1. THEORETICAL DEFINITIONS OF ARSIS/THESIS AND THE PERCEPTION OF RHYTHM
Since the beginning of modern investigations on ancient rhythmics, scholars have faced significant problems in interpreting the technical terminology employed by ancient rhythmicians, especially in relation to two of the most basic terms attested in the sources: ἄρσις and θέσις, which indicate the two fundamental components of a rhythmical foot.
Part of the difficulty derives from the fact that the extant Greek definitions of these concepts are very concise. Despite their brevity, however, they depict a very consistent scenario: the term ἄρσις denoted the ‘lifting’ of the foot, which was also described as the up-beat (τὸ ἄνω), while θέσις indicated the ‘placement’ or ‘step’ of the foot on the ground, or the down-beat (τὸ κάτω).Footnote 1 Their combination produced a rhythmical πούς (‘foot’), which Aristoxenus defined as the element that marks the rhythm of a musical piece and makes it intelligible to perception.Footnote 2 A slightly more complex case is that of the term βάσις (‘step’): except for Aristoxenus, who employed it to refer to the down-beat (τὸ κάτω),Footnote 3 the term βάσις generally denoted a dance step or rhythmical unit as a whole,Footnote 4 corresponding to the Aristoxenian πούς.
Even though the Greek sources clearly indicate that the distinction into arsis/thesis was essential in defining the rhythmical structure of a musical piece, as we will see in §1 only a few sparse passages give us some hints about how they were expressed in musical terms. This is all the more remarkable given the emphasis put by Aristoxenus on the importance of perception (αἴσθησις) at the beginning of the Elementa Rhythmica, where he defined nothing less than ‘the origin and fundamental principle of the science of rhythmics' as the study of ‘time divisions and their perception’.Footnote 5 Apparently, the perception of rhythmic divisions into arseis and theseis was so natural for the Greeks that they did not feel the need to discuss it extensively.
The reticence of the sources led many modern scholars to apply an oversimplified interpretation of Western rhythmic conventions to ancient Greek music, assuming that θέσις/βάσις (the ‘down-beat’) was not just occasionally accompanied by clapping or stomping, as the sources indicate,Footnote 6 but was consistently emphasized by means of dynamic accents that affected both instrumental and vocal music.Footnote 7 However, a few Classical testimonies suggest that this practice was far from universally followed or even appreciated: for instance, Theophrastus mentions clapping along during an aulos performance as an example of tasteless behaviour, and Cratinus disparagingly labels the ‘Pig-Boeotians’ as a ‘race of clapper-wearers’ (κρουπεζοφόρον γένος ἀνδρῶν, fr. 77 K.–A.).Footnote 8
Furthermore, such a rigid understanding of rhythm perception does not apply to Western music either. In fact, even though in our musical tradition down-beats are indeed regarded as ‘strong’ rhythmical positions, as opposed to ‘weak’ up-beats, it is not true that all ‘strong’ beats are marked by audible dynamic accents in musical performances. This kind of consistent dynamic accentuation is mostly employed as a temporary teaching device, which helps foster an internalization of the rhythmic pulse that will eventually substitute the exaggerated accentuation.Footnote 9 And we are certainly capable of identifying the rhythmical structure of music played on the organ or the harpsichord, even though these instruments are literally unable to produce dynamic accents—an extreme case which shows that Western rhythm too is based on rules that are far more complex than a mere alternation of loud and soft tones.
More generally, modern ethnomusicological studies have shown that many non-Western rhythmical idioms are organized on the basis of completely different principles and do not necessarily grant a special role to dynamic accentuation.Footnote 10 Significantly, these results are consistent with recent findings in the field of music psychology which have demonstrated how, contrary to what is often claimed about ancient rhythm,Footnote 11 the presence of dynamic accents is not absolutely necessary in order to perceive sequences of sounds as rhythmically organized.Footnote 12
Therefore, there is no ‘natural’ reason to assume that dynamic accents regularly marked rhythmical θέσεις and we should be wary of applying this idea to ancient Greek music in the absence of explicit testimonies.Footnote 13 If anything, the evidence provided by traditional Indian and Japanese music points in the opposite direction: in these musical traditions dynamic accentuation plays a very minor role in marking rhythmical phrases,Footnote 14 a feature that might be significant for our purposes given that both Sanskrit and Japanese are mora-timed pitch-accented languages like Ancient Greek.Footnote 15
A second hypothesis often advanced in writings on ancient rhythmics maintains that long syllables enjoyed a predominant status purely by virtue of their length, as opposed to pitch or stress.Footnote 16 Once again contemporary research in music psychology can help us assess the merits of this theory, since recent experiments have shown that rhythmical events emphasized by virtue of duration alone, which bear a so-called ‘temporal’ accent, are not intrinsically more capable of conveying rhythmical information than events marked by ‘melodic’ accents. By contrast ‘melodic’ accents, which emphasize specific sound-events exclusively by means of pitch changes, have proven to be significantly more capable to elicit rhythmical perceptions than temporal accents of equal salience.Footnote 17
These results clearly show that the preference traditionally granted to dynamic or temporal accents in interpreting the evidence on ancient rhythm is unwarranted. Yet, the alternative possibility that melodic features signalled the subdivision of rhythmical phrases into arsis and thesis has hardly ever been considered.Footnote 18 As we have noted above, rhythmical salience is often associated with dynamic accents in Western music, while Western languages are mostly based on stress patterns and Western folk songs tend to place stressed syllables on rhythmically ‘strong’ beats.Footnote 19 So is it possible that a similar correspondence worked also in ancient Greek music? In other words, just as linguistic salience depended on the position of acute accents (whether simple or embedded into circumflex accents),Footnote 20 is it possible that rhythmical salience was also conveyed by melodic accents that emphasized higher-pitched arseis?
This option seems all the more promising since we know that the pitch-accents of Greek lyrics played a vital role in determining a song's melodic shape: as many extant musical fragments show, acute accents tend to be associated with melodic peaks and circumflex accents with falling melodic contours.Footnote 21 And another important feature of ancient Greek musical practice strongly suggests that arseis had greater rhythmical importance than theseis: in fact, Greek musicians marked arseis and not theseis by means of a dot (stigmē) in their scores, a fact that would be at least surprising if theseis were the most relevant rhythmical elements.Footnote 22 Finally, this hypothesis is supported by the Greek and Latin sources on rhythm, to which we will now turn.
2. ARSIS IN THE GREEK RHYTHMICAL SOURCES: FOLLOWING HIDDEN CLUES
As we said at the beginning, the Greek sources provide only a few comments on the musical features associated with rhythmical arseis and theseis. Aristoxenus, in the extant part of his Elementa Rhythmica, does not even offer a formal definition of these terms: he simply employs them as synonyms of ‘up-beat’ (τὸ ἄνω) and ‘down-beat’ (τὸ κάτω),Footnote 23 suggesting that their perception and identification was all but natural. Fortunately, Aristides Quintilianus offers a richer account of these rhythmical notions in the following passage from Book 1 of his De Musica (Aristid. Quint. De Mus. 31.8–17 W.-I.):
ῥυθμὸς τοίνυν καλεῖται τριχῶς· λέγεται γὰρ ἐπί τε τῶν ἀκινήτων σωμάτων (ὥς φαμεν εὔρυθμον ἀνδριάντα) κἀπὶ πάντων τῶν κινουμένων (οὕτως γάρ φαμεν εὐρύθμως τινὰ βαδίζειν) καὶ ἰδίως ἐπὶ φωνῆς· περὶ οὗ νῦν πρόκειται λέγειν.
ῥυθμὸς τοίνυν ἐστὶ σύστημα ἐκ χρόνων κατά τινα τάξιν συγκειμένων· καὶ τὰ τούτων πάθη καλοῦμεν ἄρσιν καὶ θέσιν, ψόφον καὶ ἠρεμίαν. καθόλου γὰρ τῶν φθόγγων διὰ τὴν ὁμοιότητα τῆς κινήσεως ἀνέμφατον τὴν μέλους ποιουμένων πλοκὴν καὶ ἐς πλάνην ἀγόντων τὴν διάνοιαν τὰ τοῦ ῥυθμοῦ μέρη τὴν δύναμιν τῆς μελῳδίας ἐναργῆ καθίστησι, παρὰ μέρος μέν, τεταγμένως δὲ κινοῦντα τὴν διάνοιαν. ἄρσις μὲν οὖν ἐστι φορὰ μέρους σώματος ἐπὶ τὸ ἄνω, θέσις δὲ ἐπὶ τὸ κάτω ταὐτοῦ μέρους. ῥυθμικὴ δέ ἐστιν ἐπιστήμη τῆς τῶν προειρημένων χρήσεως.
The term ‘rhythm’ is used in three ways. For it is applied to bodies that do not move, as when we speak of a statue having ‘good rhythm’, as well as to anything that moves, as when we speak of someone walking with ‘good rhythm’; and it is properly applied to the voice: this is what we are now to discuss.
Rhythm is a combination of durations put together in some definite order: and we call their modifications arsis and thesis, sound and calm. Given that in general the notes, through the lack of differentiation in their movement, make the interweaving of the melody meaningless and lead the mind to wander in confusion, it is the parts of rhythm that make the function of the melody clear, moving the mind part by part but in an orderly fashion. In fact, arsis is the upwards motion of a part of the body, while thesis is the downwards motion of the same part. Rhythmics is the science of the employment of the things we have mentioned.
While the bulk of Aristides’ discussion of rhythmics clearly derives from Aristoxenus,Footnote 24 his account of the notions of arsis and thesis goes well beyond the meagre sketch outlined in the extant parts of the Elementa Rhythmica. In particular, even though Aristides confirms Aristoxenus’ association of ἄρσις with upward motion and θέσις with the opposite, he presents it as the logical conclusion of a complex train of thought, which begins by defining the ‘modifications’Footnote 25 of rhythmically ordered durations as arsis and thesis, ‘sound’ and ‘calm’.
On a straightforward reading, Aristides seems to be telling us that arsis corresponds to ψόφος and thesis to ἠρεμία—an association that evidently undermines the rhythmical prominence traditionally attributed to theseis. For this reason, Aristides’ testimony looked suspicious to many scholars who either emended the text by inverting the last two terms or argued in favour of a dubious chiastic reading of this passage, which would re-establish the supposedly ‘correct’ correspondence of thesis with the rhythmically marked element.Footnote 26 But this approach is problematic under many respects. First of all, it alters the text without any apparent need, just to conform it to a preconceived and unattested notion of Greek rhythm as a sequence of dynamically marked theseis and weaker arseis. Second, it introduces a significant interpretative prejudice, namely that the word ψόφος—a hapax legomenon in Aristides—necessarily entails the presence of dynamic accents.
However, scores of passages show that in acoustic and musical contexts the term ψόφος indicated generally a sound, without any specific implications in dynamic terms.Footnote 27 This is particularly clear in a passage of Aristotle's De Anima, a text that provides us with vital information about the basic aesthetic categories related to sound perception in the Classical age. After defining ψόφος in general as the object proper to the sense of hearing and as a kind of movement (κίνησις),Footnote 28 Aristotle contrasts ‘loud sounds’ (μεγάλοι ψόφοι) with ‘soft sounds’ (μικροὶ ψόφοι),Footnote 29 showing unequivocally that the term ψόφος by itself did not carry any dynamic connotations. Furthermore, Aristotle identifies the differences proper to ψόφοι with ‘high and low pitch’ (τὸ ὀξὺ καὶ τὸ βαρύ),Footnote 30 not loudness or softness,Footnote 31 and defines φωνή, that is to say the medium of the kind of rhythm discussed by Aristides, precisely as a type of ψόφος.Footnote 32
A similar problem concerns the term ἠρεμία, which most interpreters take as referring to rhythmical rests.Footnote 33 Although this explanation makes perfect sense from a musical point of view, it is not warranted by Aristides’ usage of the term ἠρεμία and its cognates, which consistently denote gentle or calm movements;Footnote 34 by contrast, Aristides defines rhythmical rests as ‘empty durations’ (κενοὶ χρόνοι, 38.27–39.2 W.-I.), which are employed to ‘fill up’ (πρὸς ἀναπλήρωσιν) otherwise incomplete rhythmical sequences. This usage is paralleled in the treatises known as Anonyma Bellermanniana, where rests are again described as κενοὶ χρόνοι (§1, 3, 83, 85 and 102), while the expression ἠρεμία φωνῆς is employed, in a very Aristoxenian fashion, in order to describe the ‘static’ quality proper to the voice when it ‘stands still on one note’, as opposed to the voice's movement (κίνησις) between high and low pitches.Footnote 35
This evidence opens up the possibility that Aristides did not identify arsis and thesis with ‘sound’ and ‘silence’ but rather used these terms to indicate two different kinds of sounds, a connotation that is confirmed by his subsequent comments on the distinct aesthetic influence that arseis and theseis have on the mind (Aristid. Quint. De Mus. 82.4–6 W.-I.):
τῶν δὲ ῥυθμῶν ἡσυχαίτεροι μὲν οἱ ἀπὸ θέσεων προκαταστέλλοντες τὴν διάνοιαν, οἱ δ᾽ ἀπὸ ἄρσεων τῇ φωνῇ τὴν κροῦσιν ἐπιφέροντες τεταραγμένοι.
Of the rhythms, those that initially calm down the mind starting from thesis are quieter; by contrast, those that begin from arsis and bring the ‘sound’ upon the voice are agitated.
In this passage, Aristides describes the nature of arsis and thesis in terms that are remarkably similar to the ones he employed in Book 1. First, he depicts the soothing nature of rhythms that begin from theseis as ‘quieter’ (ἡσυχαίτεροι), a characterization that is evidently akin to the ‘calm’ (ἠρεμία) that he associated with thesis in the earlier passage and certainly does not refer to the presence of rhythmical rests, given that the different aesthetic effects of rhythms employing ‘complete feet’ and rhythms comprising rests are explicitly examined in the following lines (82.6–9 W.-I.).
Second, even though Aristides does not qualify the character of thesis in aesthetic terms but simply describes its relaxing effect on the mind, he presents arseis as being marked by a special kind of aesthetic prominence: rhythms beginning from arsis are agitated because of the distinctive way in which the voice is affected by κροῦσις, just like arseis corresponded to a special kind of ‘sound’ (ψόφος) in Book 1.
Once again, the interpretation of this text crucially depends on the meaning attributed to the term κροῦσις. In general, this word means ‘striking’ or ‘collision’, a connotation that led many interpreters to take this passage as referring to some sort of percussion that emphasized arseis.Footnote 36 However, in musical contexts, the word κροῦσις indicates specifically the striking or plucking of the string of an instrumentFootnote 37 and, consequently, denotes primarily differences in pitch, not volume.Footnote 38 And this is exactly how Aristides employs this term throughout his De Musica, as all the occurrences of the words κροῦσις/κρούειν in this work refer precisely to melodic features.Footnote 39 Therefore, in the absence of any signs to the contrary, we should take the word κροῦσις as indicating a particular type of pitch also in this passage, a kind of pitch that marked arseis and was perceived as being particularly moving and stimulating.
While this testimony is interesting in itself, it becomes much more significant in the light of a passage from Aristotle's De Anima that offers a remarkable characterization of high pitch (τὸ ὀξύ) as what ‘moves perception to a great extent in a short time’ (κινεῖ τὴν αἴσθησιν ἐν ὀλίγῳ χρόνῳ ἐπὶ πολύ), as opposed to the feeble influence of low pitch.Footnote 40 This idea is elaborated further in one of the pseudo-Aristotelian Problems (ps.-Arist. Pr. 19.49, 922b28–34):
διὰ τί τῶν τὴν συμφωνίαν ποιούντων φθόγγων ἐν τῷ βαρυτέρῳ τὸ μαλακώτερον; ἢ ὅτι τὸ μέλος τῇ μὲν αὑτοῦ φύσει μαλακόν ἐστι καὶ ἠρεμαῖον, τῇ δὲ τοῦ ῥυθμοῦ μίξει τραχὺ καὶ κινητικόν; ἐπεὶ δὲ ὁ μὲν βαρὺς φθόγγος μαλακὸς καὶ ἠρεμαῖός ἐστιν, ὁ δὲ ὀξὺς κινητικός, καὶ τῶν ταὐτὸ μέλος ἐχόντων εἴη ἂν μαλακώτερος ὁ βαρύτερος ἐν ταὐτῷ μέλει μᾶλλον· ἦν γὰρ τὸ μέλος αὐτῷ μαλακόν.
Why is it that, among the notes that produce a concord, the softer character is in the lower register? Is it because melody is by nature soft and calm, while it becomes rough and moving thanks to the mixture of rhythm? In fact, since the low note is soft and calm while the high one is moving, then when two notes appear in the same melody the lower note should be softer; for we have seen that melody was soft in itself.
This testimony is crucial for our purposes. Not only it confirms Aristotle's account by depicting high pitch as ‘moving’ (κινητικός), it also explicitly relates its stimulating aesthetic effect to the naturally ‘moving’ function of rhythm, presenting this association as part and parcel of a well-established belief: while melody in itself is ‘calm’, it becomes ‘moving’ thanks to the mixture created by rhythm. Aristides Quintilianus expresses the very same notion in even clearer Aristotelian terms: melody is naturally ‘female’, ‘inactive and without form’ and plays the part of the matter since it is capable of embodying opposite qualities (high and low pitch), while rhythm is ‘male’ and acts as the ‘maker’, in that it ‘gives shape to melody and moves it in an orderly fashion’.Footnote 41 In other words, rhythm affects melody by ‘mixing’ its pitches and durations in a definite order, creating a meaningful alternation of ‘moving’ and ‘relaxing’ elements. It is very significant to notice that the author of the Problem accounts for this phenomenon on the basis of the opposite natural characters and aesthetic effect of high and low notes while he completely ignores dynamic accents: according to him, low pitch is naturally ‘soft and calm’, similarly to the nature of melody in general, while high pitch is inherently moving (κινητικός), just like rhythm.
This conception matches perfectly both Aristides’ general definition of rhythm as ‘the movement of the voice’Footnote 42 and his specific characterization of arsis and thesis: in fact, the quality that Aristides attributes to unmarked theseis (ἠρεμία) is exactly the same that is associated with the relaxed nature of low pitch in the pseudo-Aristotelian Problem (ἠρεμαῖος). And, significantly, Aristides explains this contrast between ‘moving’ arseis and ‘relaxing’ theseis by giving rhythm precisely the same function it has in the pseudo-Aristotelian Problem: while abstract sequences of notes are inexpressive and obscure, rhythm clarifies the function (dynamis) of a given melody by establishing a well-defined order of sounds and guides the mind by means of meaningful sequences of high and low pitches.Footnote 43 Once again, not a word about dynamic accentuation.
One last piece of evidence shows beyond doubt that arseis were perceived as the ‘marked’ elements of rhythmical sequences already in the third century b.c. In an extended passage of his Synopsis on Pulses, Galen recounts how the famous Alexandrian physician Herophilus developed a theory of pulse rhythm on the basis of concepts normally employed by musicians to analyse musical rhythms. More specifically, Herophilus identified rhythmical arsis with the ‘dilation of the artery’, a time-duration ‘marked by a perceptible movement, when the dilated artery beats against our touch’ (τόν τε τῆς αἰσθητῆς κινήσεως, ἡνίκα πλήττει τὴν ἁφὴν ἡμῶν ἡ ἀρτηρία διαστελλομένη), while he associated thesis with the contraction of the artery,Footnote 44 a time-duration characterized by absence of movement and ‘calm’ (ἠρεμία).Footnote 45
The terminological and conceptual similarities between Herophilus’ and Aristides’ testimonies are evident. Both associate arsis with aesthetically marked durations, which are identified through a physical beat (πλῆξις) in the body and by specific melodic means (ψόφος/κροῦσις) in the voice; similarly, they both identify thesis with a kind of ‘calm’ (ἠρεμία) that corresponds to physical stillness in the body and to the relaxing effect exerted by lower pitches on the mind. The closeness of Herophilus’ and Aristides’ approaches is further revealed by the fact that Aristides mentions precisely the ‘pulsation of the arteries’ along with dance and melody as examples of the three basic types of rhythm (Aristid. Quint. De Mus. 31.18–32.7 W.-I.):
πᾶς μὲν οὖν ῥυθμὸς τρισὶ τούτοις αἰσθητηρίοις νοεῖται· ὄψει, ὡς ἐν ὀρχήσει· ἀκοῇ, ὡς ἐν μέλει· ἁφῇ, ὡς οἱ τῶν ἀρτηριῶν σφυγμοί· ὁ δὲ κατὰ μουσικὴν ὑπὸ δυεῖν, ὄψεώς τε καὶ ἀκοῆς. ῥυθμίζεται δὲ ἐν μουσικῇ κίνησις σώματος, μελῳδία, λέξις. […] διαιρεῖται δὲ ὁ ῥυθμὸς ἐν μὲν λέξει ταῖς συλλαβαῖς, ἐν δὲ μέλει τοῖς λόγοις τῶν ἄρσεων πρὸς τὰς θέσεις, ἐν δὲ κινήσει τοῖς τε σχήμασι καὶ τοῖς τούτων πέρασιν, ἃ δὴ καὶ σημεῖα καλεῖται.
Rhythm as a whole is perceived by these three senses: sight, as in dancing; hearing, as in melody; and touch, as in the pulsations of the arteries. But musical rhythm is perceived by two senses, sight and hearing. In music, bodily movement, melody and diction can be organized rhythmically. […] In diction, rhythm is divided into syllables, in melody into the ratios of arseis to theseis, and in movement into figures and their boundaries, which are also called signals.
Remarkably, in this passage Aristides does not relate arsis and thesis to dance, as one would expect if their fundamental nature truly derived from the dynamic accentuation produced by striking feet on the ground.Footnote 46 By contrast he associates rhythmical subdivisions into arseis and theseis with melody, a choice that is perfectly consistent with the melodic connotation of these two concepts that is offered throughout the De Musica as well as in the Aristotelian sources.
Therefore, it is against this background that we should interpret Aristides’ well-known characterization of arsis as the upward motion (φορά) of a part of the body, as opposed to the downward motion associated with thesis. Since this depiction is presented as a logical follow-up of his previous remarks on the vital influence that rhythm exerts on melody, it is hard to see how Aristides could refer exclusively to dance movements without saying so openly. On the contrary, we should take Aristides’ definition of arsis as referring to an upward motion that affects all the aesthetic media he mentions: the arteries, the dancers’ limbs, as well as the melodic movement of the voice. This upward motion corresponded to the ‘marked’, ‘moving’ element of a rhythmical sequence, as opposed to the static and ‘calm’ character of theseis.
3. ARSIS AS ELATIO VOCIS IN LATIN RHYTHMICAL SOURCES
Even though the evidence provided by the Greek sources combines to form a remarkably consistent picture, in order to reconstruct it we had to follow clues scattered in many different texts, often in the form of passing remarks. The perception of musical arseis and theseis seems to have been so natural for the Greeks that they did not feel the need to discuss it extensively, a surprising feature that may be understood in the light of the agreement between Greek rhythm and prosody: just as prosodic accents highlighted the most salient syllables of a sentence by raising the voice's pitch, melodic accents emphasized arseis, the most salient elements of each rhythmical phrase.
The situation is markedly different in the case of Latin musical sources, which are much more explicit in associating rhythmical arseis with higher pitches and theseis with lower ones. The earliest Latin description of this correspondence appears in Book 9 of Martianus Capella's De Nuptiis—a text that is particularly important for our purposes, as it offers a Latin translation of extensive sections of Aristides Quintilianus’ De Musica, accompanied by a few explanatory integrations.Footnote 47 This holds true also for §974, where Martianus translates Aristides’ definition of rhythmical foot and openly relates its two basic components to different kinds of vocal movement:
ποὺς μὲν οὖν ἐστι μέρος τοῦ παντὸς ῥυθμοῦ δι᾽ οὗ τὸν ὅλον καταλαμβάνομεν· τούτου δὲ μέρη δύο, ἄρσις καὶ θέσις.
(Aristid. Quint. De Mus. 33.12–13 W.-I.)A foot is the part of the entire rhythm through which we understand the whole. It has two parts, arsis and thesis.
pes uero est numeri prima progressio per legitimos et necessarios sonos iuncta, cuius partes duae sunt, arsis et thesis. arsis est elatio, thesis depositio uocis ac remissio.
(Mart. Cap. De Nupt. §974)A foot is the primary progression of rhythm, which consists of sounds that are legitimately and necessarily related to each other. It has two parts, arsis and thesis. Arsis is an elevation, thesis a lowering and relaxation of the voice.
A few pages before, Martianus had already produced a similar interpretation of Aristides’ general definition of arsis and thesis which, as we have seen above, belonged to a discussion that was programmatically devoted to vocal rhythm and dealt primarily with melodic features.Footnote 48 Once again, Martianus’ paraphrase reflects accurately both aspects by establishing a direct connection between the melodic movement of the voice and rhythmical flows based on arsis and thesis (Mart. Cap. De Nupt. §967):
numerus est diuersorum modorum ordinata conexio, tempori pro ratione modulationis inseruiens, per id quod aut efferenda uox fuerit aut premenda, et qui nos a licentia modulationis ad artem disciplinamque constringat.
Rhythm is the orderly arrangement of different measures; it organizes time according to the flow of melodic modulation, by means of which the voice is either raised or lowered, and restrains us from melodic licence, binding us to art and discipline.
If the tenor of this passage was not clear enough, many other sections of Martianus’ treatise indicate that his comments on the raising and lowering of the voice refer specifically to changes in pitch, not volume. Particularly interesting for our purposes is Martianus’ general definition of modulatio (Mart. Cap. De Nupt. §932):
constat autem omnis modulatio ex grauitate soni uel acumine. grauitas dicitur quae soni quadam remissione mollescit, acumen uero, quod in aciem tenuatam gracilis et erectae modulationis extenditur.
Moreover, melodic movement (modulatio) as a whole consists of lower- or higher-pitched sounds. That which is called low pitch soothes by a certain relaxation of the sound, while high pitch is what is stretched up to the sharp tip of a high, thin melody.
This excerpt shows not only that Martianus conceived modulatio specifically as a combination of different pitchesFootnote 49 but also that the concept of remissio soni/uocis, which he employed in §974 to gloss Aristides’ definition of thesis, unequivocally indicates low pitch.Footnote 50 In addition, Martianus provides here exactly the same characterization of the different aesthetic effects of high and low pitches that we found in Aristides and the Aristotelian Problems: low pitches are said to be relaxing and soothing, while high pitches are tense and stimulating.
In spite of this noteworthy agreement, these testimonies have generally been dismissed as irrelevant or misleading because of their alleged inconsistency with the Greek evidence. But, as we have seen at the beginning, this assessment was grounded on the misguided assumption that the opposition between arseis and theseis depended exclusively on dynamic contrasts, a notion which is never attested in ancient texts. None the less, instead of questioning this basic premise, scholars have developed complex interpretative approaches to explain the alleged contradiction between the Greek and the Latin sources on the basis of a mysterious shift in meaning of the terms arsis and thesis: according to them, the Latin term arsis came to denote the accented and dynamically stressed part of the foot, while thesis indicated the unaccented part.Footnote 51 However, to my knowledge, this bold change is nowhere mentioned in the sources—a fact that would be at the very least surprising if such a radical overturn of a well-established convention actually took place.
This hypothesis appears even more problematic if we take a closer look at a passage by Marius Victorinus/Aphthonius that is generally cited as evidence of this alleged inversion. According to traditional interpreters, Victorinus did not realize that he employed in the same sentence two opposite and contradictory notions: a fictional ‘Greek’ concept of thesis as dynamically accented down-beat and a new ‘Latin’ meaning of arsis as the stress-accented part of a rhythmical foot.Footnote 52 But this interpretation seems rather implausible not only because Victorinus was a well-respected scholar who had excellent knowledge of ancient GreekFootnote 53 but especially because neither of these fictitious concepts is corroborated by the extant sources: the Greek evidence clearly indicates that arseis were associated with raises in pitch, not volume, and the same goes for the Latin sources, as we will see in detail below. For the moment, however, let us read the text in question (Mar. Vict. Ars Gramm. 6.40.14-17 Keil):
arsis igitur ac thesis quas Graeci dicunt, id est sublatio et positio, significant pedis motum. est enim arsis sublatio pedis sine sono, thesis positio pedis cum sono: item arsis elatio temporis, soni, uocis, thesis depositio et quaedam contractio syllabarum.
The notions that the Greeks call arsis and thesis, which means lifting and placement, indicate the movement of the foot. In fact, arsis is the lifting of the foot without sound, thesis the placement of the foot with sound: likewise, arsis is the elevation of a time-duration, sound or voice, thesis the placing-down and some sort of contraction of syllables.
The first sentences are pretty straightforward: they outline the familiar Greek rhythmical definition of arsis and thesis as the raising and lowering movement of a foot, adding the intuitive remark that they are respectively accompanied by silence and sound. This last observation probably refers to the practice of interspersing musical performances with clapping, stomping or finger-snapping, which is often discussed in Latin sourcesFootnote 54 and is occasionally mentioned also by Greek authors.Footnote 55 However, in the absence of clear evidence, we should be wary of assuming that such an accompaniment necessarily and constantly replicated the dynamic accentuation of the melody or its instrumental counterpart: for instance, both the ancient Indian tradition detailed in the Natyasastra and its modern counterpart show that this need not necessarily be the case.Footnote 56
The second part of this passage is more challenging. According to many interpreters, Victorinus/Aphthonius shifts here to the ‘new’ Latin usage, so that the word arsis now indicates the rhythmic section marked by a raise in volume that is opposed to a softer thesis. However, in my view it is difficult to reconcile this interpretation with the wording of the Latin text, especially with regard to the conjunctive adverb item (‘likewise’), which emphasizes the continuity of thought between these clauses and would be a very odd choice if the author really employed two different acceptations of the terms arsis/thesis. In actual fact, though, the second clause elaborates upon the content of the previous one, exactly as the word item implied. First, Victorinus transfers the distinction between up- and down-movements from the physical domain to a purely musical one; second, he clarifies how the words arsis and thesis could be employed in connection with different musical media:Footnote 57 pure sounds, syllables sung by the voice as well as abstract time-spans may appear in arsis or thesis, which correspond respectively to their ‘elevation’ and their ‘lowering’.
One last issue makes the traditional interpretation of this passage even more problematic. As we have seen above, many interpreters take the definition of arsis as elatio temporis, soni, uocis as referring to an increase in volume of vocal sounds, which they associate with the presence of dynamic stress in Latin accentuation.Footnote 58 However, there is no evidence that the expression elatio uocis had any dynamic connotations: in itself, the word elatio means simply ‘raise’ and could well indicate a change in pitch and not in volume. And this is exactly the meaning attested by late Latin grammarians, who consistently use the word elatio to refer to the variation in pitch that characterized accented syllables.Footnote 59 On a few occasions, they literally define these pitch variations as arsis and thesis, as for example in the following passage by pseudo-Priscian (ps.-Priscian 3.521.25–31 Keil):Footnote 60
ad hanc autem rem arsis et thesis sunt necessariae. nam in unaquaque parte orationis arsis et thesis sunt, non in ordine syllabarum sed in pronuntiatione: uelut in hac parte, natura, ut quando dico ‘natu’, eleuatur uox, et est arsis intus; quando uero sequitur ‘ra’, uox deponitur, et est thesis deforis. quantum autem suspenditur uox per arsin, tantum deprimitur per thesin. sed ipsa uox, quae per dictiones formatur, donec accentus perficiatur, in arsin deputatur; quae autem post accentum sequitur, in thesin.
Arsis and thesis are necessary in relation to this matter. For arsis and thesis are present in each and every part of speech, with regard to the pronunciation and not the order of syllables: for instance, in this part of speech, natura, when I say ‘natu’ the voice is raised and arsis is here. And indeed when ‘ra’ follows, the voice is lowered and thesis appears. As much as the voice is elevated in arsis, that much it is lowered in thesis. But the voice itself, which is created by uttering words, is attributed to arsis up until the point where the accent is completed; instead, the voice that follows after the accent is assigned to thesis.
Interestingly a passage from Quintilian's Institutio oratoria shows that this ‘vertical’ conception of pitch, which we take for granted but was not universal in the ancient world, was attested already in the first century a.d. Commenting on the different vocal qualities proper to orators and singers, Quintilian describes the orator's voice as powerful and enduring (fortis ac durabilis) but limited in pitch range, since orators did not train their voice to ‘reach the lowest and highest sounds’ (nec praeparare ab imis sonis uocem ad summos, 11.3.22.5–6); by contrast, singers are said to ‘sweeten even the highest notes with the melody of their voice’ (illi omnes etiam altissimos sonos leniant cantu oris, 11.3.23.3–4). The same approach is shown also in the work of Ptolemy's contemporary Aulus Gellius. In defining the science of ‘canonics’, a term that normally describes the type of harmonic investigation based on the use of the monochord (κανών),Footnote 61 Gellius tells us that the difference between rhythm and melody consists in the fact that the former is concerned with a sound's ‘length’ and the latter with its ‘height’ (Gell. NA 16.18.4–6):
κανονική autem longitudines et altitudines uocis emetitur. longior mensura uocis ῥυθμός dicitur, altior μέλος. est et alia species, quae appellatur μετρική, per quam syllabarum longarum et breuium et mediocrium iunctura et modus congruens cum principiis geometriae aurium mensura examinatur.
Canonics, on the other hand, measures the durations and pitch heights of the voice. The ‘longer’ measure of the voice is called rhythm, the ‘higher’ melody. There is also another science called metrics that examines with the aid of the ears the combination of long and short and intermediate syllables as well as the verse, in keeping with the principles of geometry.
Two points are particularly important for our current purposes. First, these testimonies show that the concept of altitudo was associated with higher pitches as early as the first century a.d. and that this usage was firmly established by the second century.Footnote 62 Second, Gellius outlines a stark contrast between rhythmics and harmonics on the one hand, metrics on the other: while the former are conceived as different parts of the same discipline, which we would call ‘music’, the discipline of metrics is presented as a separate science that is concerned with different sequences of long and short syllables and not with strictly ‘musical’ sounds. This theoretical opposition between rhythmics and metrics is reflected also by significant differences in the terminology employed in these two disciplines, differences which affect also the Latin usage of the term arsis: in fact, while musical treatises consistently relate arseis to raises in pitch, metrical texts show a very different understanding of this term, which we will examine in the next section.
4. ARSIS AS SVBLATIO IN LATIN METRICAL TERMINOLOGY
Differently from rhythmicians, metricians employed the term arsis to indicate the syllables placed at the beginning of a foot or metrical sequence; in such contexts, the word thesis designated the syllables appearing at the end of the same foot or metrical sequence. The fact that the metrical usage of the words arsis and thesis coexisted side by side with their pitch-related one is revealed in the pseudo-Priscian passage just discussed (3.521.25–31 Keil), where the author specifies that the words arsis and thesis refer to ‘the pronunciation’ and not ‘the order of the syllables’ (non in ordine syllabarum sed in pronuntiatione): this remark shows plainly that, in principle, the term could indicate two different concepts. Diomedes spells this out even more clearly (Diomedes 1.474.30–475.1 Keil):
pes est sublatio ac positio duarum aut trium ampliusue syllabarum spatio conprehensa. pes est poeticae dictionis duarum ampliusue syllabarum cum certa temporum obseruatione modus recipiens arsin et thesin, id est qui incipit a sublatione, finitur positione.
A [metrical] foot consists in the raise (sublatio) and placement (positio) that takes place in the space of two, three or more syllables. In poetic diction, a foot is a measure comprising two or more syllables; observing a specific relation between time-units, it receives arsis and thesis, which means that it starts from the raise (sublatio) and ends with the placing down (positio).
This passage is crucial not only because it tells us that the words arsis and thesis in metrical contexts referred specifically to the order of syllables, but also because it provides the key to understanding better the usage of other Latin authors: in fact, Diomedes translates this metrical acceptation of arsis as sublatio while, a few pages earlier, he used the term elatio in order to define the raise in pitch associated with the natural accentuation of words.Footnote 63
This use of the Latin word sublatio as a translation of the metrical acceptation of the term ἄρσις is confirmed both in theoretical reflections and in practical analyses of metrical sequences: for instance, Latin grammarians describe dactyls (– ⁝ ⏑⏑) as comprising a long syllable in sublatio and two short syllables in positio. Footnote 64 Modern scholars took these testimonies as evidence of a supposed contradiction between Latin and Greek sources: in fact, if sublatio is taken to mean arsis in a rhythmical sense, then the Latin metrical description indeed seems to revert the Greek rhythmical definition according to which the sequence – ⁝ ⏑⏑ comprises a long syllable in thesis and two short ones in arsis (Aristid. Quint. De Mus. 35.8–9 W.-I.). However, in my view this interpretation is mistaken because it mixes terminology and concepts that belong to two different disciplines: metrics and rhythmics.Footnote 65
In order to clarify these terminological differences, it may be useful to compare and contrast the different definitions of ‘dactyl’ developed by the two disciplines. On the one hand, as far as we can tell, Greek rhythmicians did not use the term δάκτυλος at all to describe the sequence – ⁝ ⏑⏑, which is called ἀνάπαιστος ἀπὸ μείζονος by Aristides Quintilianus. They employed the terms δάκτυλος/δακτυλικός exclusively to designate a rhythmical genus defined by an equal duration of arsis and thesis and not by a fixed sequence of long and short syllables.Footnote 66 By contrast, Greek metricians employed the term δάκτυλος to indicate specifically the syllable sequence – ⏑⏑, while they seem to ignore its rhythmical subdivision into arsis and thesis. Footnote 67
Given that the Latin grammarians clearly state that the term sublatio refers to the ordo syllabarum and, therefore, place their contribution squarely within the field of metrics and not rhythmics,Footnote 68 it is not really surprising to find out that their metrical definition of dactylic sequences differed from the Greek rhythmical one. That this discrepancy does not depend on the language (Latin vs Greek) but derives from the different terminology employed by metricians and rhythmicians is further shown by the fact that the metrical concept of arsis/sublatio as the beginning of a syllabic sequence appears also in late Greek metrical treatises: for instance, a passage from the so-called Anonymus Ambrosianus defines arsis as ‘the beginning of a verse, while thesis is the end’, and Choeroboschus relates these terms to the beginning and the end of a metrical foot.Footnote 69
One final piece of evidence shows that we are on the right track in distinguishing the concept of sublatio as the first part of metrical sequence from that of elatio as rhythmical arsis marked by a raise in pitch. In his Latin translation of Aristides Quintilianus, Martianus Capella never employs the term sublatio but, in keeping with Aristides’ usage,Footnote 70 consistently translates the rhythmical concept of arsis with the word elatio, as is shown by the following examples:
ἐν δὲ τῷ ἰαμβικῷ γένει ἁπλοῖ μὲν πίπτουσιν οἵδε ῥυθμοί· ἴαμβος ἐξ ἡμισείας ἄρσεως διπλασίου θέσεως, τροχαῖος ἐκ διπλασίου θέσεως καὶ βραχείας ἄρσεως, ὄρθιος ὁ ἐκ τετρασήμου ἄρσεως καὶ ὀκτασήμου θέσεως, τροχαῖος σημαντὸς ὁ ἐξ ὀκτασήμου θέσεως καὶ τετρασήμου ἄρσεως.
(Aristid. Quint. De Mus. 36.1–5 W.-I.)The following simple rhythms fall within the iambic genus: the iamb, made of an arsis half the length of the thesis and a thesis double the arsis; the trochee, made of a double thesis and a short arsis; the orthian, consisting of a four-time arsis and an eight-time thesis; the ‘marked’ trochee, consisting of an eight-time thesis and a four-time arsis.
qui igitur incompositi errant, isti sunt: iambus ex dimidia elatione et positione quae gemina est; trochaeus ex duplici positione et elatione quae breuis est. orthius uero, qui ex tetrasemi elatione [id est arsi] et octasemi positione constabit, ita ut duodecim tempora hic pes recepisse uideatur, atque habet propinquitatem aliquam cum iambico pede; quattuor enim primis temporibus ad iambum consonat, reliquis octo temporibus adiunctis. dehinc trochaeus, qui semanticus dicitur, id est qui e contrario octo primis positionibus constet, reliquis in elationibus quattuor breuibus artetur.
(Mart. Cap. De Nupt. §985)So these are the incomposite [iambic] measures: the iamb, consisting of an elatio half of the length of the positio and a positio which is double; the trochee, consisting of a double positio and a short elatio; and then the orthian, which will contain a four-time elatio, which is the arsis, and an eight-time positio, so that this foot appears to comprise twelve times in total and has some kinship with the iambic foot; for it agrees with the iamb by having four times first, while the remaining eight times are added afterwards. Next is the so-called ‘marked’ trochee, which is the reverse of the orthian and comprises the first eight times as positiones, while it is rounded off by the remaining four shorts in elationes.
This clear terminological distinction between metrical sublatio and rhythmical elatio explains also why Marius Victorinus shifts between these two terms in his discussion of the concept of arsis. Given that his whole section on arsis and thesis (6.40.5–41.21 Keil) focusses mainly on metrical questions, he mostly employs the metrical term sublatio.Footnote 71 However, when he introduces a comparison with the strictly musical domain and explains how arsis is expressed in song, music and dance, he shifts to the proper rhythmical term, elatio. Footnote 72
5. ANCIENT VERSE SCANSION AND THE ORIGIN OF THE METRICAL CONCEPT OF ARSIS/SVBLATIO
The previous sections have shown how the Latin sources present us with two different and well-defined concepts of arsis, which are respectively translated as elatio and sublatio. The word elatio corresponds to the rhythmical concept of arsis, which entailed a raise in pitch as well as a physical upward movement, while the term sublatio appears in metrical contexts and indicates the first syllables of a given metrical sequence. The obvious question now is: why was this double usage of the word arsis developed in the first place?
Of course, given the scantiness of the evidence, we are not in a position to provide a definite answer to this question but some aspects of the ancient practice of verse scansion allow us to formulate a reasonable hypothesis. First, it is important to note that the verb scandere acquired the specialized meaning of ‘scanning verses’ only in late antique sources,Footnote 73 while in Classical Latin it simply indicated the activity of ‘climbing’, in keeping with the basic meaning of the Indo-European root skand-. This difference in usage suggests, on the one hand, that verse scansion was perceived as being related to some kind of ‘elevation’, on the other that it was introduced at a relatively late stage. Second, our sources indicate that this practice belonged specifically to the educational sphere.Footnote 74 For instance, Priscian reports the following fictional dialogue between a teacher and a student, who is asked to scan the first line of the Aeneid and analyse its metrical structure (Priscian 3.461.15–17 Keil):Footnote 75
scande uersum. ‘arma ui rumque ca no Troi ae qui primus ab oris’
quot caesuras habet? duas. quas? semiquinariam et semiseptenariam.
But how did they mark these metrical subdivisions while reading verses aloud in class? To my knowledge, the earliest evidence on this matter is provided by Terentianus Maurus. In accordance with his interest in metre, Terentianus employs the terms arsis/thesis in their metrical acceptation, indicating respectively the beginning and the end of a metrical sequence, and significantly characterizes arsis as the part of a foot that subleuetur,Footnote 76 an expression which foreshadows the term that was to become standard in later Latin metrical sources, sublatio. Within this framework, he mentions two different devices employed to distinguish arsis and thesis: on the one hand, he describes as ponere the action that identified metrical theseis and tells us that teachers of metre marked these positions by finger-snapping or stomping;Footnote 77 on the other, he associates the concept of arsis with a raise in the voice's sound (attollit sonorem) and, interestingly, relates this meaning to the Greek usage—a choice that is perfectly consistent with the Greek evidence we have examined above.Footnote 78 The following passage by Sacerdos describes in greater detail what this kind of vocal accentuation entailed exactly (Sacerd. 6.448.20–5 Keil):
hoc tamen scire debemus, quod uersus percutientes, id est scandentes, interdum accentus alios pronuntiamus, quam per singula uerba ponentes. toro et pater, acutum accentum in ‘to’ ponimus et in ‘pa’; scandendo uero inde toro pater Aeneas in ‘ro’ et in ‘ter’. haec igitur in metro ideo suam non continent rationem, quia in ipsis nulla intellectus ratio continetur: nam ropater nihil significat.
But we also need to know the following: when striking verses, i.e. scanning them, sometimes we pronounce accents differently than when we put them on individual words. For instance, in the case of the words toro and pater, we put an acute accent on ‘to’ and ‘pa’. By contrast, when we scan the line inde toro pater Aeneas, we put the accent on ‘ro’ and ‘ter’. Hence these metrical subdivisions do not reflect the sense of the words, as no discernible meaning is contained in them: for rópater does not mean anything.
Here Sacerdos describes very clearly how, in scanning a dactylic hexameter, the beginning of each metrical foot was marked by an artificial acute accent, which was shifted from its natural prosodic position to the first long syllable of each foot. However, one aspect of this passage poses a serious riddle: why does he state that ‘ter’ is accented, a short syllable which closes a dactyl? Several options and emendations have been proposed in order to solve this problem. According to some interpreters, Sacerdos mis-scanned the word pătĕr as if it comprised two long syllables,Footnote 79 thus re-establishing a correspondence between accent and the beginning of the foot (ī́ndĕtŏ rṓpā tḗraē ne͜ā́s). However, even though similar prosodic errors caused by the influence of stress accents on vowel quantities appear elsewhere in Sacerdos,Footnote 80 the final remark of this passage makes this interpretation very unlikely: in fact, the group rópater is presented as one ‘word’, which means that it should bear only one accent,Footnote 81 while an erroneous scansion of pătĕr would produce two (rṓpātḗr). A promising alternative has been recently advanced by Karin Zeleny, according to whom the reading -ter is a corruption of the original Aen- caused by the similar appearance of these two syllables in cursive handwriting.Footnote 82 If this attractive explanation is correct, Sacerdos’ original text read as follows:
toro et pater, acutum accentum in ‘to’ ponimus et in ‘pa’; scandendo uero inde toro pater Aeneas in ‘ro’ et in ‘Aen-’.
This emendation not only leads to a consistent accentuation of the beginning of each metrical foot—ī́ndĕtŏ rṓpătĕr a ḗnē ā́ssīc ṓrtŭsăb ā́ltō, Verg. Aen. 2.2—but also explains why Sacerdos singled out specifically the words toro and pater. In fact, they exemplify the two kinds of artificial accentuation that affected words in verse scansion: the accent could either be shifted from its natural position to the beginning of the foot, as in the case of toró as opposed to the natural tóro; or it disappeared entirely, as in the case of the word pater. By contrast, Sacerdos did not have any reasons to comment on the first word índe, since its natural accent corresponded to the artificial one employed in metrical scansion.
But why was it necessary to devise such a complex method in order to teach metre in Late Antiquity? The answer readily suggests itself: the new practice of verse scansion was introduced in didactic contexts to compensate for the weakened perception of syllable quantities caused by the rise of stress accents,Footnote 83 which undermined a ‘natural’ identification of metrical feet on the basis of pure quantities. The grammarian Sergius comments explicitly on this point: after admitting that ‘it is difficult to know which syllables are long by nature’, he states that this ambiguity can be resolved ‘only by means of examples drawn from authority, when you have proceeded to scan (scandere) the verses of the poet’.Footnote 84 Not being able to perceive syllabic quantities correctly anymore, students learned to distinguish metrical feet by marking the beginning of each unit with an artificial vocal accent—something that is ironically very similar to what is still taught in most courses in Greek and Latin metre nowadays. By contrast, as Terentianus Maurus and Augustine testify, the second part of each metrical sequence was identified by means of a clap or stomp.Footnote 85
From this perspective, we can understand better a famous, if obscure, remark by Terentianus, namely that ‘a long syllable by itself is not sufficient to produce a foot, because a foot is created by two “beats” and not by a double time’.Footnote 86 If the present reconstruction is correct, Terentianus is referring to the two different types of ictus that distinguished arseis and theseis: a vocal accentuation that identified arsis/sublatio as opposed to a physical beat that corresponded to metrical thesis/positio. This hypothesis would also explain why late metrical treatises described both ‘descending’ and ‘ascending’ rhythms, that is to say rhythms that started respectively from thesis or arsis, as beginning with arsis/sublatio and ending with thesis/positio:Footnote 87 according to the new practice of verse scansion, the beginning of any given metrical sequence was marked with the same ‘elevation’ of the voice, independently of its original rhythmic character.
CONCLUSIONS
At the beginning of this study, we have seen how traditional approaches to ancient rhythmics have tacitly interpreted the evidence on the basis of a very simplified and misleading understanding of the rhythmic conventions of Western music, an approach that led many scholars to assume that theseis enjoyed a predominant status and were consistently marked by dynamic accents. However, as I showed in sections 1–4, these notions are nowhere attested in the ancient sources, nor are they supported by contemporary studies in ethnomusicology and music perception. If we let go of these preconceived ideas and interpret the evidence following the indications explicitly provided in the sources, it becomes clear that both Greek and Latin authors consistently depict a very different picture: theseis were regarded as relaxing and calm, while arseis were perceived as exciting and ‘moving’, and the rhythmical prominence of arseis was associated with the stimulating effect of high pitches, as opposed to the soothing influence of low notes.
Second, we have seen that the rhythmical notion of arsis as pitch-raise differs radically from the metrical concept of arsis, which indicates the beginning of a syllabic sequence, and these two concepts are designated by different Latin words: elatio and sublatio. This terminological and conceptual distinction allowed us to dismiss another long-lived tenet of scholarship on ancient rhythm, namely that the meaning of the rhythmical terms arsis/thesis underwent a radical overturn in the Latin sources, coming to denote exactly the opposite of what they indicated in the Greek world. As shown in sections 2–4, this alleged ‘inversion’ is not substantiated by the sources but derives from a modern misinterpretation of the meaning of the terms arsis/thesis in two independent disciplines: rhythmics and metrics.
This approach has several advantages. First, it produces a coherent explanation of the different usage of the terms arsis/thesis without resorting to unwarranted textual emendations or mysterious and unattested ‘inversions’ in the meaning of these words. Second, it accounts for the sudden appearance of the metrical concept of arsis/sublatio in late antique sources as a consequence of the new teaching practice of verse scansion. Finally, it clarifies why late antique metricians described both ‘ascending’ and ‘descending’ rhythms as beginning from arsis/sublatio and ending with thesis/positio: in scanning verses, the first part of any metrical sequence was marked with the same ‘elevation’ of the voice independently of its original rhythmical character, since the aim of this practice was not reflecting what happened in musical performances but helping students to identify the metrical structure of a given line.
By contrast both Greek and Latin musical sources consistently indicate that rhythmical arseis and theseis were expressed by meaningful alternations of high and low pitches in musical performances.Footnote 88 By establishing a ‘recognizable and orderly’ sequence of pitches and durations, rhythm transformed a shapeless scale into a moving melody which, in turn, ‘moves us in an orderly fashion’.Footnote 89
