processes begin when a theory has
been accepted, not before. And so, a corollary of his lesson is that we ought
to revisit—and be prepared to drastically revise—Kuhn’s old picture of normal
science. Thereby, his lesson resonates with the dominant note of the papers in this
volume.
De magistro
George’s work is unique as well in a respect that makes it hard to present
synoptically in any introduction, not just this one. He has conveyed much of his
philosophy of science by a route that goes beyond the standard of our time, viz. the
journal article or book chapter qua discrete, printed units of research. In particular,
that route has been his legendary two-semester course on Newton’s Principia—
really, a master class in the history and philosophy of evidential reasoning developed
at Tufts University but offered at a number of other institutions (including Stanford,
Notre Dame, and Duke). Roughly, the first semester puts the student in a position to
read the Principia by studying seventeenth-century primary texts (including Galileo,
Kepler, Huygens, and pre-Principia Newton). The second is a close reading of
the Principia, theorem by theorem; and then an overview of Newton’s impact on
mechanics after him.
The course is pitched to undergraduate students, but often the auditors include
graduate students and faculty. For many years, it was offered on Wednesday
evenings with a three-hour time slot interrupted by modest breaks. George lectured
by partially reading from amazing lecture notes and using the blackboard when
necessary. (The lecture notes would be made available after class, and after further
careful editing.)
What made George’s lectures mesmerizing was that he took all the students at all
levels seriously as genuine interlocutors in the shared adventure of understanding
Newton’s method. And what made the whole point even more remarkable: many
sessions would start with his excitement of his latest discovery—sometimes an
overnight discovery—of the evolution and development of Newton’s thought.
The paper assignments for students taking the course for a grade were all clearly
designed to foster a collective endeavor to understand the evidential status of
particular works at a given time. An assignment could read: “what was the status
of Kepler’s laws in 1680?” This could open the door to more metaphysical papers
on what exactly the nature of a Keplerian law was in the late seventeenth century; or
to examinations of seventeenth-century discussions of Kepler in astronomical texts
of the period.
Introduction
xxv
George’s pedagogic methods center on very high expectations from his students
by assigning challenging primary texts (and a lot of them) without flipping the
classroom. What he does do, and he does this amazingly well, is prepare the student
qua student to be a co-equal in his astounding intellectual adventure. He makes
sure they acquire all the technical background, one firm step at the time, and then
puts them in the position to contribute to active research, if they so wish. (Many
students end up writing term papers that could be the basis of a journal article.)
Subsequently, George would often come to co-author with students, by drawing
on their specialized mathematical, linguistic, or research skills. In turn, his lectures
along the years become enriched by what he learns or discovers while collaborating
with them or grappling with their questions. Along the way, he invites them out on
frigid winter nights to stare through a telescope, so as to experience what Galileo
might have felt when he turned one toward the Moon. (George makes sure to let
them try to see anything with the magnification that Galileo had available.) Through
his course, which he has taught for nearly three decades, George has reached and
influenced some four generations of research, from senior luminaries to current
graduate students. Inter alia, the chapters in this volume attest to the enduring
influence of his teachings.
Marius Stan
Chris Smeenk
Eric Schliesser
References
Buchwald, J.Z., and G. Smith. 2002. Incommensurability and discontinuity of
evidence. Perspectives on Science 9: 463–498.
Earman, J. 1989. World enough and space-time. MIT Press.
Friedman, M. 1992. Kant and the exact sciences. Harvard University Press.
Kuhn, Th. 1996. The structure of scientific revolutions, 3rd ed. University of
Chicago Press.
Miyake, T., and G.E. Smith. 2021. Realism, physical meaningfulness, and molecular
spectroscopy. In Contemporary scientific realism, ed. T.D. Lyons and P. Vickers,
159–182. Oxford University Press.
Newton, I. 1999. The Principia: Mathematical principles of natural philosophy, ed.
and trans. I.B. Cohen with A. Whitman. University of California Press.
Newton, I. 2004. Philosophical writings, ed. A. Janiak. Cambridge University Press.
Rynasiewicz, R. 1995. ‘By their properties, causes and effects.’ Newton’s scholium
on space, time, place and motion – II. The context. Studies in History and
Philosophy of Science 26: 295–321.
Schliesser, E. 2011. Newton’s challenge to philosophy: A programmatic essay.
HOPOS: The Journal of the International Society for the History of Philosophy
of Science 1: 101–128.
Publications by George E. Smith
Several engineering papers and more than 140 engineering reports issued by various
government agencies, and by NREC and Turbomachinery Solutions, Inc.
1979a. [with M.L. Kean] Issues in core linguistic processing. Behavioral and Brain
Sciences 2 (3): 469–70.
1979b. [with S.M. Kosslyn, S. Pinker, and S.P. Shwartz] On the demystification of
mental imagery. Behavioral and Brain Sciences 2(4): 535–48.
1980. [with S.M. Kosslyn] An information-processing theory of mental imagery:
a case study in the new mentalistic psychology. PSA 1980, Vol. 2, 247–66.
University of Chicago Press.
1983. [with N. Daniels] The plasticity of human rationality. Behavioral and Brain
Sciences 6(3): 490–1.
1986. The dangers of CAD. Mechanical Engineering 108(2).
1994a. [with S. Cohen, G. Smith, R. Chechile, and B. Cook] Designing curricular
software for conceptualizing statistics. Proceedings of the First Scientific Meeting of the International Association of Statistics Education, eds. L. Brunelli and
G. Cicchitelli, 237–45. University of Perugia.
1994b. [with S. Cohen, R. Chechile, F. Tsai, and G. Burns] A method for evaluating
the effectiveness of educational software. Behavior Research Methods, Instruments, and Computers 26(2): 236–41.
1995. [with W. Harper] Newton’s new way of inquiry. The Creation of Ideas in
Physics: Studies for a Methodology of Theory Construction, ed. J. Leplin, 113–
66. Kluwer/Springer.
1996a. [with S. Cohen and R. Chechile] A detailed, multisite evaluation of curricular
software. Assessment in Practice, ed. Trudy W. Banta et al., 220–2. San
Francisco: Jossey Bass.
1996b. [with S. Cohen, R. Chechile, and G. Burns] Impediments to learning
probability and statistics from an assessment of curricular software. Journal of
Educational and Behavioral Statistics 21(1): 35–54.
xxvii
xxviii
Publications by George E. Smith
1996c. [lead author; with L. Brown, R. Chechile, S. Cohen, R. Cook, J. Ennis, D.
Garman, S. Lewis] ConStatS: Software for Conceptualizing Statistics. Englewood Cliffs: Prentice-Hall.
1996d. Chandrasekhar’s Principia: an essay review. Journal for the History of
Astronomy 27: 353–62.
1996e. [with E. Schliesser] Huygens’ 1688 report to the VOC. De Zeventiende Eeuw
12: 196–214.
1997a. [with J. Z. Buchwald] Thomas S. Kuhn, 1922–1996. Philosophy of Science
64: 361–76.
1997b. J. J. Thomson and the electron, 1897–1899: an introduction. The Chemical
Educator 2(6): 1430–71.
1998a. [with W. Harper] Isaac Newton. Encyclopedia of Philosophy. London:
Routledge.
1998b. Newton’s study of fluid mechanics. International Journal of Engineering
Science 36: 1377–90.
1999a. The achievements of Book 2. I. Newton, Mathematical Principles of
Natural Philosophy, trans. I.B. Cohen and A. Whitman, 188–194. University of
California Press.
1999b. The motion of the lunar apsis. I. Newton, Mathematical Principles of Natural
Philosophy, trans. I.B. Cohen and A. Whitman, 257–64. University of California
Press.
1999c. Planetary perturbations: the interaction of Jupiter and Saturn. I. Newton,
Mathematical Principles of Natural Philosophy, trans. I.B. Cohen and A.
Whitman, 211–17. University of California Press.
1999d. Newton and the problem of the Moon’s motion. I. Newton, Mathematical
Principles of Natural Philosophy, trans. I.B. Cohen and A. Whitman, 252–57.
University of California Press.
1999e. A puzzle in Book 1, Prop. 66, Coroll. 14. I. Newton, Mathematical Principles
of Natural Philosophy, trans. I.B. Cohen and A. Whitman, 265–68. University of
California Press.
1999f. How did Newton discover universal gravity? The St. John’s Review XLV (2):
32–63.
2000a. Fluid resistance: why did Newton change his mind? Foundations of Newtonian Scholarship, eds. R. Dalitz and M. Nauenberg, 105–36. Singapore: World
Scientific.
2000b. [with D. Mindell] The emergence of the turbofan engine. Atmospheric Flight
in the Twentieth Century, eds. P. Galison and A. Roland, 107–55. Dordrecht:
Kluwer.
2001a. The Newtonian style in Book 2 of the Principia. Isaac Newton’s Natural
Philosophy, eds. J.Z. Buchwald and I.B. Cohen, 249–97. MIT Press.
2001b. [with I.B. Cohen, A. Whitman, and J. Budenz] Newton on fluid resistance
in the first edition: English translations of the passages replaced or removed
in the second and third editions. Isaac Newton’s Natural Philosophy, eds. J.Z.
Buchwald and I.B. Cohen, 299–313. MIT Press.
Publications by George E. Smith
xxix
2001c. J.J. Thomson and the electron, 1897-1899. Histories of the Electron: The
Birth of Microphysics, eds. J.Z. Buchwald and A. Warwick, 21–76. MIT Press.
2001d. La scoperta dell’elettrone [Discovery of the electron] Storia della scienza,
Vol. 7, La scienza dell’Ottocento. Istituto della Enciclopedia Italiana.
2001e. Review of J.R. Christianson’s “On Tycho’s Island: Tycho Brahe and His
Assistants, 1570–1601.” Journal of Interdisciplinary History 32: 130–1.
2001f. Comments on Ernan McMullin’s “The impact of Newton’s Principia on the
philosophy of science.” Philosophy of Science 38: 327–38.
2001g. [with J.Z. Buchwald] An instance of the fingerpost: review of M. Christie,
“The Ozone Layer: A Philosophy of Science Perspective.” American Scientist
89(6): 546–9.
2002a. From the phenomenon of the ellipse to an inverse-square force: why not?
Reading Natural Philosophy, ed. D. Malament, 31–70. La Salle: Open Court.
2002b. The methodology of the Principia. The Cambridge Companion to Newton,
eds. I.B. Cohen and G.E. Smith, 138–73. Cambridge University Press.
2002c. [with I.B. Cohen] The Cambridge Companion to Newton. Cambridge
University Press.
2002d. [with I.B. Cohen] Introduction. The Cambridge Companion to Newton, eds.
I.B. Cohen and G.E. Smith, 1–32. Cambridge University Press.
2002e. [with I. B. Cohen] Newton and the lunar motion. Review of N. Kollerstrom,
Newton’s Forgotten Lunar Theory. Journal for the History of Astronomy 33: 212–
3.
2002f. [with J. Z. Buchwald] Incommensurability and discontinuity of evidence.
Perspectives on Science 9: 463–98.
2003. Review of “Meanest Foundations and Nobler Superstructures: Hooke, Newton, and ‘The Compounding of the Celestiall Motions of the Planetts.’” Physics
Today 56(9): 61–2.
2005a. [with S.R. Valluri, P. Yu, and P.A. Wiegert] An extension of Newton’s apsidal
precession theorem. Monthly Notices of the Royal Astronomical Society 358:
1273–84.
2005b. Was wrong Newton bad Newton? Wrong for the Right Reasons, eds. J.Z.
Buchwald and A. Franklin, 127–160. Springer.
2006. The vis viva dispute: a controversy at the dawn of dynamics. Physics Today
59(10): 31–6.
2007a. Isaac Newton. The New Dictionary of Scientific Biography, ed. Noretta
Koertge, 48–53. Farmington Mills, MI: Charles Scribner’s Sons.
2007b. Isaac Newton. Stanford Encyclopedia of Philosophy. (http://plato.stanford.
edu)
2007c. Newton’s Philosophiae Naturalis Principia Mathematica. Stanford Encyclopedia of Philosophy. (http://plato.stanford.edu)
2008. [with J. Maienschein] What difference does history of science make, anyway?
Isis 99: 318–21.
2009. [with J. W. Dauben and M. L. Gleason] Seven decades of history of science:
I. Bernard Cohen (1914–2003). Isis 100: 4–35.
xxx
Publications by George E. Smith
2010. Revisiting accepted science: the indispensability of the history of science. The
Monist 93: 545–79.
2012. How Newton’s Principia changed physics. Interpreting Newton, eds. A.
Janiak and E. Schliesser, 360–95. Cambridge University Press.
2013. On Newton’s method. Book symposium on W. Harper’s “Isaac Newton’s
Scientific Method.” Metascience 22: 215–246.
2014. Closing the loop: testing Newtonian gravity, then and now. Newton and
Empiricism, eds. Z. Biener and E. Schliesser, 262–351. Oxford University Press,
2014.
2016a. [with R. Iliffe] eds. The Cambridge Companion to Newton, second edition.
Cambridge University Press.
2016b. [with R. Iliffe] Introduction. The Cambridge Companion to Newton, second
edition, 1–32. Cambridge University Press.
2019. Newton’s numerator in 1685: a year of gestation. Studies in History and
Philosophy of Modern Physics 68: 163–77.
2020a. [with Raghav Seth] Brownian Motion and Molecular Reality: A Study in
Theory-Mediated Measurement. Oxford University Press.
2020b. Experiments in the Principia. The Oxford Handbook of Newton, eds. E.
Schliesser and Chr. Smeenk. Oxford University Press.
2020c. The Principia: from conception to publication. The Oxford Handbook of
Newton, eds. E. Schliesser and Chr. Smeenk. Oxford University Press.
2020d. Newton’s laws of motion. The Oxford Handbook of Newton, eds. E.
Schliesser and Chr. Smeenk. Oxford University Press.
2021a. [with T. Miyake] Realism, physical meaningfulness, and molecular spectroscopy. Contemporary Scientific Realism and the Challenge from the History
of Science, ed. T. Lyons and P. Vickers, 159–80. Oxford University Press.
2021b. [with the assistance of J.M. Musca] Du Châtelet’s commentary on Newton’s
Principia: an assessment. Epoque Emilienne, ed. R. Hagengruber, 255–310.
Springer.
Forthcoming, a. Newtonian relativity: a neglected manuscript and understressed
corollary. Newtonian Relativity.
Forthcoming, b. [with E. Schliesser] Huygens’ 1688 report to the directors of
the Dutch East India Company on the measurement of longitude at sea and
the evidence it offered against universal gravity. Archive for History of Exact
Sciences.
In preparation, a. Isaac Newton’s De Motu Corporum, Liber Secundus: a Variorum
Translation of the Manuscript and Related Manuscripts from 1685, trans. G. E.
Smith and A. Whitman, with the assistance of R. Strobino, and with commentary
by S. Hesni and G.E. Smith, in preparation.
In preparation, b. [with Chr. Smeenk]. Newton on constrained motion: a commentary on Book 1, Section 10 of Newton’s Principia. Archive for History of Exact
Sciences.
Contents
1
Smith, Smith and Seth, and Newton on “Taking to Be True” . . . . . . . .
Jody Azzouni
2
‘To Witness Facts with the Eyes of Reason’: Herschel
on Physical Astronomy and the Method of Residual Phenomena . . . .
Teru Miyake
21
Newton on the Relativity of Motion and the Method
of Mathematical Physics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Robert DiSalle
43
3
1
4
Henry Cavendish and the Density of the Earth . . . . . . . . . . . . . . . . . . . . . . . .
Allan Franklin
65
5
Does the Present Overdetermine the Past? . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Craig W. Fox
83
6
Newton’s Example of the Two Globes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Monica Solomon
95
7
On Metaphysics and Method in Newton . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Howard Stein
115
8
Working Hypotheses, Mathematical Representation,
and the Logic of Theory-Mediation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Zvi Biener and Mary Domski
139
9
Newton’s Principia and Philosophical Mechanics. . . . . . . . . . . . . . . . . . . . . .
Katherine Brading
163
10
Newton on Quadratures: A Brief Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Niccolò Guicciardini
197
11
A Tale of Two Forces: Metaphysics and its Avoidance
in Newton’s Principia. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Andrew Janiak
223
xxxi
xxxii
Contents
12
Theory-Mediated Measurement and Newtonian Methodology . . . . . .
Michael Friedman
13
Immediacy of Attraction and Equality of Interaction
in Kant’s “Dynamics” . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Katherine Dunlop
243
281
14
Remarks on J. L. Lagrange’s Méchanique analitique . . . . . . . . . . . . . . . . .
Sandro Caparrini
307
15
Ptolemy’s Scientific Cosmology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
N. M. Swerdlow
327
16
Revisiting Accepted Science: The Indispensability
of the History of Science . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
George E. Smith
349
Editors and Contributors
About the Editors
Marius Stan is Associate Professor of Philosophy at Boston College. He works in
history and philosophy of science, with an emphasis on eighteenth century physics.
He is co-author, with Katherine Brading, of Philosophical Mechanics in the Age of
Reason (Oxford), and author of Kant’s Natural Philosophy (Cambridge).
Christopher Smeenk is Professor of Philosophy and Director of the Rotman
Institute at Western University, Ontario. He works in general philosophy of science,
philosophy of cosmology, and history of physics. In addition to numerous articles,
he is co-editor of the Oxford Handbook of Newton (with Eric Schliesser) and
co-author of The Aim and Structure of Cosmological Theory (with James Owen
Weatherall).
Contributors
Jody Azzouni Tufts University, Medford, MA, USA
Zvi Biener University of Cincinnati, Cincinnati, OH, USA
Katherine Brading Department of Philosophy, Duke University, Durham, NC,
USA
Sandro Caparrini Politecnico di Torino, Chestnut Hill, MA, USA
Robert DiSalle Department of Philosophy, Western University, London, ON,
Canada
Mary Domski University of New Mexico, Albuquerque, NM, USA
Katherine Dunlop University of Texas at Austin, Austin, TX, USA
xxxiii
xxxiv
Editors and Contributors
Craig W. Fox Edelstein Center for History and Philosophy of Science, Technology
and Medicine, Hebrew University of Jerusalem, Jerusalem, Israel
Allan Franklin University of Colorado, Boulder, Boulder, CO, USA
Michael Friedman Stanford University, Stanford, CA, USA
Niccolò Guicciardini Università degli Studi di Milano, Milano, Italy
Andrew Janiak Department of Philosophy, Duke University, Durham, NC, USA
Teru Miyake School of Humanities, Nanyang Technological University, Singapore, Singapore
Eric Schliesser University of Amsterdam, Amsterdam, the Netherlands
George E. Smith Tufts University, Medford, MA, USA
Monica Solomon Department of Philosophy, Faculty of Humanities and Letters,
Bilkent University, Ankara, Turkey
Howard Stein Department of Philosophy, University of Chicago, Chicago, IL,
USA
N. M. Swerdlow University of Chicago, Chicago, IL, USA
California Institute of Technology, Pasadena, CA, USA