¹ The first important distinction to make is that not all scientific realists would subscribe to this view of phenomena. Some scientific realists, for instance, have a more robust and sophisticated conception of phenomena than mere empirical manifestations of what there is. For instance, Bogen and Woodward (1988) have famously introduced the distinction between data and phenomena and argued that while data must be observable records of occurrences, phenomena need not be empirically accessible in any relevant sense. The main worry for scientific realists is to show how our inferential practices can bridge the gap between the two: if we are justified in believing in unobservable phenomena on the basis of reliable data, we can similarly be justified in believing in unobservable entities. Although Bogen and Woodward's analysis is arguably unorthodox in the realist panorama, it is nonetheless worth a specific mention. In Massimi (2007), I latched onto Bogen and Woodward's distinction between data and phenomena and expanded on it by showing how unobservable phenomena may appear in data models. However, the aim of my paper was not to defend traditional scientific realism. I wanted to defend instead a mild form of Kantianism, according to which we ‘construct’ phenomena in data models in such a way that the gap between the phenomena and the underlying entities is not as wide as both empiricists and realists have typically portrayed it.

² Another important distinction is in order here. Van Fraassen would not agree with my characterization of phenomena as empirical manifestations of what there is, and would maintain that phenomena are to all intents and purposes ‘objects’, observable and perceivable ‘objects’ (but not copies, or images, or empirical manifestations of objects). For instance, van Fraassen would argue that the phenomenon is the tree in front of my window, not the image of the tree reflected in a nearby pond: ‘If you see a reflection of a tree in the water, you can also look at the tree and gather information about the geometric relations between the tree, the reflection, and your vantage point. The invariances in those relations are precisely what warrant the assertion that the reflection is a picture of the tree. (…) But now you are postulating that these relations hold, rather than gathering information about whether that is so’ (2001), p. 160 (see on this point also van Fraassen 2008; I am very grateful to Carlo Gabbani for this quotation in his comments to Massimi (2007) at the seminar in Florence). Yet I think that there is some lingering ambiguity in this apparently very simple and intuitive characterization of phenomena. How can we say that we can look at the tree, and also look at the reflection of the tree in the water, without ‘postulating that these relations hold’? Or, how can we say that we have the image of an amoeba in a microscope, but we should not really ask whether this is the image, or empirical manifestation of ‘an amoeba’? Is not it tantamount to saying that we do have after all empirical manifestations of things—be they observable things, like a tree, or unobservable ones, like an amoeba—but that from an empirical point of view we should not ask what they are ‘empirical manifestations’ of? And why is the reflection of the tree in the water, or for that matter on my retina, any different from the reflection of the amoeba (through the microscope) on my retina? Once we start using these examples, I am not sure that we can consistently speak of phenomena as rough-and-ready observable/perceivable ‘objects’, as opposed to empirical manifestations of objects. But there is also another important distinction to be made. Constructive empiricists of course acknowledge that sometimes phenomena may manifest themselves in data models and hence that there may be an element of construction in the phenomena that science saves (see on this point the exchange between Teller (2001) and van Fraassen (2001)). However, constructive empiricists would maintain that despite this, we can still draw a distinction between observable phenomena, on the one hand, and whatever goes beyond the observable phenomena, on the other hand. As it will become clear later in this paper, the alternative Kantian conception of phenomena that I am going to propose, insofar as it stresses the non-ready made nature of phenomena, has the effect of blurring that very same distinction dear to constructive empiricists and realists alike. In other words, despite the fact that Kantians and constructive empiricists may share the idea that sometimes phenomena are something that we make (rather than ready-made in nature), nevertheless they draw very different epistemological lessons about what we should believe and how to bridge the gap between what we believe there is and what there is (which is precisely what the problem of knowledge is all about).

³ ‘Among these hypotheses there are some which save the phenomena by means of epicycles, others which do so by means of eccentrics, still others which save the phenomena by means of counterturning spheres devoid of planets. Surely, the gods’ judgment is more certain. But as for us, we must be satisfied to “come close” to those things, for we are men, who speak according to what is likely, and whose lectures resemble fables' Proclus In Platonis Timaeum commentaria. Quotation from Duhem (1908), English translation (1969), p. 21.

⁴ Duhem (1908), English translation (1969), pp. 21–2. Emphasis added.

⁵ Ibid., p. 105.

⁶ Ibid., p. 117.

⁷ Van Fraassen (2006), pp. 281, 287.

⁸ Kant (1936, 1938), English translation (1993), 22: 322.

⁹ The adjective ‘ready-made’ has been intentionally chosen to echo Goodman (1978) and Putnam (1982). This paper is indeed Goodmanian–Putnamian in spirit, in urging to abandon traditional realist views in favour of a Kantian one (although I am not going to subscribe to either Goodman's ‘worldmaking’ view or Putnam's ‘internal realist’ one).

¹⁰ Kant (1781, 1787). English translation (1997), Preface to the second edition, Bxx.

¹¹ Ibid., Bxviii.

¹² These fascicles are published under the title ‘How is physics possible? How is the transition to physics possible?’ (Ak 22: 282–452) in the English translation of the Opus postumum by Eckart Förster and Michael Rosen (1993). All citations are taken from this English translation.

¹³ Friedman (1992a), chapter 5, has illuminatingly pointed out how Lavoiser's chemical revolution, and the recent discoveries of pneumatic chemistry underlie and prompted the ‘Transition’, whose specific aim was to bridge the gap between the Metaphysical Foundations on the one hand, and the vast realm of empirical forces recently discovered, on the other hand (e.g. forces responsible for the solidification, liquefaction, elasticity, and cohesion of objects, which could not be accounted for within the Newtonian paradigm of MAN). On this specific issue, see also Pecere (2006).

¹⁴ Friedman (1992a), ch. 4.

¹⁵ Kant's second law of mechanics states that ‘Every change in matter has an external cause. (Every body persists in its state of rest or motion, in the same direction, and with the same speed, if it is not compelled by an external cause to leave this state)’, MAN, chapter 3, Proposition 3, ibid., p. 82. This law seems to encompass both Newton's I law (i.e. a body persists in its state of rest or uniform motion) and Newton's II law (F = ma) in requiring an impressed force F as the cause of any change of uniform motion into accelerated motion. But this is in fact questionable because it would require to show that Kant's second law entails Newton's second law; and, this is not evident, given Kant's vague assertion about the existence of an external cause of change of motion (I thank Roberto Torretti for raising this point).

¹⁶ Indeed, already in the General Observation to Chapter 2 of MAN, Kant had stressed the inadequacy of Cartesian physics and the superiority of the Newtonian ‘dynamical explanatory scheme’, not least because it avoids feigning hypotheses by contrast with the mathematical-mechanical scheme that ‘gives the imagination far too much freedom to make up by fabrication for the lack of any inner knowledge of nature’ Kant (1786), English translation (2004), p. 71. On the difference between the mathematical-mechanical scheme and the metaphysical-dynamical one, see again Friedman (1992a), pp. 180–3.

¹⁷ Indeed Kepler called his Astronomia nova ‘aitiologetos’ (I thank Roberto Torretti for pointing this out). However, to Kant's eyes, Kepler's view fell short of introducing the right sort of dynamic causes, namely those that could bridge the gap between kinematics and dynamics and pave the way to a mathematical physics of Newton's type, those same dynamic causes that could led us to infer a priori to a law such as Newton's law of gravitation. In other words, according to Kant, what Kepler did not have is the notion that external force does not cause just motion but change in motion (acceleration); whereas Galileo's description of free fall as uniformly accelerated motion (under a presumably constant force) contributed decisively to Newton's discovery, as we shall see below. On the Kepler–Newton relationship as Kant portrayed it in the ‘Transition’, see Caygill (2005).

¹⁸ Galileo (1638), English translation (1914), p. 162.

¹⁹ Ibid., pp. 166–7. Emphases added.

²⁰ Ibid., p. 169.

²¹ As Domenico Bertoloni Meli (2008) has pointed out, in the second and third day of Two New Sciences Galileo's main concern was with establishing an axiomatic science of motion on the example of Archimedes. Despite a voluminous historical literature in recent times on Galileo's experiments and machines, ‘his foundational efforts have attracted less attention, yet they constitute a major episode in the history of science.’ It was precisely in his life-long strive to achieve a formal axiomatic presentation of the new science of motion that in Two New Sciences Galileo was looking for a natural and self-evident principle from which to deduce his law of free fall (already found on experimental grounds in 1604).

²² The Italian ‘esperienza’ is translated in Crew and de Salvio as ‘experiment’. I translate it as ‘thought experiment’ instead because there is an element of idealisation as indicated by the verb ‘imagine’ in the following discussion about arcs reaching the horizontal plane (we are assuming that there is no air resistance, or friction, etc.).

²³ Ibid., p. 171–2.

²⁴ Ibid., p. 172.

²⁵ He imagines the flow of time between any initial and final instant A and B as a vertical line AB, in which we can identify some time intervals AD and AE. He then represented space with another vertical line going from H to I, such that the space interval HL is run through in the first time interval AD and the space interval HM in the time interval AE. How can he prove that HM:HL = AE²:AD²? He imagined another time line AC drawn from A at any angle whatever with AB. Suppose we now draw parallel lines that from points D and E intersect the new time line AC in O and P, respectively. The parallel line DO now represents the maximum degree of speed acquired at instant D of time AD, and EP the maximum degree of speed acquired at instant E of time AE. In the previous Theorem I, the so-called mean speed theorem, Galileo had proved that the time in which a certain space is traversed by a moveable in uniformly accelerated motion from rest is equal to the time in which the same space would be traversed by the same moveable carried in uniform motion whose degree of speed is one-half the maximum and final degree of speed of the uniformly accelerated motion. Then, he can now conclude that the spaces HM and HL are the same spaces that would be traversed in times AE and AD by a moveable in uniform motion whose degree of speed is one-half EP and DO (which represent the maximum degree of speed at instant E and D respectively). Therefore, the spaces HM and HL are in duplicate ratio of the times AE and AD. QED.

²⁶ ‘The method of the Two New Sciences is clearly not that of hypothesis, deduction and experiment in the modern sense. In fact, Galileo was quite unable to treat the principles of a demonstrative science as hypothetical for they must be true and evident’ Wisan (1978), p. 43.

²⁷ Feyerabend (1975).

²⁸ I intend here ‘a priori’ in the sense of being constitutive of the object of experience, which as Michael Friedman (2001a) has illuminatingly pointed out, is the relevant Kantian meaning of ‘a priori’ that still applies after Kant.

²⁹ It is worth noting in the demonstration above Galileo's interchangeable use of ‘momento’ and ‘impeto’, where by ‘impeto’ Galileo does not mean the Medieval impetus of Oresme and Buridan (i.e. an internal force keeping the projectile in motion). In Galileo, ‘impeto’ is almost synonymous with ‘momento’, and it is the product of a body's weight and speed. Already in the Pisan work Le meccaniche in 1597, working on balances, Galileo had defined the ‘momento’ as the propensity of a body to move downwards because of its weight and its position on the balance. In Koyré's words, ‘the impetus of the moving body is nothing other than the dynamic impulse given to it by its gravity’ Koyré (1939), English translation (1978), p. 185. As Hooper (1998), pp. 159–160, has illuminatingly pointed out ‘In motion on inclined planes, the momenta gravitatis, which are due to the angle of descent, are shown to be congruent to the momenta velocitatis given by the rules of speed, and are taken as the explanation and cause of the latter’.

³⁰ This Kantian moves is still vulnerable to the following objection: once we build causes in the phenomena (via suitable dynamic forces), we can eschew the underdetermination problem at the cost of facing another problem, namely that of explaining how we know what the phenomena are (I thank Peter Lipton for pointing this out). Of course, this is not much of a problem for Kant himself: his task was to justify retrospectively the universal and necessary validity of Newtonian physics. He knew (or, at least, he believed to know) what the phenomena were. But the problem remains for us, after the scientific revolutions of twentieth century physics and after Kuhn. From Newton to Einstein, the dynamical analysis of free falling bodies has changed; Newton's gravitation is not quite Einstein's gravitation, and quantum gravity may in turn be different from both. This raises of course very serious issues for any Kantian philosopher of science, and no wonder it has been a debated topic in the most recent literature. How to reconcile Kant with Kuhn (see Friedman 2001a for a possible answer to this question)? And was Kuhn himself right in ‘relativising’ Kantianism to paradigms or scientific lexicons? Can we really say with Kuhn that whenever a scientific revolution occurs, scientists ‘live in a different world’, presumably populated by different phenomena? I intend to investigate this further issue and the problems it raises in future research. I have intentionally left it out of this paper, because the aim of this paper was to analyse Kant's conception of phenomena itself, rather than its implications for scientific revolutions.

³¹ Friedman (1992b), p. 182.

³² Kant (1781, 1787), A645/B673 – A647/B 675.

³³ See on this point Friedman (1991), pp. 74–5.

³⁴ Friedman (1992a), p. 245, gives a penetrating analysis of how the chemical revolution revealed a gap in Kant's critical philosophy, in particular a gap between the top-down approach typical of the Metaphysical Foundations (which moved from Kant's transcendental principles to metaphysical principles of natural science, from which the empirical law of universal gravitation could then be derived) and the bottom-up approach typical of systematicity as a regulative principle of reflective judgment, which embraces the variety of empirically given forces in nature and strives to subsume them under higher-order concepts. According to Friedman, the ‘Transition’ project tried to reconcile the top-down approach with the bottom-up one, i.e. it tried to reconcile the constitutive aspect inherent the Metaphysical Foundations with the regulative aspect championed in the Critique of Judgment, and to show that these two opposite paths intersect at some point, namely that the increasing empirical variety subsumed under the principle of reflective judgment eventually leads up to the two fundamental forces of attraction and repulsion envisaged in the Metaphysical Foundations. Eckart Förster (2000) compares and contrasts Friedman's interpretation with that of Mathieu and Tuschling, and offers an alternative analysis for the role of systematicity in the ‘Transition’ as rooted in Kant's principle of a formal purposiveness of nature disclosed solely by aesthetic judgments of natural beauty, in continuity with the Critique of Judgment.

³⁵ See for details Opus postumum, 22:516. We reach at this point a controversial part of the ‘Transition’, where several passages suggest that Kant was actually taking the distance from Newton to the point of even rejecting the very same Newtonian expression ‘natural philosophy’ in favour of scientia naturalis. A propos of this, Caygill (2005), pp. 34–5 writes: ‘Kant questioned not so much the fact of Newton's dynamic history of planetary motion as its philosophical or scientific character (…). The Principia performs the amphiboly of making philosophy and metaphysics into a branch of mathematics rather than recognising that philosophy “must set the philosophical foundations prior to mathematical ones”.’ First of all, let me note that Caygill's analysis of Kant's on Newton in the XIth fascicle of the Opus postumum fundamentally agrees with my interpretive line in this paper about how Kant saw the transition from kinematics to dynamics operated by Newton, and the pivotal role that in Kant's eyes Newton played towards a science of moving forces in nature. So, it is not the fact of Newton's dynamics that is at stake here. What is at stake instead is the ‘philosophical’ character of Newton's Principia, according to Caygill, and why Kant saw his ‘Transition’ as a fundamental challenge to Newton's Principia. Let me just briefly note that I agree with Caygill that according to Kant the philosophical foundations must be set prior to the mathematical ones; but this is precisely what Kant had already done with respect to Newton's Principia back to MAN in 1786. What is at stake in the ‘Transition’ is the specific search for a transition to ‘physics’ intended as a system of empirically given moving forces in nature, which starting with a bottom-up approach could possibly meet at some point the top-down approach laid out in MAN. And I do not see this as a challenge to Newton's Principia, but more as a way of complementing and expanding on Newton's project.

³⁶ On this point, see Friedman (1992a), pp. 290–341, and Pecere (2006). In a forthcoming paper of mine jointly authored with Silvia De Bianchi, we investigate precisely the role that the aether played as a medium for the transmission of attractive and repulsive forces in an account of a variety of phenomena back in Kant's mid-1750s works (On Fire, Universal Natural History and Theory of Heavens, Physical Monadology, and New Elucidation). In particular, we reconstruct the tradition of mixed sciences (experimental physics and pneumatic chemistry with Boerhaave, Musschenbroek, s'Gravesande, Hales among others) as an important tradition somehow complementary to the Newtonian one (especially, the Newton of the II edition of Principia and of the Optiks), from which Kant's view of the aether as a medium of forces seems to have derived. There are some surprising analogies between some of Kant's early views on dynamics exposed in these 1750s works (in reference to specific chemical problems such as combustion, liquefaction, spirituous airs, etc.) and the views exposed at the end of his life in the ‘Transition to physics’, as if the late Kant felt the need to go back to some of the pressing physical problems which prompted his original philosophical investigations in the pre-critical period.