Show Summary Details

Page of

PRINTED FROM OXFORD HANDBOOKS ONLINE ( © Oxford University Press, 2018. All Rights Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in Oxford Handbooks Online for personal use (for details see Privacy Policy and Legal Notice).

date: 24 April 2019

The Predictive Sciences: Measuring and Forecasting Weather Conditions

Abstract and Keywords

This article examines the science and folklore of Greek and Roman methods of weather prediction, dividing techniques into astrometeorological practices (those that looked at the motions of the stars and planets to predict the weather) and “Theophrastan” practices (those that looked at signs on earth or in the atmosphere, from the behavior of animals to the colors of clouds). It includes some discussion of how and when the ancients thought about the causal links between both kinds of signs and the ensuing weather, and discusses the uses of weather prediction in agriculture and seafaring, and its place in ancient didactic literature.

Keywords: science, astrometeorology, folklore, didactic literature, agriculture, seafaring, material culture, animals, Theophrastus

Like any other kind of prediction, weather forecasting is at its heart a practice of looking at signs in order to infer possible, probable, or inevitable events in the future.1 The links between the signs and their significations can be directly causal, as when a physician sees a woman with the plague and predicts that she will likely die. They can also be more distantly causal, as when a red sky tonight anticipates a pleasant day tomorrow: the red sky does not cause the nice weather tomorrow in the same way as a plague causes death, but instead the red sky and tomorrow’s weather are both presumably caused by the same general atmospheric state. Thirdly, signs can—at least in theory—be noncausal: we either don’t know or don’t care why this event always seems to follow on that one, but we recall that it does and, so, make our prediction accordingly. We find ourselves in this situation often enough in the modern world, when, for example, a statistical correlation has been found between two things before the causal mechanism that correlates them has been discovered.

When reading ancient sources, it is not always clear which of these three situations an author sees as applying in any given case. Other times, however, an author is explicit, as when Ptolemy explains the physics that connects planetary motions to the weather. This example should serve to remind us that the causal mechanisms and physical connections that people see between signs and the events that they purport to foretell are very much culturally and historically dependent: Ptolemy sees astrological causation where a modern astronomer would see mistaken attribution or observational error. For our present purposes, this means that ways of forecasting the weather—what signs to look for, what kinds of events could be predicted, how those signs may be connected to their predictions—will vary across time and place. Also, the intellectual and institutional contexts in which weather prediction finds itself will vary: Is weather forecasting seen as a form of divination? Of astronomy? Of rural or maritime folklore? Given the strong intellectual and cultural ties that we find between Greece and Rome in antiquity, we should expect considerable overlap in their methods of predicting weather, although, as we shall see, there is also change and development over time.

I. Observing and recording

Given the great deal of material on weather prediction that survives from classical antiquity, it is perhaps surprising that we have no direct textual evidence of attempts to measure or record day-to-day weather by Greeks and Romans.2 Contrast this to the situation in Mesopotamia, where the Babylonians made a concerted effort to record weather in a set of remarkable texts known as the Astronomical Diaries, which include observations (and calculations) of heavenly phenomena, observations of weather, as well as river levels and monthly commodity prices.3 The Diaries also include important historical and political events, and it is through an entry in one diary that we know the date of Alexander the Great’s death. The relevant entry shows the dispassionate nature of these texts, as well as their regular pairing of the important and the mundane: “the king died, clouds.” The diaries were made, continuously or very nearly so, over the course of 800 years at Babylon (from some time in the mid-eighth century BC until well into the first century AD). The composing of the Diaries represents—with no word of exaggeration—the single longest-running scientific research program in human history. They are a product of no small investment on the part of the researchers and their institutional sponsors. The purpose or purposes of the Diaries are still not entirely clear, but it has been speculated that at least one of their functions was omenological: to try and determine whether or to what extent the non-astronomical phenomena they tracked were periodic in the same way as astronomical motions were, or perhaps to see whether the historical events, weather events, and commodity prices were linked to the astronomical motions or to each other.4 Weather phenomena recorded in the diaries was not limited to weather that impeded astronomical observation, and indications are often given of how intensive the weather was on any particular day, with phrases like “light rain,” or “a little thunder,” or, charmingly, “rain, [so much that] the sandal was removed.”

Greek and Roman sources, by contrast, make no attempt to systematically record weather, although they give considerable attention to its prediction. Modern scholars generally divide classical weather-prediction methods into two camps, which were probably not mutually exclusive in practice. One, widely evident in the literary and epigraphic record, is the prediction of weather by means of astronomy-astrology, commonly called astrometeorology. The second, likely to have been widespread in the oral tradition, is the prediction of coming weather from atmospheric and animal signs (I will call this type of weather prediction “Theophrastan,” after one its most prominent ancient sources).

II. Astrometeorology

Astrometeorology in the classical tradition consists predominantly in the practice of correlating weather with what are called the phases of the fixed stars. Stellar phases occur because, in addition to the sun’s daily motion relative to a given location’s horizons (its daily risings and settings),5 it also has a slow west-to-east motion over the course of the year relative to the stars and constellations (the sun moves approximately one degree per day against the background of the fixed stars, through each equinox and solstice and back around again). This means that, as the sun creeps along through the signs of the zodiac, a given star will rise earlier and earlier each day relative to the sun.6 If we imagine a star at a time of year when it happens to be too close to the sun to be visible, we can see that, with the star rising earlier and earlier each day, there will come a date on which it is sufficiently far from the sun as to finally become visible, just above the horizon, a little before sunrise. This is one of its phases, called its heliacal rising, and it happens every year at the exact same point in the seasonal cycle.7 The three other significant phases of a star’s year are similarly reliable seasonal indictors: its first failure to appear in the sky just after sunset; its final rising on the eastern horizon just after the sun sets (the next day it will come into visibility in the darkening sky, but it will already be above the horizon); and its first setting on the western horizon just before sunrise (the previous day it will have disappeared into the increasing morning brightness before it touched the horizon).

The reliability of the timing of stellar phases relative to the seasonal year matters because, until the introduction (and subsequent correction) of the Julian calendar, almost all ancient calendars wandered with respect to the seasons.8 This meant that anyone interested in when to plant barley, when to harvest turnips, when it was safe to go out on the water, among other activities, needed some non-calendrical means of tracking the change of season. When does winter really begin? Using the Gregorian calendar (remarkably accurate for its apparent simplicity), we have only a three-day variation in the timing of the solstices and equinoxes from year to year. In lunar calendars, such as the various ancient Greek ones, there is considerably more wandering from one year to the next, depending on if and how a particular calendar was intercalated (that is, when and if leap-months were inserted). In actual practice, Greek calendars complicate the matter even further, insofar as, at least by the Hellenistic period if not earlier, there was occasional but significant administrative manipulation of dates, taking what was nominally a lunar calendar and making it subject to the whims of officials. There is, for example, one known instance in which, anticipating an annual festival in which a travelling dramatic troupe was supposed to perform but for which they had not yet arrived, the Athenians in 270 BC delayed the festival by interfering with the calendar. The festival was supposed to take place on the tenth of the month Elaphebolion, and when the troupe had not arrived by the night of the ninth, the archons simply declared the next day would also be the ninth of Elaphebolion, and they repeated the practice until, the performers finally arriving, they allowed the tenth of Elaphabolion at last to have its day. Such practices led to what appear to be multiple calendars overlapping in the same location, with inscriptional evidence that sometimes distinguishes between dates “according to the archons” and dates “according to the moon.”

The Greeks, therefore, turned to stellar phases as the most reliable way of marking the seasons. There are different schema extant, but the one from the Hippocratic On Regimen gives the flavor:

τὸν μὲν οὖν ἐνιαυτὸν ἐς τέσσαρα μέρεα διαιρέω, ἅπερ μάλιστα γιγνώσκουσιν οἱ πολλοί, χειμῶνα, ἦρ, θέρος, φθινόπωρον· χειμῶνα μὲν ἀπὸ πλειάδων δύσιος ἄχρι ἰσημερίης ἠαρινῆς, ἦρ δὲ ἀπὸ ἰσημερίης μέχρι πλειάδων ἐπτιολῆς, θέρος δὲ ἀπὸ πλειάδων μέχρι ἀρκτρούρου ἐπιτολῆς, φθινόπωρον δὲ ἀπὸ ἀρκτούρου μέχρι πλειάδων δύσιος.9

I divide the year in four parts, the ones that most people know best: winter, spring, summer, [and] autumn. Winter is from the setting of the Pleiades to the vernal equinox, spring from the equinox to the rising of the Pleiades, summer from the Pleiades to the rising of Arcturus, and autumn from Arcturus to the setting of the Pleiades.

Use of the stars, equinoxes, and solstices as seasonal and agricultural markers goes back in the literary record to the earliest periods of Greek poetry. Hesiod, writing in the eighth or seventh century BC, divides the agricultural year of the Works and Days using just these kinds of markers, although he also mixes in a number of animal and other signs as well.10 We can see the prominence of seasonal indicators in the poem when, after 300-odd lines of moral introduction, he finally comes to discuss the “works” and their “days.” He opens the agricultural section of the poem thus:

  • Πληιάδων Ἀτλαγενέων ἐπιτελλομενάων
  • ἄρχεσθ᾽ ἀμήτου, ἀρότοιο δὲ δυσομενάων.11

  • Begin the harvest at the rising of the Pleiades, and the ploughing when they set.

Hesiod goes on to tell us that wood felled in the “rainy autumn,” as Sirius appears higher and higher in the sky each night, will be most free from worms (417). He warns us that waiting too long past the setting of the Pleiades to plough, specifically doing so at the winter solstice (479), will yield a poor crop. Sixty days after the winter solstice, he tells us, will bring the evening rising of Arcturus and, with it, the duty to prune vines (564–567). Grain is to be winnowed at the rising of Orion in the summer (597–598), and grapes to be harvested at the morning rising of Arcturus, which is the same time, he tells us, as Orion’s coming to mid-heaven (an unusual astrometeorological indicator, as it turns out, 609–610). He reminds us again to plough at the setting of the Pleiades, which he associates with the setting of the Hyades and that of Orion as well (615–617). He then goes on to tell us of the relevance of stars for seafaring: bring ships in for winter at the setting of the Pleiades (618–621), and do not return to the sea until fifty days after the winter solstice (663).

A few weather-signs in the poem are not astrometeorological but are instead of the “Theophrastan” type, to be discussed more thoroughly in the last section of this article. One sign tells us that hearing the call of the crane in the autumn (on its migration southward, presumably) signals rainy weather (448), and another that heavy rain just after the appearance of the cuckoo happily negates any potential losses faced by those who had ploughed too late during the previous autumn (486). Hesiod adds that when the north wind Boreas collects clouds thickly, one can expect rain toward evening (552–553). A few interesting “hybrid” signs combine astrometeorological and other signs. One such indication is from the activity of snails, which they undertake in apparent response to the rising of the Pleiades: when the snails climb up houses, “fleeing” that star cluster, harvest time has come (571–573). Another associates the flowering of artichokes with the enfeebling influence of Sirius on men (women, by contrast, are made “most lustful,” μαχλόταται, 582–588). Hesiod also includes one calendar reference, to the winter month Lenaeon, which he says is miserable. He associates its frosts with the north wind Boreas.

Associations of stellar phases and seasons or weather show up regularly in the Greek literature after Hesiod, and there is a group of texts known as parapegmata, many of which are concerned exclusively with stellar astrometeorology.12 Parapegmata appear both in literary texts and in epigraphy, but their name, meaning “pegging-beside-devices,” derives from their epigraphic instantiation, where holes were drilled beside daily entries for stellar phases or weather, and a peg was moved, day by day, to indicate the current position in the cycle. There are a number of extant parapegmata of both inscriptional and literary types. Inscriptional parapegmata were often used to track cycles other than the astrometeorological year. We have inscriptional parapegmata to track the days of the week, the phases of the moon, the sun’s motion through the zodiac, and we even have several extant parapegma for tracking the Roman calendar. This all makes perfect sense, since, in an age before the advent of inexpensive paper, the use of holes and a moveable peg allowed for the construction of what was effectively a perpetual wall calendar for tracking days and dates. Some were magnificent and beautifully carved (the Latium parapegma, for example), some were crude graffiti.

From Miletus we have two different inscribed astrometeorological parapegma, one of which dates from the early first century BC, the other probably not far from that. Both are fragmentary. One of the two (Miletus I) lists only stellar phases with their peg-holes, but it does include the entry “Sagitta sets, it is the season of the continuous west wind.” The other Miletus parapegma (Miletus II), however, is rich with astrometeorological content, and, like many literary parapegmata, it attributes its stellar phases and weather forecasts to various older authorities (here each “●” marks a hole for a peg):

  • αἲξ ἀκρώνυχος δύνει κατὰ καὶ Φίλιππον καὶ Αἰγυπτίους.

  • αἲξ ἑσπερία δύνει κατὰ Ἰνδῶν Καλλανέα ●

  • ἀετὸς ἀπιτέλλει ἑσπέρας κατ᾽ Εὐκτήμονα

  • ἀρκτοῦρος δύεται ἕωθεν καὶ ἐπισημαίνει κατ᾽ Εὐκτήμονα. τῆιδ᾽ ἀετὸς ἐπιτέλλει ἑσπέρας καὶ κατὰ Φίλιππον.

  • Capella sets acronychally according to both Philippus and the Egyptians.

  • Capella sets in the evening according to Callaneus of the Indians.

  • Aquila rises in the evening according to Euctemon.

  • Arcturus sets in the morning and there is a change in the weather according to Euctemon. On this [day] Aquila rises in the evening also according to Philippus.13

Here we notice an effort being made to preserve differences in wording as presumably found in the original sources, calling the same phase variously “acronychal” and “evening.” Note also the peg hole at the end of the second line, marking a day with no associated stellar phase or weather. In other fragments of this same parapegma, we find references to hail, thunder, and winds.

The references to older authorities such as Euctemon and Philippus are common in astrometeorological parapegmata, and astronomers cited by other texts are a veritable who’s who of early Greek astronomy: Meton, Callippus, Conon, Dositheus, Eudoxus, and Hipparchus, among others. Roman-era parapegmata include other sources such as Varro and Caesar, probably either Julius (in connection with his calendar reform) or Germanicus (in connection with his translation of Aratus). The inclusion of earlier authorities, particularly the early Greek astronomers, has led many scholars to believe that those early sources were themselves authors of parapegmata, now lost, from which our extant entries are derived. This is certainly possible, but the earliest attested astrometeorological parapegma is as late as the third century BC, and it is difficult to say with certainty what form the earlier sources may have taken. In particular, there are wide discrepancies among all attributive parapegmata, sometimes diverging by many days in the timing of phases according to one and the same authority, or even leaving entries out entirely, which has led some scholars to be more cautious about what the older sources may have looked like, or even if they were systematic and exhaustive in the same way as parapegmata are meant to be.14 It is also possible that attributions to any single author were culled from multiple sources, which would explain some of the inconsistencies across citations.

After the third century BC, however, the picture becomes much clearer, and we begin to have complete parapegmata extant from literary sources. Astronomical and agricultural authors both preserve astrometeorological parapegmata for us. We find one appended, for example, to the end of Geminus’ astronomical textbook, the Introduction to the Phenomena, and Ptolemy himself dedicates a little book to one.15 There is a parapegma keyed to the zodiacal dial on the Antikythera mechanism, although in its present state it is too fragmentary to tell whether it included any meteorological information or whether it only listed stellar phases (the latter is probably more likely). We find instances of astrometeorological parapegmata in Pliny’s Natural History, Columella’s On Agriculture, and material related to them strewn throughout Ovid’s Fasti.

Because literary parapegmata were written on papyrus or parchment, the simple expedient of drilling a hole for a moveable peg was no longer feasible, and so the user needed some other method of tracking their place in the astrometeorological cycle. This problem was solved variously in our sources. Geminus, writing probably in the first century BC16 (and possibly before the advent of the Julian calendar), linked the stellar phases to the sun’s motion through the zodiac:

  • τὴν δὲ παρθένον διαπορεύεται ὁ ἥλιος ἐν ἡμέραις λʹ.
  • ἐν μὲν οὖν τῇ εʹ ἡμέρᾳ Εὐδόξῳ ἄνεμος μέγας πνεῖ, καὶ ἐπιβροντᾷ.
  •             Καλλίππῳ δὲ οἱ ὦμοι τῆς παρθένου ἐπιτέλλουσι· καὶ ἐτησίαι λήγουσιν.
  • ἐν δὲ τῇ ιʹ ἡμέρᾳ Εὐκτήμονι προτρυγητὴρ φαίνεται· ἐπιτέλλει δὲ καὶ
  •             ἀρκτοῦρος, καὶ οἰστος δύεται ὄρθρου· χειμὼν κατὰ θάλασσαν· νότος.
  •              Εὐδόξῳ ὑετός, βρονταί· ἄνεμος μέγας πνεῖ.
  • The sun travels through Virgo in thirty days.
  • On the fifth day: According to Eudoxus a strong wind blows and it thunders.
  •  According to Callippus the shoulders of Virgo rise and the Etesian winds stop.
  • On the tenth day: According to Euctemon Vindemiatrix appears, Arcturus rises, and
  • Sagitta sets at dawn; storm at sea, south wind. According to Eudoxus, rain, thundery, a strong wind blows.

Exactly how an ancient reader would understand the different predictions ascribed to the different authorities is not entirely clear, nor how reliable specific predictions or authorities were thought to be. Perhaps predictions ascribed to Euctemon were thought to be more applicable to Attica, and those ascribed to Eudoxus more applicable to Rhodes, where each was known to have worked? This seems reasonable enough, though we have no explicit evidence either way.

By the first century AD, we start seeing literary parapegmata keyed according to the Julian calendar, although Ptolemy works with the Alexandrian calendar, which was intercalated in the same way as the Julian and so stayed locked to the seasons for many years at a stretch.

How the associations between weather and stellar phases were made in the first place by the various authorities is unknown, although an observational basis has been widely supposed. There are two issues of interest in this regard. One is the question of whether the meteorological predictions were averaged or approximated based on observations made repeatedly over the course of several years (and if so, one wonders how many years), or whether some of the sources may have been reports of single years. A second, related issue has to do with predicted weather conditions that would, by definition, preclude astronomical observations. Whatever the observational status of a prediction like “heavy rain” might be, it is clear that if such rain was observed, then the stellar phase with which it is associated could not, in all likelihood, have been seen directly on that night. The ordering and timing of the phases must have been known from other (rain-free) years, or else the phase was inferred from a star’s position once visibility resumed. No matter how these impediments to observation were handled, it is clear that, if we do hold that astrometeorological weather predictions were derived “observationally,” we must qualify that term carefully.

We have in total about twenty extant astrometeorological parapegmata, dating from the early third century BC to the fifteenth century AD. Most modern scholars believe the tradition goes back further than the third century BC, insofar as the authorities mentioned in many parapegmata date to as early as the fifth century, but I have pleaded caution. We do know that the basic technological feature of the inscriptional parapegmata—which is to say the drilling of peg-holes in stone to track some kind of recurring cycle—appears to date from as early as the fifth century BC, but the one parapegma we have that may be that old, the Athenian Ceramicus parapegma, simply tracks a fragmentary numbered list: “fifth, sixth, seventh, eighth, ninth.” We unfortunately have no idea what it was counting.17 It certainly does not appear to have been meteorological.

Fixed-star astrometeorology also appears in many sources apart from parapegmata, showing the tradition to have been very widespread. Thus Aratus (third century BC) tells us that sailors watch for stellar phases in the morning and at nightfall:18

  • καὶ μὲν τις καὶ νηὶ πολυκλύστου χειμῶνος
  • ἐφράσατ᾽ ἢ δεινοῦ μεμνημένος Ἀρκτούροιο
  • ἠέ τεων ἄλλων, οἵ τ᾿ ὠκεανοῦ ἀρύονται
  • ἀστέρες ἀμφιλύκης, οἵ τε πρώης ἔτι νυκτός.19
  • And many a man onboard ship has noticed signs of a surging storm by paying heed to either dread Arcturus or some other stars that are drawn from the ocean at morning twilight and when it is still early night.

Similar passages are scattered throughout Pliny’s Natural History, Vergil’s Georgics, Varro’s Rerum Rusticarum, and many others.

II. i. Planetary Astrometeorology

Ptolemy, in addition to having written one of our longest and most detailed parapegmata (the Phaseis), also included three chapters on weather signs in his great astrological treatise, the Tetrabiblos. One chapter concerns fixed-star astrometeorology such as we have seen in parapegmata (although in this particular instance the treatment is more cursory), one chapter concerns Theophrastan-type signs (to be discussed in the next section), and one discusses how to use planetary positions to predict weather. This tradition of planetary astrometeorology has a long history after Ptolemy, and in fact it continues to be practiced even today by professional astrologers and almanac makers.

Ptolemy’s chapter on fixed-star astrometeorology is a short description of what to expect when the sun is in each sign:

τὸ μὲν οὖν τοῦ Κριοῦ δωδεκατημόριον καθόλου μέν ἐστι διὰ τὴν ἰσημερινὴν ἐπισημασίαν βροντῶδες καὶ χαλαζῶδες· κατὰ μέρος δὲ ἐν τῷ μᾶλλον καὶ ἧττον ἀπὸ τῆς τῶν κατ᾽ ἀυτὸν ἀπλανῶν ἀστέρων ἰδιότητος τὰ μὲν προηγούμενα αὐτοῦ ὀμβρώδη καὶ ἀνεμώδη, τὰ δὲ μέσα εὔκρατα, τὰ δ᾽ ἑπόμενα καυσώδη καὶ λοιμικά.20

The sign of Aries is on the whole thundery and full of hail, because of the change in weather at the equinox. Part by part (in increase and decrease from the character of the fixed stars in that part21): its leading parts are rainy and windy, its middle parts are temperate, its last parts are hot and pestilential.

He elaborates each remaining sign in a similar way. He repeats the formulaic καθόλου, “on the whole,” for each subsequent sign, qualifying the reliability of the semiotic and causal relationships.

Ptolemy then outlines a more detailed method in II.13 that uses the new and full moons that immediately precede the solstices and equinoxes, taking these in conjunction with planetary positions at those moments to determine general weather conditions. He also offers a more specific method that looks to either the new or full moon just before the solstice or equinox, or else (in the case where the new or full moon after the solstice or equinox is closer to that equinox or solstice than the preceding new or full moon had been) to all new or full moons throughout the season. He then casts what is essentially a horoscope for the coming season. As with a natal horoscope, the astrologer looks at the signs in which the sun and moon are found, their angles relative to the other planets, and the qualities of those planets and the signs in which they are found to determine the general weather conditions. Finally, Ptolemy offers an even more detailed method that builds on the previous, looking now to quarter moons in addition to new and full moons. All of this builds on his previous remarks about signs and fixed stars, where the annual appearances and disappearances of important fixed stars have now the power to increase or decrease the effects we should otherwise expect from our reading of the moon’s position, particularly as they move through what modern astrologers call the four “angles,” that is, the two horizons, the midheaven, and the imum coeli (the unobservable point in the sky directly opposite the midheaven). Ptolemy even says that the intensifications and decreases caused by the fixed stars at these points can give us “hour-by-hour” (καθ᾽ ὥραν) predictions, and he compares their effects to the way the moon influences the tide. Specifically, he adds, winds are intensified in particular directions depending on the latitude of the moon.

These methods and others like them have a venerable tradition in astrology, reaching their summit, perhaps, in the practices of Renaissance and Early-Modern almanac makers. However, even today farmer’s almanacs sometimes cryptically allude to their secret methods of reading the moon for long-term weather forecasting, so something like these methods may still be in use.22

III. Theophrastan Weather Prediction

Theophrastus of Eresus is by no means the only or even the earliest source for terrestrial and atmospheric weather signs, but his work on this type of prediction is our only extant and complete ancient standalone text on the subject,23 and so I hope the name will suffice if only to distinguish signs from animals, birds, clouds, and the air, from the astrometeorological signs examined above.

Theophrastan prediction uses a wide range of terrestrial and atmospheric phenomena as weather signs. We have already seen examples of birds and snails signaling changes in Hesiod. In the Tetrabiblos, Ptolemy rounds out his discussion of fixed-star and planetary astrometeorology with a list of skyward signs to watch for (note that these are not strictly speaking astrological, in the sense that they do not directly involve the stars as causes): if the sun rises unclouded, it signifies good weather; if it is “spotted” (ποικίλος) or has a special kind of cloud beside it (τὰ καλούμενα παρήλια νέφη, “the so-called parhelian clouds,” mentioned also by Aratus), then strong winds can be expected. Other similar observations of the moon’s brightness, mottledness, and its haloes, taken just before and after its significant phases, give weather predictions. Halos around stars and planets, as well as their clarity on any given night, are likewise worth observing for weather prediction. Finally, cloud formations, rainbows, shooting stars, and comets (the latter taken to be an atmospheric phenomenon in antiquity) can all predict winds and storms.

The classic ur-text for this kind of prediction is, as I have said, the De Signis of (probably) Theophrastus, dating from either the fourth or the third century BC.24 The text itself is a puzzling one in many ways. It is organized by the type of weather signified (signs for storm, signs for rain, signs for winds) rather than by the signs themselves. This means that if someone were to see, say, a strange kind of cloud, one would not know where in the text to look for what this particular cloud signified, unless one already knew what it signified (since the text is organized according to the weather type signified, not by the signs one observes). This seems rather to defeat the purpose of a reference work, assuming of course that its purpose was to serve as this kind of reference work, which is not at all certain.25

The author himself tells us little except that he has collected as many weather signs as he was able (καθ᾽ ὅσον ἦν εφικτόν), some observed by himself, others by unnamed people of no small reputation (παρ᾽ ἑτέρων οὐκ ἀδοκίμων). He immediately distances his work from astrometeorology, telling his reader that those things can be learned from the astronomers, although he does give us a few interesting astronomical signs. One tells us that the weather at the setting of the Pleiades will remain stable until the summer solstice, reminiscent of a rule in Pliny (HN xviii.231) that says the weather on the day before a solstice, the day of, and the day after it will give us the general weather conditions for each of the following three months, respectively.26 So, too, we are told that the significant lunar phases divide up monthly weather periods.

For the terrestrial and atmospheric signs, which make up the bulk of the book, we are treated to a smorgasbord of signs that we might watch for. There is some structure to each section, generally beginning with signs from the appearance of the sun and moon, followed then by miscellaneous happenings. Many of the former are very like our modern “red sky at night” rules: if the sun sets in a clear sky in winter or just after a storm, then it promises fair weather. If it rises very hot but not bright, this is a sign of wind. If the young moon is “upright” (σελήνιον ὀρθόν, presumably referring to the orientation of its horns) until the fourth day of the month, then it will storm until the middle of the month. (Pliny attributes a very similar sign to Varro, telling us that if the moon is directa on the fourth, there will be a great storm at sea [NH xviii.348], and Aratus explicitly points to the upright horns on the fourth day as a sign of wind.)27

One is often tempted with Theophrastus’ animal signs to hypothesize that he may have thought there to have been some causal correlation understood as operating between the animal’s behavior and the impending weather, as if the animal simply sensed the coming storm before we did. Burt all he really says is things like: if geese honk more than normal, it is a sign of storm. Similarly, cattle digging at the ground before bedding down implies a stormy winter is coming. But if the link to perception is never explicitly made in Theophrastus, we do know from other sources that, in at least some instances, the Greeks and Romans explicitly say that an animal sign happens because the animal “perceives” the coming weather. So, for example, Plutarch’s delightful short work Whether Terrestrial or Aquatic Animals Are Smarter mentions early on that animals often have superior senses to our own, and contains a number of examples of animals explicitly being said to perceive coming changes in weather.28 Pointing to a passage in Aratus (Arat. 956), for example, where it is said that ants quickly bring up their eggs to the surface before it rains, Plutarch adds that they do so if they perceive (αἴσθωνται) that their eggs are becoming dank (967F-968A). So also hedgehogs sense (προαίσθωνται) that the wind is about to change (972A), and tuna perceive the equinoxes and solstices:

ὁ δὲ θύννος οὕτως ἰσημερίας αἰσθάνεται καὶ τροπῆς, ὥστε καὶ τὸν ἄνθρωπον διδάσκειν μηδὲν ἀστρολογικῶν κανόνων δεόμενος· ὅπου γὰρ ἂν αὐτὸν χειμῶνος αἱ τροπαὶ καταλάβωσιν, ἀτρεμεῖ καὶ διατρίβει περὶ τὸ αὐτὸν τρόπον ἄχρι τῆς ἰσημερίας.29

The tuna senses the solstice and equinox so well that it even shows them to man, without the need for astronomical tables. For wherever the solstice catches it, the tuna remains and spends its time in that same area until the equinox.

What is really remarkable about this passage is that the tuna has here become the sign that tells people that the solstice and equinox are happening through its behavior (though Plutarch declines to tell us precisely how that behavior is observed or how small an area the individual tuna now occupies).

Lest we think all animal precognition was thought to be due to sensory superiority, though, we need to keep in mind another passage in the same work. There, the crocodile’s ability to foretell the exact width to which the Nile will flood is described in curiously noncausal language: it is said to be a “guess” (στοχασμός), and its foreknowledge to be “divinatory” (μαντική) rather than logical (λογική).30 Nor, Plutarch adds for good measure, is the animal’s knowledge the product of intelligence or deliberation (ἐπίνοια, συλλογισμός).31 Even if Plutarch is perhaps being figurative by calling the crocodile’s abilities divinatory in this passage—and we must keep in mind that he may not be—this still raises caution about our ascribing any and all animal foreknowledge to superior sensory apparatus in the ancient sources, even ones so effusive about animal abilities as Plutarch is.

Returning to Theophrastus, we can also see a number of signs that are not likely causal in any direct sense, given the extent to which they rely on incidental or seemingly random details. For example: the observation of a white sparrow signals a great storm. Clearly sparrows do not turn white just in time for stormy weather. So too, if one hears a seal make a loud bark in the harbor while it is holding an octopus, this signals a storm. Or, a personal favorite: if a crow standing on a white rock lifts its head while that rock is being washed by a wave, this promises rain.

Another major source for Theophrastan signs, and one much more familiar to Hellenistic and Roman readers, is the Phaenomena by Aratus of Soli. Aratus collects a great number of signs, but, like Theophrastus, does not speculate on how or why they work. Aratus begins his section on weather signs with observations of the moon: the orientations of its horns, its color, and its haloes, adding some rules that correlate the age of the moon (the number of days since new moon) with periods for the validity of its signs. Signs from the appearance of the sun, sky, and clouds follow. He completes his survey of skyward signs with the observation of the constellation of the Manger and two stars in Cancer called the Asses, which foretell weather depending on how they look in the sky or if one or the other of them becomes difficult to see. From there he moves on to a wide variety of signs from birds, insects, and animals, the sea, thunder and lightning, shooting stars, and more. In some instances we may discern hints at an implied causal connection, as when a heron calling repeatedly while flying awkwardly (οὐ κατὰ κόσμον) to shore is said to be moving on a storm-bearing wind (913–915).

In addition to his list of short-term signs, ones that we should watch for in order to know what weather is likely today or in the next few days, Aratus also offers us some rules of thumb from observations of trees and plants that allow us to forecast whole seasons and crop-yields.

Signs like these and others are prominent in many genres of ancient literature, though usually in more scattered form, as in Roman agricultural texts where Theophrastan rules of thumb for weather forecasting pop up individually or in small clusters rather than in longer, focused discussions as in Theophrastus and Aratus.

IV. Conclusion

It is clear that many of the ancient weather signs, of both the Theophrastan and astrometeorological types, date back to rural and maritime oral traditions, passed on by farmers, sailors, and fishermen from time immemorial. And indeed, we find these kinds of signs prevalent in agricultural and seagoing contexts in antiquity. From Hesiod’s first happy intermingling of both types of sign in his poem about rural life, to more specific concentration on one type or another as becomes common in later texts, utility is never very far from the surface. So, for example, Aratus mentions farming but focuses even more on the use of his Theophrastan signs by those who take to the sea. Three centuries later, there began a wave of Roman agricultural writing that included Varro, Vergil, Columella, and Pliny, a phenomenon that has no real equivalent in the Greek-speaking world. All of these authors set their astrometeorological material firmly in the context of farming, and often speak of the utility of such signs to increase yields and to prevent disasters such as spoiled crops or non-germinating seeds. Their astrometeorology was built, as we have seen, on what had become by Pliny’s day a several-centuries-old textual tradition of parapegmata.


Allen, J. (2001) Inference from Signs, Oxford.Find this resource:

Clay, J. S. (2009) “Works and Days: Tracing the Path to Arete,” in F. Montanari et al., eds., Brill’s Companion to Hesiod, Leiden, pp. 71–90.Find this resource:

Clay, J. S. (2003) Hesiod’s Cosmos, Cambridge.Find this resource:

Ercolani, A. (2010) Opere e giorni, Rome.Find this resource:

Evans, J. (1998) The History and Practice of Ancient Astronomy, Oxford.Find this resource:

Evans, J. and J. L. Berggren (2006) Geminos’s Introduction to the Phenomena, Princeton, NJ.Find this resource:

Fakas, C. (2001) Die hellenistische Hesiod: Arats Phainomena und die Tradition des didaktischen Epos, Wiesbaden.Find this resource:

Fortenbaugh, W. W. and D. Gutas, eds., (1992) Theophrastus: His Psychological, Doxographical, and Scientific Writings, New Brunswick, NJ.Find this resource:

Gee, E. (2013) Aratus and the Astronomical Tradition, Oxford.Find this resource:

Gee, E. (2000) Ovid, Aratus, and Augustus, Cambridge.Find this resource:

Hannah, R. (2009) Time in Antiquity, New York.Find this resource:

Hannah, R. (2005) Greek and Roman Calendars, London.Find this resource:

Jones, A. (1999) “Geminus and the Isia,” Harvard Studies in Classical Philology 99, pp. 255–267.Find this resource:

Kidd, D. (1997) Aratus: Phaenomena, Cambridge.Find this resource:

Lehoux, D. (2004) “Impersonal and intransitive ἐπισημαίνει,” Classical Philology 99, pp. 78–85.Find this resource:

Lehoux, D. (2007) Astronomy, Weather, and Calendars in the Ancient World, Cambridge.Find this resource:

Rehm, A. (1941) Parapegmastudien, Munich.Find this resource:

Rehm, A. (1949) “Parapegma,” Paulys Realencyclopädie de classischen Altertumswissenschaft, vol. xviii.4, pp. 1295–1366.Find this resource:

Rochberg, F. (2004) The Heavenly Writing, Cambridge.Find this resource:

Sachs, A. and H. Hunger, eds. (1988–2006) Astronomical Diaries and Related Texts from Babylonia, Vienna.Find this resource:

Samuel, A. E. (1972) Greek and Roman Chronology, Munich.Find this resource:

Scofield, B. (2010) A History and Test of Planetary Weather Forecasting, Doctoral diss., University of Massachusetts, Amherst.Find this resource:

Sider, D. (2002) “On On Signs,”[Should the second “On” be deleted? Or, should it be italicized, part of the title “On Signs?] in W. W. Fortenbaugh and G. Wöhrle, eds., On the Opuscula of Theophrastus, Stuttgart, pp. 99–112.Find this resource:

Sider, D. and C. W. Brunschön (2007) Theophrastus of Eresus: On Weather Signs, Leiden.Find this resource:

Swerdlow, N. (1998) The Babylonian Theory of the Planets, Princeton.Find this resource:

Taub, L. (2003) Ancient Meteorology, New York.Find this resource:

Volk, K. (2002) “Letters in the Sky: Reading the signs in Aratus’ Phaenomena,” American Journal of Philology, 133, pp. 209–240.Find this resource:

Volk, K. (2010) “Aratus,” in Clauss and Cuypers, eds., A Companion to Hellenistic Literature, Malden, MA, pp. 197–210.Find this resource:

West, M. L. (1978) Hesiod, Works and Days, Oxford.Find this resource:


(1) On theories of signs in antiquity, see Allen, 2001.

(2) On classical weather prediction, see Taub, 2003; Sider and Brunschön, 2007, Lehoux, 2007.

(3) The edition of the extant Diaries is by Sachs and Hunger, 1988–2006.

(5) Pretending for the moment that we live in the ancient geocentric cosmos.

(6) With the exception of those stars that are close enough to the celestial pole that they never rise or set for the observer’s latitude.

(7) The only changes in its timing are (in the short term) the apparent variation that is due to intercalations in the Gregorian calendar (the same reason the solstices and equinoxes happen on slightly different dates over the course of four years), and (in the long term) the slow slippage that is due to the fact that the sidereal year is about four one-hundred-thousandths of a day longer than the tropical year on which we base our calendar.

(8) On ancient calendars, see Hannah, 2005, 2009; Samuel, 1972.

(9) Reg. III.68.

(10) On the Works and Days, see Ercolani, 2010; Clay, 2009, 2003; Hannah, 2009; West, 1978.

(11) Op. 383-384.

(13) For “there is a change in the weather” as a translation of the Greek ἐπισημαίνει, see Lehoux, 2004.

(14) Lehoux, 2007, but contrast e.g., Hannah, 2009, 2005.

(15) For Geminus, see Evans and Berggren, 2006.

(16) See Jones, 1999.

(17) For what it is worth, the ordinal adjectives are feminine: πέμπτη etc. For the date and speculations on its function, see most recently Lehoux, 2007.

(18) On Aratus, see e.g., Gee, 2013, 2000; Volk, 2010, 2002; Fakas, 2001; Kidd, 1997.

(19) Aratus, Phaenomena, 744–47, translation Kidd’s.

(20) Ptolemy, Tetr. II.12, p. 95.18–23. These descriptors are followed closely in an expanded form by Ptolemy’s younger contemporary, Vettius Valens, I.2. It is unclear whether they collectively rely on a third source.

(21) In this phrase, I have taken the κατ᾽ αὐτόν, “in” or “with respect to it,” to refer back to τὸ μέρος, the part (of the sign of the zodiac). Robbins in the Loeb edition has changed the phrase to read ἀπὸ τῆς τῶν κατὰ τῶν ἀπλανῶν ἀστέρων ἰδιότητος, which I find very difficult to make sense of, and he in fact glosses it in the English, as “due to the special quality of the fixed stars.” In Cumont’s Teubner edition the text is the same as I have used and the only variant he notes in the apparatus is that one stemma of manuscripts omits the τῶν before κατ᾽ αὐτὸν. Unrelatedly, I follow Robbins in reading ἐν τὸ μᾶλλον καὶ ἧττον as referring to increase and decrease in intensity of influence in this passage. Compare for example, Tetr. I.8, p. 22.1.

(22) See, e.g., Scofield, 2010. Explaining the limitations on his knowledge of the methods used by modern almanac makers, Scofield tells us that, on inquiring with the editors of the Old Farmer’s Almanac, he received “an astonishingly hostile response” (n. 115, p. 51). My curiosity is piqued.

(23) For an admirably detailed list of known authors, attributions, and fragments, see Sider and Brunschön, 2007, pp. 5–29.

(24) On the question of the book’s authorship, which is not entirely ironclad, see Sider and Brunschön, 2007, pp. 40–43. On Theophrastus’ scientific work in general and the De signis in particular, see Sider and Brunschön, 2007; Sider, 2002; Fortenbaugh and Gutas, 1992.

(25) On the organization of this text and what it can tell us about the author’s intentions, see Sider and Brunschön, 2007, pp. 32–38.

(26) This latter rule I discovered, much to my surprise, still to be in circulation among a rural community in Southern Ontario at the turn of this century. See Lehoux, 2007, pp. 3–4.

(27) Aratus, Phaenomena, 788–791.

(28) I thank the anonymous referee for the suggestion.

(29) Plutarch, Moralia, 979C-D.

(30) Plutarch, Moralia, 982C.

(31) By contrast, though, he says a little later that the crocodile uses judgment rather than emotion in deciding which babies it will rear and which it will kill (982D).