ARTICLE IN PRESS

Studies in History and Philosophy of Modern Physics 39 (2008) 687– 693

Contents lists available at ScienceDirect

Studies in History and Philosophy

of Modern Physics
journal homepage: www.elsevier.com/locate/shpsb

Essay review

Conceptual Foundations of Yang– Mills Theories

Alexandre Guay
CIRST, Université du Québec à Montréal, Case postale 8888, Succursale Centre-ville, Montréal, Québec, Canada H3C 3P8

Gauging What’s Real: The Conceptual Foundations of Contemporary Gauge Theories

R. Healey. Oxford University Press (2007), 320pp., £40.00, hardcover, ISBN 978-0-19-928796-3

1. The book and a few commentaries

Richard Healey’s Gauging What’s Real is something not frequently seen in the landscape of
philosophy of physics. It is not a book about the interpretation of quantum mechanics or quantum
field theory, nor is it about space and time or causality. Rather it focuses on the conceptual
foundations of specific theories of interaction, namely gauge theories of the Yang–Mills (YM) type
(including electrodynamics).1 Since three of the four fundamental interactions are modelled by
theories of this kind, the importance of the subject is obvious for the elaboration of a scientific image
of interaction. The book aims at a delicate balance of readability while being exhaustive. Most
subjects already present in the philosophical literature are discussed, although sometimes briefly.
Healey takes remarkable steps to make his book accessible to a large community of researchers. For
example, there are no less than six pedagogical appendixes. Not surprisingly, in this context many
questions would require a much more detailed discussion to satisfy the specialist. But as Healey says
himself: ‘‘[T]he book does not represent the last word—or even my last word—on this topic . . . If I
have conveyed a sense of its intellectual interest and importance, while provoking some reader to do
a better job, then I will be satisfied’’ (p. x). The relative rarity of books about fundamental interaction
theories gives us more than enough reason to forgive some short discussions. The introductory
chapter presents, among other things, the principal fibre bundle formalism. This geometrical
framework is the main conceptual item in Healey’s toolbox when discussing classical YM theories. It
is obviously a particularly appropriate tool in this context. After this first chapter follow three
chapters focusing mostly on classical YM theories. The first of these is about the Aharonov–Bohm
(AB) effect, a quantum phenomenon predicted by a model of quantized charged particles subjected to
the influence of a classical electromagnetic gauge potential. The importance of the AB effect should
not be underestimated. In the philosophical literature it is clearly one of the main empirical reasons
to reconsider the usual field ontology of classical electromagnetism. Moreover, even though a

E-mail address: [email protected]

1
On many occasions Healey makes valuable points about General Relativity (GR), another gauge theory. But GR is
obviously not his main target. Most of the time comments are made in order to contrast GR with YM theories. In this review I
discuss only YM theories.

1355-2198/$ - see front matter & 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.shpsb.2008.05.004
 ARTICLE IN PRESS
688 A. Guay / Studies in History and Philosophy of Modern Physics 39 (2008) 687–693

quantum effect, it has a significant impact on the interpretation of classical YM theories. The model
of a quantized charged particle (for example represented by a wave function) subjected to a classical
gauge potential can be modelled by a Uð1Þ principal fibre bundle. This geometrical construction is
also the natural setting to represent the classical Uð1Þ YM theory (the classical theory associated with
quantum electrodynamics). Therefore an interpretation of the AB effect can possibly be transposed to
the Uð1Þ YM theory and vice versa. Moreover a convincing interpretation of the Uð1Þ YM theory can
serve as a foundation for interpreting classical non-Abelian YM theories. With respect to the AB
effect, Healey proposes three interpretations, each transposable as an ontological interpretation
of all classical YM interactions: (1) no new electromagnetic properties view (for YM theories, no
gauge potential properties view), (2) new localized electromagnetic properties view (new localized
gauge properties), and (3) new non-localized electromagnetic properties view (non-localized gauge
potential properties view). For the AB effect, and for classical YM theories in general, Healey supports
the third position.
Healey’s main argument against the first position is that it seems to bring back action at a
distance, ostensibly a bad thing, to field theory (see the next section for a discussion about this
point). But also it seems that in non-Abelian theories there may be physically distinct situations in a
region even though the gauge field is the same throughout the region in each situation. New
properties seem needed to distinguish these situations. Since there is no empirical application of a
non-Abelian classical YM theory, this argument relies on a certain interpretation of the mathematical
formalism that could be contested.
In the recent literature numerous philosophers defend one version or another of the second
position, for example Leeds (1999) and Maudlin (1998). Most of Healey’s attacks on this position
consist in reframing these positions in the formalism of the principal fibre bundle. If this geometrical
formalism represents adequately classical YM theories (and we have no reason to believe the
contrary), it excludes the possibility of localized gauge properties. For a recent enlightening analysis
of the ontological implications of the principal fibre bundle formalism, see Catren (2008).
So it seems we are left with the third position. One of the more interesting theses of Healey’s book
is that the new non-localized gauge properties are captured by holonomies in the appropriate
principal fibre bundle. Thus, the holonomy interpretation of classical YM theory asserts that the
theory represents intrinsic holonomy properties of regions of space-time, each of which consists of
all points on a loop. If this interpretation is correct, the main metaphysical consequence is the
unavoidable non-separability of physical processes in classical physics.
Healey answers many objections to the holonomy interpretation but some questions remain:

(1) Holonomy is not the unique way to characterize the non-localized gauge properties. Other
representations can be formulated. For example, if the asymptotic boundaries are empty
Minkowskian we can use, as variables, gauge invariant line integrals coming from infinity. For
other boundary conditions other choices are possible. Does this freedom in representation have
an impact on the ontological conclusions?
(2) Holonomies for non-Abelian principal fibre bundles are not gauge invariant. As Healey explains, for
holonomies to capture intrinsic properties of a principal bundle, one or more reference points in a
particular fibre have to be chosen. The apparent space-time realism implied by the introduction of
loops and reference points in the interaction ontology could be problematic if we aim, like many
physicists, for a background free theory. The fact that the paradigmatic example of a locally defined
entity, the matter field, cannot be considered realistically, since it is not gauge invariant, makes one
suspicious about an interpretation relying on a realist position about space-time points.
(3) Knowing all this, are holonomies the appropriate variables to identify gauge orbits in the space of
histories of the theory (presumably the physical variables)? Should we not look for a way to
identify gauge orbits that involve, in a transparent way, matter degrees of freedom?

This said, even if we are suspicious about the loop interpretation, this does not in any way weaken
Healey’s claim that classical YM theories involve a non-localized interaction process. This is a
significant assertion.
 ARTICLE IN PRESS
A. Guay / Studies in History and Philosophy of Modern Physics 39 (2008) 687–693 689

The second half of Healey’s book concentrates on quantum YM theories. Healey discusses an
impressive number of subjects that often are presented for the first or almost for the first time in the
philosophical literature. Not surprisingly many sections feel too compact. Let us briefly expose the
main content of each chapter.
The fifth chapter is an introduction to the quantization of YM theories, mainly YM free field
theories. This chapter is a good introduction but does not allow the reader to grasp all the complexity
and the subtleties involved in the quantization of a gauge theory. For example, the fact that the BRST
symmetry is not discussed when ghost fields are introduced is a weakness. Also, only two pages are
devoted to theories containing interacting terms with matter fields. This is unfortunate since one of
the great strengths of Healey’s analysis of classical YM theories is that it does not loose sight of
matter fields. In fact this is one of the advantages of using the principal fibre bundle formalism in
which matter fields are easily represented.
Chapter six is a defence of the formal character of gauge symmetry. Familiar approaches to this
question are extensively discussed, like the possible observation of gauge transformations and the
empirical implications of the gauge argument. More interestingly, much less discussed approaches
are exposed and criticized: the empirical status of ghost fields, the possible gauge dependence of the
spontaneous gauge symmetry-breaking, the possible empirical distinction between large and small
gauge transformations and finally the status of anomalies. In each case Healey argues in a concise
way against the position that gauge symmetry is empirical.
The seventh chapter is an introduction to what a gauge reduced quantum YM theory would look
like. The version the author is presenting is based on a loop representation of the YM fields. Healey
forthrightly does not hide the problems encountered by this representation, especially when
interaction with a matter field is included. Nevertheless, the potential significance of such a
representation is not to be underestimated. Philosophers, as Healey is rightly pleading, should take a
particular interest in this area of research since these representations are presumably based on
physical—in other words, gauge invariant—variables.
YM quantum theories are models of the more general QFT. It is now a cliché to assert that the
ontology implied by QFT is not obvious. Could this problem be clarified if we focus on YM theories?
The author explores briefly this question in the eight and final chapter of the book, pushing his loop
interpretation in the framework of Bohm, Copenhagen and modal interpretations. This chapter is not
the strongest but it makes an interesting general point. Philosophers of physics have the tendency to
focus on the more general theories. Healey suggests that specific theories, like YM theories, can give
us a new perspective on interpretative questions about quantum physics. This chapter could induce
new research to clarify this possibility. The remainder of the book consists of a concluding chapter,
asserting the importance of acknowledging the possibility of non-localized properties, and of six
useful appendixes.
The next section of this essay discusses in more detail the interpretation of the AB effect. The third
section deals with the language in which questions about conceptual foundations are usually stated.

2. The Aharonov–Bohm effect

Most of the discussions about the interpretation of the Aharonov–Bohm effect follow a similar
path, for example Aharonov and Bohm (1959), Feynman, Leighton, and Sands (1964), Healey (1997)
and Belot (1998). Schematically, (1) since there is an observable effect (a phase shift in the
interference pattern) on quantized charged particles, something, let us call it X, must have acted on
them. (2) X cannot be the electromagnetic field because charged particles cannot reach non-zero
electromagnetic field regions. Action at a distance is excluded. (3) Therefore new electromagnetic
properties (localized or not) have to be invoked to explain the effect. This is why the AB effect is a
pure quantum effect. (4) Furthermore, once new properties are believed to be necessary, in order to
have a unified bearer of interaction, we interpret all electromagnetic interaction as the result of these
new properties. In consequence the electromagnetic field becomes a derived entity.
First a remark about this common argument. A reader might worry about (2) because the
localization of quantum particles is not a well-defined concept. However this proposition, already
 ARTICLE IN PRESS
690 A. Guay / Studies in History and Philosophy of Modern Physics 39 (2008) 687–693

present in Aharonov and Bohm (1959), can be expressed in a less problematic way. For example, in
the Feynman functional quantization method, the AB effect can be predicted by not including in the
sum over histories any path crossing a region where the electromagnetic field is not zero. In
consequence the effect does not depend on a local interaction with F mn but rather on a relative phase
shift between path contributions that depends on the electromagnetic flux. Therefore no direct
electromagnetic field interaction is involved in the production of the AB effect. Now I would like to
present an alternative, but complementary, line of reasoning that, to my knowledge, is absent from
the philosophical literature and from Healey’s book.
There are many ways to modify, with an electromagnetic field, an interference pattern obtained
by a two-slit scattering of charged particles. For example, (A) we can introduce an enclosed magnetic
flux between the slits. In this case we obtain a fringe shift with respect to the unperturbed pattern,
while the pattern envelope remains the same. This is a case of the magnetic AB effect. Or (B) we could
add a uniform magnetic or electric field behind the slits and then obtain a displacement of the
pattern without an envelope or relative phase change. Also (C) we could install an electromagnetic
source, like a light, behind the slits. If the source is sufficiently intense, it can destroy the interference
pattern. In this case the envelope and the pattern become identified. We could imagine other
examples, but these will suffice for my purpose. The point is that only in cases B and C it is a net
change, caused by the addition of an electromagnetic field, of average momentum and energy
recorded. In case A, momentum, energy and angular momentum are conserved as an average (Olariu
& Iovitzu Popescu, 1985). In B, a net transfer of transversal momentum explains the displacement of
the pattern. In C, the destruction of the interference pattern implies a strong enough electromagnetic
interaction in order for the charged particles to behave classically. From this point of view the AB
effect is a pure quantum (beyond classical) effect because it is a measurable effect that does not
require the transfer of a physical quantity that we usually associate with a classical interaction.
Before returning to the AB effect let us discuss briefly the ontological status of the
electromagnetic field in classical physics. The best reason we have to include a field in our ontology
is clearly provided by Einstein discussing the special theory of relativity:

We now shall inquire into the insights of definite nature which physics owes to the special theory
of relativity. (I) There is no such thing as simultaneity of distant events; consequently there is also
no such thing as immediate action at a distance in the sense of Newtonian mechanics. Although
the introduction of actions at a distance, which propagate with the speed of light, remains
thinkable, according to this theory, it appears unnatural; for in such a theory there could be no
such thing as a reasonable statement of the principle of conservation of energy. It therefore
appears unavoidable that physical reality must be described in terms of continuous functions in
space. (Einstein, 1970, p. 61)

For Einstein, some conservation principles are more fundamental than assumed ontology. Retarded
action as a mode of electromagnetic interaction violates conservation of energy and momentum
(Lange, 2002, Chapter 5). Furthermore, violation of conservation of energy in this context entails a
violation of determinism (Guay, 2004, Chapter 5). The introduction of the electromagnetic field in the
ontology of the theory solves all these problems. On the other hand, since a non-zero gauge potential
does not always carry energy it is not necessary to add this entity to the ontology. It would be too
strong to say that in classical physics to exist is to carry energy. Nevertheless, a very strong reason to
justify the introduction of a new entity into the ontology of a theory would be to save certain (local)
conservation principles that we believe to be essential. In the same way, action at a distance is
rejected not because it is unthinkable but rather because it endangers the conservation principles of
energy and momentum.
How is this discussion in classical physics relevant to the interpretation of the quantum AB effect?
First, the AB effect is modelled as the influence of classical electromagnetism on quantized charged
particles. The interaction is not theoretically understood in a quantized way. Second, as already said,
the AB setting is a model of a Uð1Þ principal fibre bundle, the geometrical setting of the Uð1Þ classical
YM theory. In many aspects the AB effect is giving us information about classical interaction in a
semi-classical context. Authors like Belot (1998), Lyre (2004) and Healey go as far to assert that the
 ARTICLE IN PRESS
A. Guay / Studies in History and Philosophy of Modern Physics 39 (2008) 687–693 691

AB effect is a good ground to discuss the ontology of classical electromagnetism. Consequently, in

this context an Einsteinian ontological approach is acceptable.
Since the AB effect does not imply a modification of average momentum or energy of particles, the
necessity to reject action at a distance as a mean of interaction is not clear in this case. Thus, an
electromagnetic interaction can take two forms:

 If there is a transfer of energy the bearer of the interaction is the electromagnetic field. This entity
is still necessary to guarantee the conservation principles even in semi-classical contexts (Olariu
& Iovitzu Popescu, 1985).
 If there is no average transfer of energy or momentum the interaction is mediated by a retarded
action at a distance of F mn (or possibly from the charges themselves).

So there are two available stories. If your interaction concept implies some kind of locality
condition that excludes action at a distance, the AB effect justifies the addition of new gauge
properties. If your concept of a bearer of interaction only implies the local conservation of energy you
have the option to not introduce new gauge properties and resort to retarded action at a distance. I do
not deny that the imposition of a locality condition on interaction seems reasonable but I emphasise
the point that a locality condition on interaction which is not grounded in the local conservation of
energy is not naturalistic and therefore must be grounded in some metaphysical conception of
causality. Such a discussion is not available in existing papers about gauge interaction.

3. The choice of a formal language

Faced with the plurality of not clearly related problems discussed in the second half of Healey’s
book, it seems appropriate to long for a Carnap-style philosophical approach. By that I mean the
reframing of the different problems and questions in the same ideal language that allows a clear
identification of their relations, to expose and eliminate pseudo-problems, and to provide tools to
transpose interpretative solutions from one theory to another. In fact, for quantum YM theories we
need the equivalent of what the principal fibre bundle formalism is doing for Healey when he
discusses classical gauge theories. This formal apparatus should satisfy a few requirements:

(1) Naturalistic: it should come from physics itself.

(2) Not realm specific: it should give a transparent transition from classical to quantum YM theories.
(3) General: it should be flexible enough to treat all applications of YM theory in quantum physics.

Let us briefly examine each of these requisites. The first is designed to comply with methodological
naturalism. The language for studying conceptual foundations of science should come from science
itself. Having said that, it is not necessary to adhere to naturalism to aim for a formal framework
coming from science. Recent history shows many examples where technical formalisms of scientific
origin enrich in a significant way philosophical discourse. For example, the use of geometrical models
in Earman (1989), of fibre bundles in Maudlin (2007) or of a mathematical approach to objectivity in
Nozick (2001). Moreover since concepts like gauge and field are heavily theory laden, a scientific
formal framework may be better suited to represent them.
The second requisite would allow discussing in the same framework interpretations of classical
and quantum YM theories. Ideally we want our philosophy of classical and quantum physics not to be
completely distinct from one another. Some difficulties seem specific to quantum physics, like
renormalization and anomalies, but a discussion of the variables of a gauge reduced classical theory
could clarify the interpretation of the associated quantum YM theory. If the passage from a classical
gauge theory to its quantized version is transparent enough it would be a clear advantage.
Finally, a good technical formalism must be general enough to discuss many theories in a similar
way. For example, a technical framework that would only be able to represent well Uð1Þ YM theory
and not non-Abelian YM theories would be interesting but could in fact mislead us. One requires that
 ARTICLE IN PRESS
692 A. Guay / Studies in History and Philosophy of Modern Physics 39 (2008) 687–693

the chosen technical formalism should be able to represent all YM theories but should also be a good
framework to examine more specific features like the vacuum, renormalization, and anomalies.
I believe two formal apparatuses fill all these requirements: Dirac quantization (DQ) based on the
representation of a YM classical theory as a constrained Hamiltonian2 and the Feynman functional
sum over histories quantization (FQ) which starts with a Lagrangian representation of the classical
YM theory. Both formalisms come from physics itself. Both formalisms express the transition from
classical to quantum theory in a relatively clear manner. Schematically, DQ is an extension of
canonical quantization techniques. In the classical theory the gauge transformations correspond to
first-class Hamiltonian constraints. During the quantization process the first-class constraints are
promoted to operators on a Hilbert space that can identify vectors in the physical sector. In FQ the
quantization process consists in producing a functional sum over possible classical histories of the
studied system. Each history contributes to a probability amplitude through its associated action.
I will return to this shortly. Finally, both formalisms are general in two senses. First, they can
accommodate all YM theories and provide a representational framework to discuss more specific
questions, like the status of the vacuum. Second, they can serve as a formal language to formulate an
interpretation of quantized gauge theories. In their actual form, these formalisms are far more than
simple quantization tools. Unfortunately, one would not notice this reading Healey’s book. Since the
landmark publications of Dirac (1950) and Feynman (1948), these formalisms became sophisticated
apparatuses able to frame in an interesting way most of our questions about YM theories. See for
example Henneaux and Teitelboim (1992) or DeWitt (2003).
It is not the place here to discuss in depth all questions that can be clarified by adopting one or the
other of these formalisms. For examples about the use of DQ I refer the reader to Earman (2003) and
Pons (2005). As for FQ, since this formalism is almost absent from philosophical literature I will
briefly present two examples where FQ can help to clarify some aspects of quantum YM theories. As
the first example let us discuss briefly the status of holonomies in quantum YM theory. Healey pleads
for a holonomy interpretation of classical YM theories. He also maintains that the holonomy concept
is essential to understand the AB effect because this phenomenon can be modelled as a Uð1Þ principal
fibre bundle. The FQ formalism can help to defend this point and to generalize it. In the context of
non-relativistic quantum mechanics the transition probability amplitude of a system, let us say of
one charged particle, is computed by summing the phase contributions of all possible trajectories
(histories). If we add electromagnetism to the picture we can note that the relative change of phase
between two paths, caused only by electromagnetic interaction, is exactly equal to the holonomy
computed for a closed curve formed by the two already mentioned paths, with the electromagnetic
potential playing the role of the connection. In other words, since in FQ a quantum phenomenon is
the result of the interference between contributions of different histories, holonomy is the relevant
entity to qualify in general electromagnetic interaction in semi-classical contexts.
How to extend this reasoning to a fully quantized YM theory with matter fields is not obvious.
However, if we confine ourselves to models built on a lattice, Wilson (1974) has shown that the gauge
interaction can in significant contexts be characterized by the trace of what, in a geometrical setting,
would be called holonomies. As a consequence holonomies are not just a nice tool to characterize
classical gauge properties but are also a fundamental mathematical entity to characterize gauge
interaction in non-relativistic quantum physics and in certain models of quantum field theory on a
lattice.
The second example is more technical. To quantize a non-Abelian YM theory is a tricky business.
First, we have to get the gauge surplus under control. If we just fix the gauge the obtained quantized
theory is not unitary. As explained by Healey (p. 167) one of the convenient ways to quantize a non-
Abelian YM theory is to add to the gauge fixed Lagrangian terms involving new fields, called ghosts,
that guarantee unitarity and renormalizability.3 Physicists consider these fields to be fictive. Using
the FQ formalism, Bryce DeWitt proposes a proof that shows that this quantized theory (with gauge

2
Healey uses the constrained Hamiltonian formalism to argue effectively for certain points but he does not do this
systematically.
3
This procedure is equivalent to imposing a new global symmetry, called the BRST symmetry, to the YM theory (Becchi,
Rouet, & Stora, 1976). The role of this symmetry in classical YM theory is discussed in Catren (2008).
 ARTICLE IN PRESS
A. Guay / Studies in History and Philosophy of Modern Physics 39 (2008) 687–693 693

fixing and ghost terms) is equivalent to an eventual quantization of a gauge reduced theory.4 In other
words, the added degrees of freedom coming from the ghost fields are just what is needed to
neutralize the effect in the quantum theory of the added degrees of freedom coming from the gauge
surplus. To my knowledge such a proof exists only in the FQ. The implications of this proof for the
status of the gauge principle, as a pragmatic principle, are discussed in Guay (2008).
Even if this Carnap style approach is appealing, an important question remains about the
compatibility of these two frameworks. If it is proven that, starting with the same classical theory,
they do not produce compatible quantum theories, a choice will have to be made. But in the mean
time both should be exploited by philosophers in a systematical way.

4. Conclusion

In spite of the few already mentioned weaknesses, reading Gauging What’s Real is very
stimulating. Since the measure of any book is the questions it makes us ask, Healey’s work is
exemplary. The diversity of problems and approaches discussed in the book will undeniably provoke
responses and comments from many readers. Is this not what a philosophy book should do?

Acknowledgements

The writing of this essay was supported by a postdoctoral grant from the Canada Research Chair in
the History and Sociology of Science.

References

Aharonov, Y., & Bohm, D. (1959). Significance of electromagnetic potentials in the quantum theory. Physical Review, 115(3),
485–491.
Becchi, C., Rouet, A., & Stora, R. (1976). Renormalization of gauge theories. Annals of Physics, 98(2), 287–321.
Belot, G. (1998). Understanding electromagnetism. British Journal for the Philosophy of Science, 49(4), 531–555.
Catren, G. (2008). Geometric foundations of classical Yang–Mills theory. Studies in History and Philosophy of Modern Physics,
doi:10.1016/j.shpsb.2008.02.002.
DeWitt, B. (2003). The global approach to quantum field theory (Vol. 1). Oxford: Oxford University Press.
Dirac, P. A. M. (1950). The Hamiltonian form of field dynamics. Canadian Journal of Mathematics, 3, 1–23.
Earman, J. (1989). World enough and space-time. Cambridge, MA: MIT Press.
Earman, J. (2003). Tracking down gauge: An ode to the constrained Hamiltonian formalism. In K. Brading, & E. Castellani (Eds.),
Symmetries in physics: Philosophical reflections (pp. 140–162). Cambridge: Cambridge University Press.
Einstein, A. (1970). Autobiographical notes. In P. A. Schilpp (Ed.), Albert Einstein: Philosopher-scientist (pp. 1–95). Cambridge:
Cambridge University Press.
Feynman, R. P. (1948). Space-time approach to non-relativistic quantum mechanics. Reviews of Modern Physics, 20(2), 367–387.
Feynman, R. P., Leighton, R. B., & Sands, M. (1964). The Feynman lectures on physics: Mainly electromagnetism and matter.
Reading, MA: Addison-Wesley Publishing.
Guay, A. (2004). Symétrie: Réflexions sur les formes naturelles. Doctoral dissertation, Université de Montréal.
Guay, A. (2008). A partial elucidation of the gauge principle. Studies in History and Philosophy of Modern Physics, 39(2), 346–363.
Healey, R. (1997). Nonlocality and the Aharonov–Bohm effect. Philosophy of Science, 64(1), 18–41.
Henneaux, M., & Teitelboim, C. (1992). Quantization of gauge systems. Princeton, NJ: Princeton University Press.
Lange, M. (2002). An introduction to the philosophy of physics. Oxford: Blackwell Publishers.
Leeds, S. (1999). Gauges: Aharonov, Bohm, Yang, Healey. Philosophy of Science, 66(4), 606–627.
Lyre, H. (2004). Holism and structuralism in Uð1Þ gauge theory. Studies in History and Philosophy of Modern Physics, 35,
643–670.
Maudlin, T. (1998). Healey on the Aharonov–Bohm effect. Philosophy of Science, 65(2), 361–368.
Maudlin, T. (2007). The metaphysics within physics. Oxford: Oxford University Press.
Nozick, R. (2001). Invariances: The structure of the objective world. Cambridge, MA: Harvard University Press.
Olariu, I., & Iovitzu Popescu, I. (1985). The quantum effects of electromagnetic fluxes. Reviews of Modern Physics, 57(2),
339–436.
Pons, J. M. (2005). On Dirac’s incomplete analysis of gauge transformations. Studies in History and Philosophy of Modern Physics,
36, 491–518.
Wilson, K. G. (1974). Confinement of quarks. Physical Review D, 10(8), 2445–2459.

4
The proof can be found in DeWitt (2003, Chapter 24).