- Robert Hannah
Time, or the passage of time, was told through a variety of means in antiquity—via one’s own body, through the actual or calculated movement of celestial bodies (sun, moon, and stars), and by means of artificial instruments, including sundials, water clocks, and various forms of timers. While the natural or built environments could provide large-scale, immobile backdrops to aid with telling the time, there were also miniature instruments that could be carried by hand around the known world with remarkable confidence in their accuracy. And while the simplest form of timing might be provided by one’s own body—such as through its hunger or its shadow—there were also artificial mechanisms of such extraordinary ingenuity and complexity that their like would not be seen for another millennium, and whose remains still elude complete explanation.
- Science, Technology, and Medicine
Updated in this version
Text and bibliography expanded to reflect current scholarship. Images, keywords, and summary added.
The Body as a Clock
At the popular level, marking time in the day could utilize simply the shadow cast on the ground by a person and measured by the person’s own feet (cf. Ar. Eccl. 651–652) (see Figure 1). Fixed shadow lengths to represent the time when something should be done would lead throughout the year to a variable time (in modern terms) for that activity, something that may have suited people’s biological clocks, since the length of daylight itself varied throughout the year with the seasons. Greek shadow tables surviving from the Hellenistic period to Byzantine times show these foot measurements formalized into seasonally sensitive shadow lengths that were organized according to the zodiacal or calendar months.
Another form of natural timer—the body’s hunger—is mentioned by satirists who disparage sundials as slave-driving instruments that force people to eat only at fixed times (Gell. NA 3.3.5. quoting from Plautus, three hundred years earlier). The comedic trope must have originated in the Hellenistic period when manufactured sundials were becoming more common.
So long as the sun is shining, a human figure on its own and of whatever height can serve as a pointer to indicate the time of day. As we have seen, it may do this as an entirely personalized and mobile clock, since the shadow cast by the body is a function of the height of that particular body. Each person could pace out their own cast shadow and measure its length in terms of their own feet. One person’s “ten feet” would then tell the same time of day as another person’s “ten feet,” even though the length in standard terms, such as our modern meters, would differ between people of different metrical heights. Everything is assumed to be proportional, so a taller person will necessarily cast a longer shadow than a shorter person, but he/she may also be assumed to have longer feet, so that their longer shadow will still comprise ten of his/her longer feet.
A stick, or similar instrument, could serve as a proxy for all individuals. If it were set up in a communal or otherwise public space, it would lend itself to the development of a common or standardized notion of time, not dependent on any individual person’s height. Every passerby could consult this communal shadow and agree on the indicated time. Such an instrument is called a gnomon.
The first documented use of a gnomon by the Greeks is associated with Anaximander and therefore situated in the first half the 6th century bce. Our source, Diogenes Laertius, indicates that it was used to display the time of year rather than the time of day (Diog. Laert. 2.1). So it would appear that the gnomon was developed by the Greeks initially as a calendrical tool to mark out the passage of time in the year, and not as a clock signalling the time of day.
Diogenes Laertius says that Anaximander set his gnomon “on the shadow-catchers (ἐπὶ τῶν σκιοθήρων)” in Sparta, “showing the solstices and equinoxes” (cf. Suda, Ἀναξίμανδρος Α 1986). We know that, by the late 5th century bce. the astronomer Oinopides had adopted the term γνώμων to refer to a perpendicular line when drawn to another straight line from an external point (Procl. In Eucl. p. 283, 7–10), and so we can infer that in Anaximander’s time a gnomon was a shadow-casting rod, of a more-or-less permanent material, placed vertically in the ground. The site of the shadow-catchers (τὰ σκιόθηρα) in Sparta might be associated with the linguistically related Skias (Σκιάς), a construction mentioned by Pausanias (3.12.10) as standing in the public assembly area of Sparta. Both share the root σκια in their names, connecting them with shadow in some way. The Skias is unlikely to have been a building for the Spartan assembly, as no traces have been found archaeologically, and the assembly seems to have met on open ground. Instead it has been proposed that the Skias might have been a pillar that naturally cast a shadow on the assembly area and also carried on its top a smaller object that served as a gnomon to better define the end of the shadow.1
A gnomon of this form could certainly provide a measure of the shadow lengths, both through the day and through the year. Indeed, shadow-catcher (σκιαθήρας) is itself specifically identified by Vitruvius as a type of a sundial, comprising an upright gnomon set up on a flat surface (De arch. 1.6.6). However, we are not told that Anaximander’s gnomon displayed anything except the times of the solstices and equinoxes. A midday shadow-line (which we call nowadays a meridian line) runs north to south, and it can indicate the time of the solstices—that is, the northern and southern limits of the sun in the sky. The summer solstice, when daylight is longest and nighttime shortest, would coincide with the southern end of the line closest to the gnomon, while the winter solstice, when daylight is shortest and nighttime longest, would occur when the shadow struck the northern end of the line furthest from the gnomon. A single point midway between these extremes would do double duty for the spring and autumn equinoxes, when day and night are almost equal. This meridian line therefore acts as a calendar, marking time for the sun’s course through the year, and hence through the seasons of the year, via the three date points. More elaborate forms of such lines would later increase the points to include divisions between twelve months of the year, identified often with the passage of the sun through the signs of zodiac. Large-scale examples survive from Roman imperial times, most famously in the form of an inlaid bronze meridian line set up by Augustus in the Campus Martius in Rome in 10/9 bce and mentioned by Pliny (HN 36.72). This used as its gnomon an obelisk over 30m tall, brought from Egypt and re-erected in Rome to commemorate Augustus’s takeover of Egypt in 30 bce and his correction of the Julian calendar.
Herodotus notes that the Greeks learned of the gnomon from the Babylonians (Hdt. 2.109.3). He also attributes to the same source Greek knowledge of “the twelve parts of the day.” He still lacks a specific word for hour, and it is not until the first half of the 4th century bce, in the Hippocratic medical texts, that we find the word for “season,” ὥρα, being adopted to express “hour.”2 It is from the second half of the same century that the earliest Greek sundials survive, and these include lines to mark out the hours.
In antiquity, there were two types of hours: seasonal hours and equinoctial hours. The former change in length with the seasons, as they are a twelfth of the period of daylight, which itself changes with the seasons in temperate zones. Equinoctial hours, on the other hand, are equal in length, representing a twelfth of daylight at the equinoxes, when daytime and nighttime are practically equal. Seasonal hours became the norm on sundials, while equinoctial hours became the preserve of astronomers, who used them because they simplified calculations of celestial motions over longer periods of time. They are the hours that we still use nowadays.
The earliest surviving Greek sundials are two fragmentary plane dials, one from Olympia and the other from the Amphiareion at Oropus in Attica. Both are flat blocks of marble incised with hour lines that radiate in a fan-shaped formation from the gnomon point to demarcate the start and end of the twelve hours of daylight.3 In appearance, the two dials resemble the much older Egyptian vertical dial type, and Egypt might be the ultimate source of influence for the Greek type of plane sundial. However, whereas the Egyptian vertical sundial displayed seasonal hours, the two Greek dials show equinoctial hours. The Olympia dial displays only the hours, while the Oropus sundial adds the date lines for the solstices and equinoxes, and furthermore includes an inscription that explains these lines, which suggests that the lines are a novelty. So the Olympia dial type, by lacking date lines, might predate the Oropus form.
The plane sundial, which occurs usually in horizontal or vertical forms but also in a tilted equatorial form as at Oropus, is the easiest type to construct but technically the most difficult to mark out, because the hemispherical dome of the sky has to be projected on to a flat surface. The sundials from Olympia and Oropus belong to this type. The plane type has a very long life: Figure 2 shows a vertical example from the Byzantine period (9th–10th centuries ce), in which the hours are numbered, according to the Greek fashion, by letters of the alphabet: Α = 1, Β = 2, Γ =3, Δ = 4, Ε = 5, Ƨ = 6, Ζ = 7, Η = 8, Θ = 9, Ι = 10, ΙΑ = 11, ΙΒ = 12. This simple form would be placed upright against a south-facing wall to capture the sun.
More sophisticated versions of plane sundials appear on the Tower of the Winds in Athens, built by Andronicus from Cyrrha in Macedonia and dated to the 2nd century BCE. Here the dials were incised one on each of the eight sides of the octagonal building, and they vary in format so as to enable the capture of sunlight on the shadow-casting gnomons on each side through the day (Figures 3 and 4). These are among the earliest surviving examples of the vertical plane type, but their complexity and accuracy imply an older ancestry.4
By the 2nd century bce. a wide variety of types of sundial was available to the Greeks and Romans. Vitruvius lists fourteen types, along with their supposed inventors, the earliest being Eudoxus (De arch. 9.8.1). The principal kinds surviving in the archaeological record are spherical, cylindrical, conical, and plane. Other types are known, including eventually small, portable ones, which were functional over much of the Roman empire. In her ground-breaking and still relevant study, Gibbs (1976) recorded 256 sundials.5 In an initial analysis, Schaldach (1998) listed over 340. Bonin (2015) recorded 586 sundials, as well as numerous representations of them in Greek and Roman art.6 Now, about 879 have been entered into an online database, classified into six types and more than 30 subcategories, at the Berlin research project “Ancient Sundials.”7
Two types of sundials, the spherical and the conical, dominate production from the Hellenistic period onward. These present dial faces whose geometry encourages, in practical terms, subdivisions that create unequal seasonal rather than equinoctial hours.
The spherical type of sundial entailed carving out usually a quarter-sphere of stone (Figure 5). It was theoretically the simplest to mark out, because it captured the celestial dome inversely onto a matching concave surface. Its gnomon hung out over the hollow part-sphere. It is generally assumed that Vitruvius refers to this spherical type when writing about the scaphe or hemisphaerium—the latter name particularly recommends the identification (De arch. 9.8.1). Typically, such sundials present two date lines marking out the winter and summer solstices, while an intermediate line, which serves for both equinoxes, is often marked out in between. A lack of care about the precise positions of these lines is not unusual. Other lines may be added for other days in the year, sometimes for the change of months, sometimes for particular festivals. The hour lines are sometimes also numbered in Greek fashion: Α Β Γ Δ Ε S Ζ Η Θ Ι ΙΑ ΙΒ, as in Figure 5.8
A variation on this type is the roofed spherical sundial (Figure 6). Instead of casting a shadow via a gnomon, this sundial allowed sunlight to penetrate through a small hole carved in the roof of the hemispherical dial and so cast a spot of light on the shadowed, concave interior. The Berlin database BSDP has also started to include digital 3D images of some of the sundials, a facility recently exploited in a study of the geometry of the roofed spherical sundial.9 The formal similarity between roofed sundials and the Pantheon in Rome has led several scholars to see in the latter a time-related function resulting from the infiltration of sunlight through the oculus of the building’s domed roof.10
The conical type of sundial competes in the surviving archaeological record with the spherical as the most popular type (Figure 7). It provides a simpler, partial hollowing out of the stone block. The earliest surviving example is from Heraclea ad Latmum in Turkey.11 An inscription tells that it was dedicated to a king Ptolemy by an Apollonios, son of Apollodotos, and was made by Themistagoras, an Alexandrian. Once thought to date to the end of the 3rd century bce, because of a supposed association with the mathematician Apollonios, who wrote on conic sections, it is now considered likely to belong to the first half of the century, and the king to be Ptolemy II, Philadelphus (reigned 284–246 bce). The conical sundial was much easier to construct in stone than the spherical, because its generating line was straight, not curved, while its theoretical underpinning was kept to a minimum—and indeed obviously simplified, or even not understood by many of the makers, to judge by their inaccuracies.
Even among miniature portable sundials, there is complexity.12 Dating to the 1st century ce is a portable sundial from Herculaneum.13 Known as the “Ham Dial” because its distorted bronze plate looks just like a small leg of ham, it consists of a spike on one side, which threw a shadow on to a series of crisscrossing lines on the plate, from which one could read the hour of the day. Other portable bronze dials are circular in shape, and some come with extra plates to suit different latitudes. One small dial of perhaps mid-3rd century ce date consists of just two plates and a gnomon, yet permits the reading of the time of day anywhere between latitudes 30˚ and 60˚ N; 30 locations are specifically listed.14
Some of these dials are simply very small versions in limestone of the stone spherical or conical types.15 On one of the smallest and earliest of such miniatures, an ivory conical dial, measuring only 2.8cm in height and dating probably to the 1st century bce, the accuracy of the hour lines is remarkable: it was made for a latitude of 33˚, which corresponds reasonably well with the latitude of its provenance, Tanis in Egypt at 31˚.16 It was found in the private house of perhaps an official who worked at the nearby temple of Amun. Fixed in place, the dial appears to have been used by a temple official to keep him on time in his duties. Not all portable dials were intended to be taken far.
One other major form of timing mechanism existed in antiquity: the water clock. In Greece the name given to the instrument was κλεψύδρα (clepsydra: “water thief”), a term apparently borrowed from a device that worked like a large, bulbous pipette (e.g., Arist. Cael. 294b.14–30; Arist. [Pr.] 914b). The name clepsydra was simultaneously applied to different types of containers that were used at least from the 5th century bce so as to time activities, notably speeches in the law courts (cf. Ar. Ach. 694, Vesp. 93). By the mid-4th century bce. the orator Demosthenes could use the word “water” (ὕδωρ) as a synonym for time itself (Dem. Poly. 2; De cor. 139; De fal. leg. 57). An example of a judicial clepsydra has been excavated in the Agora in Athens, dating to ca. 400 bce: it is a bucket-like vase with a piped hole near the base for the outflow and marked on the outside with the Greek letter χ twice (ΧΧ, i.e., two choes), and was found by experiment to hold 6.4 liters or the equivalent of just six minutes’ worth of water, a refinement of time unattainable by sundial.17 In the Roman law courts also, a speech for the prosecution or defense was timed against the water clock (cf. Plin. Ep. 2.11.14).
The standardization of time through the clepsydra extended also to the whole of the legal day in Athens, in those cases that demanded more time. This so-called “measured day” (ἡμέρα διαμεμετρημένη) was made to correspond, regardless of the time of year, to the length of the shortest days of the year ([Arist.] Ath. Pol. 67.4; Harp. s.v. ἡμέρα διαμεμετρημένη). This was then subdivided into a certain quantity of water, with thirds being given to the prosecution, the defense, and the judgment.
In Rome, Scipio Nasica was credited with introducing the first water clock, in 158 bce, just a few years after the first accurate sundial was set up in the Forum (Plin. HN 7.214). The clepsydra divided the hours of the day and night equally, Pliny records, and it can therefore be classed, along with its nearby sundial, as a clock, allowing the time to be told through the day. Julius Caesar measured the length of the nights in Britain “by water,” presumably with clepsydrai (Caes. BGall. 5.13.4). Clepsydrai of some kind were also used to measure out the length of night watches in the military world. Attempts were made to ensure that watches, and hence responsibilities, were equalized by adjusting the volume of water that the clocks held through coating the inner surface of the clepsydra with varying layers of wax (Aen. Tact. 22.24–25), and the popular tables of the lengths of day or night in terms of hours on the longest day reported by, for instance, Geminus (6.7–8), Strabo (2.5.38–42), Pliny (HN 2.186), and Cleomedes (2.1.438–451), may still reflect the results of this older technology.
Two very large outflow water clocks with a capacity of about 1,000 liters have been found in the Athenian Agora and the Amphiareion at Oropus, dating to the second half of the 4th century bce. Both specimens would probably take about 17 hours to empty, long enough to operate uninterruptedly over the whole of a summer’s day of under 15 hours. At Oropus, the clock may have been used to time rituals or performances at the nearby theatre. Dramatic plays were apparently timed before Aristotle’s time by the clepsydra, in this case a container presumably on a larger scale (Arist. Poet.1451a7–9).
An alternative approach to calibrating time against the seasonal hours was to tie the passage of time explicitly to the stars through what Vitruvius called an anaphoric clock. The term “anaphoric” comes from the Greek for the rising of a star, “anaphora” (ἀναφορά). An anaphoric clock was a mechanical device that told the time via an automated representation of the stars, indicated by images of the constellations that rise in sequence through the day. The automation could be driven most readily by a stream of constantly running water. Vitruvius is our principal source for the description of this form of clepsydra (De arch. 9.8.8–15). Physical remains of such complex machinery are rare. Fragments have been found of the star-plates of two or three such water clocks, dating perhaps to the 2nd century ce.18 Inside the Tower of the Winds in Athens, there are traces on the floor of channels for piping water from a reservoir set at the back of the building. On the basis of Vitruvius’ description of such water clocks, this piping has been interpreted as servicing a water clock.
The Antikythera Mechanism
One final instrument deserves special mention, as it combines in a space no bigger than a modern shoebox a great variety of means of marking time. This is the so-called Antikythera Mechanism, dating to the second half of the 2nd century bce, contemporary with the Tower of the Winds.19 Its fragmentary remains have revealed under x-rays and CT scanning that the Mechanism managed to correlate, in a system of about 30 geared wheels, the motions of the sun and the moon via the 19-year Metonic Cycle and probably of the five planets known to antiquity in epicyclic motion through the zodiac. The instrument could also be used to compute eclipses, and it had a dial to signal the two- and four-yearly games festivals at Olympia, Isthmia, Dodona, and Delphi. A parapegma, or star calendar, also coordinated with dials giving the zodiacal year, the Egyptian calendar, and even a civil calendar. It was probably a prestige item for a wealthy patron who lived in the Doric Greek-speaking world, perhaps in the sphere of influence of Corinth, a city famous for bronze-working.
Links to Digital Materials
- Freeth, Tony. “The Antikythera Mechanism: A Shocking Discovery from Ancient Greece.” Stanford, CA: Stanford Humanities Center, 2016.
- Gerd Graßhoff, Elisabeth Rinner, Karlheinz Schaldach, Bernhard Fritsch, Liba Taub. Ancient Sundials. 2016, Edition Topoi.
- Bonnin, Jérôme. La Mesure du Temps dans l’Antiquité. Paris: Les Belles Lettres, 2015.
- Freeth, Tony, Yanis Bitsakis, Xenophon Moussas, John H. Seiradakis, Agamemnon Tselikas, Helen Mangou Mary Zafeiropoulou, et al. 2006. “Decoding the Ancient Greek Astronomical Calculator Known as the Antikythera Mechanism.” Nature 444 (November 30, 2006): 587–591.
- Frischer, Bernard, John Fillwalk, Paolo Albèri Auber, David Dearborn, Mika Kajava, and Stefano Floris. “Edmund Buchner’s Solarium Augusti: New Observations and Simpirical Studies,” Rendiconti della Pontificia Accademia Romana di Archeologia. Serie III 89 (2016–2017): 3–90.
- Gibbs, Sharon L. Greek and Roman Sundials. New Haven: Yale University Press, 1976.
- Graßhoff, Gerd, Elisabeth Rinner, Karlheinz Schaldach, Bernard Fritsch, and Liba Taub. Ancient Sundials. Berlin: Edition Topoi, 2016.
- Hannah, Robert. Time in Antiquity. London: Routledge, 2009.
- Hannah, Robert. “Greek Government and the Organization of Time.” In Companion to Ancient Greek Government, ed. Hans Beck, 349–365. Oxford: Blackwell, 2013.
- Hannah, Robert. “Time-Telling Devices.” In Companion to Greek Science, Medicine, and Technology, ed. Georgia L., Irby, 923–940. Oxford: Wiley-Blackwell, 2016.
- Haselberger, Lothar. The Horologium of Augustus: Debate and Context. Supplement 99. Portsmouth, RI: Journal of Roman Archaeology, 2014.
- Jones, Alexander, ed. Time and Cosmos in Greco-Roman Antiquity. Institute for the Study of the Ancient World at New York University. Princeton, NJ: Princeton University Press, 2016.
- Jones, Alexander. A Portable Cosmos: Revealing the Antikythera Mechanism, Scientific Wonder of the Ancient World. Oxford: Oxford University Press, 2017.
- Price, Derek John De Solla. Gears from the Greeks: The Antikythera mechanism, a calendar computer from ca. 80 B.C. Transactions of the American Philosophical Society 64. Philadelphia: American Philosophical Society, 1974.
- Schaldach, Karlheinz. Römische Sonnenuhren: eine Einführung in die antike Gnomonik. Frankfurt am Main, Germany: H. Deutsch, 1998.
- Schaldach, Karlheinz. Die antiken Sonnenuhren Griechenlands: Festland und Peloponnes. Frankfurt am Main, Germany: H. Deutsch, 2006.
- Schaldach, Karlheinz. “Zu den Sonnenuhren des Andronikos.” In Der Turm der Winde in Athen, ed. H. J. Kienast, 197–226. Wiesbaden, Germany: Reichert Verlag, 2014.
- Schaldach, Karlheinz. “Measuring the Hours: Sundials, Water Clocks, and Portable Sundials.” In Time and Cosmos in Greco-Roman Antiquity, ed. Alexander Jones, 63–94. Princeton, NJ: Princeton University Press, 2016.
- Talbert, Richard J. A. Roman Portable Sundials: The Empire in Your Hands. Oxford: Oxford University Press, 2017.
- Winter, Eva. Zeitzeichen: Zur Entwicklung und Verwendung antiker Zeitmesser. Berlin: De Gruyter, 2013.
1. Philip Thibodeau, “Anaximander’s Spartan Sundial,” The Classical Quarterly 67 (2017): 374–379.
2. V. Langholf, “Hōra = Stunde: zwei Beilage aus dem Anfang des 4. Jh. v. Chr.,” Hermes 101 (1973) 382–384.
3. Olympia, National Museum S 373, and Oropus, Archaeological Museum A392: K. Schaldach, “Measuring the Hours: Sundials, Water Clocks, and Portable Sundials,” in Alexander Jones, ed., Time and Cosmos in Greco-Roman Antiquity (Institute for the Study of the Ancient World at New York University; Princeton, NJ: Princeton University Press, 2016): 65–69, Figs. III–2 (Olympia), III–5A,B (Oropos).
4. Karlheinz Schaldach, “Zu den Sonnenuhren des Andronikos,” in H. J. Kienast, Der Turm der Winde in Athen (Wiesbaden, Germany: Reichert Verlag, 2014): 197–226. See also Gerd Graßhoff, Elisabeth Rinner, Karlheinz Schaldach, Bernhard Fritsch, Liba Taub, Ancient Sundials
6. Karlheinz Schaldach, Römische Sonnenuhren: eine Einführung in die antike Gnomonik (Frankfurt am Main: H. Deutsch, 1998); and Jérôme Bonnin, La Mesure du Temps dans l’Antiquité (Paris: Les Belles Lettres, 2015).
9. See Alexander Jones, “The Roofed Spherical Sundial and the Greek Geometry of Curves,” in J. Steele and M. Ossendrijver, eds., Studies on the Ancient Exact Sciences in Honour of Lis Brack-Bernsen (Berlin: Edition Topoi, 2017).
10. Robert Hannah and Giulio Magli, “The Role of the Sun in the Pantheon’s Design and Meaning,” Numen 58 (2011): 486–513; and Robert Hannah, “The Orchestration of Time in Ancient And Medieval Buildings,” in Giulio Magli, Antonio César González-García, Juan A. Belmonte, Elio Antonello, eds., Archaeoastronomy in the Roman World (Berlin: Springer, 2019), 37–56.
12. See Richard J. A. Talbert, Roman Portable Sundials: The Empire in Your Hands (Oxford: Oxford University Press, 2017).
14. Oxford, Museum for the History of Science 51358; Derek J. de Solla Price, “Portable Sundials in Antiquity, including an Account of a New Example from Aphrodisias,” Centaurus 14 (1969): 253–256; Karlheinz Schaldach, Römische Sonnenuhren: eine Einführung in die antike Gnomonik (Frankfurt am Main: H. Deutsch, 1998), 45–47; and Talbert, Roman Portable Sundials, 52–59.
16. London, British Museum EA 68475; and J. Evans and M. Marée, “A Miniature Ivory Sundial with Equinox Indicator from Ptolemaic Tanis, Egypt,” Journal for the History of Astronomy 39 (2011): 1–17. See also: Dialface ID 317.
17. Robert Hannah, “Greek Government and the Organization of Time,” in Companion to Ancient Greek Government, ed. Hans Beck (Oxford: Blackwell, 2013): 352, Figure 23.2; and Schaldach, “Measuring the Hours,” 63–64, 66 Fig. III–1.
18. For example, Salzburg Museum SM 3985: Schaldach, “Measuring the Hours,” 81–83, Fig. III–15.
19. Tony Freeth et al., 2006. “Decoding the Ancient Greek Astronomical Calculator known as the Antikythera Mechanism,” Nature 444 (30 November 2006): 587–591; and Tony Freeth, “The Antikythera Mechanism: A Shocking Discovery from Ancient Greece,” (Stanford, CA: Stanford Humanities Center, 2016); and Alexander Jones, A Portable Cosmos: Revealing the Antikythera Mechanism, Scientific Wonder of the Ancient World (Oxford: Oxford University Press, 2017).