The Music of Hindostan c13.s3: III. Musical Time

The Music of Hindostan c13.s3

Language: English

III. Musical Time

c13.s3.1

The simplest unity after the solitary beat (āditālaḥ) is, as we saw, dvitīyaḥ (◡ ◡ —, ◡ — ◡, — ◡ ◡). This gives us the three ordinary distributions of the longs and shorts (crotchets and quavers) within the four-unit foot (beat), a certain number of which, generally four, make up the bar (āvard) or section of common time. The name Ādi is used for common time (transferred from the solitary beat). Further development within the beat takes place by resolving any long or short into any one of these feet. Thus:

c13.s3.3

and similarly with

c13.s3.5

and

c13.s3.7

So that such a combination of notes as the look of which frightens us and the sound of which does not particularly appeal to us, appears to an Indian as quite an ordinary distribution of his accepted metrical units

c13.s3.9

|| 1 ◡ ◡ | 2 ◡ ◡ | 3 ◡ ◡ | 4 ◡ ◡ ||

c13.s3.10

in an *āvart*¹ of *Ādi tāla*. In fact, the main difficulty of realizing and enjoying the nice distinctions of Indian rhythm is that we have not acquired the habit of resolving mentally every unit into any distribution of its constituent elements so that we could sing them at a moment's notice or that they would instantly appeal to us when we heard them sung.

c13.s3.11

¹ This word is translated by ‘bar’ or ‘section’ as the context seems to require : its literal meaning is ‘enclosure’. The smaller unit, or bar or half-bar, is a *vibhāg* (lit. division). Sanskrit, *āvarta* and *vibhāga*.

c13.s3.12

Here are some instances of *Ādi*:

c13.s3.13

Tāla Ādi. Rāg Jhinjoti. (1)

c13.s3.14

Ex. 308.

c13.s3.15

&c.

c13.s3.16

(2) Rāg Dhīraśankarābharaṇa.

c13.s3.18

Ma - ri - yā - - - - da gā - - - - - du-rā

c13.s3.20

Ma - ri - yā - - - - da gā - - - - - du-rā

c13.s3.22

Ma - ri - ya - - - da gā - - - - - &c.

c13.s3.23

Rāg Dhīraśankarābharaṇa.

c13.s3.24

Musical notation for (3) Rāg Dhīraśankarābharaṇa. Do-ru-ku-ṇa yi-ṭu vaṇ-ti śe-va Do-ru-ku-ṇa &c.

c13.s3.25

Musical notation for (4) 1st. 2nd. O . . . . . . . Ran-ga śā-yi . . śā-yi . .

c13.s3.26

Tālam Ādi. This Rāgam has E and A omitted in descent.

c13.s3.27

(5) d = 72. Vivace.

c13.s3.28

Musical notation for (5) d = 72. Vivace. Nin nē . . . . . . . bha-ja-na sē - - - yū . . vā - ḍa-nu Nin - - nē . . . . . . bha-ja-na . . sē - - - - yū . . vā - ḍa-nu Nin - - - nē . . . . . . . bha-ja - - na . . . . . . . . .

c13.s3.29

But the section (āvard) need not synchronize with the bar (vibhāg); it may begin in the middle of the vibhāg, though, as with us, it seldom does. A more usual place for it to begin is on the last unit of the four. This kind is called Deśādi (foreign Ādi):

c13.s3.30

Deśādi Tālam. (1) Ex. 309.

c13.s3.31

Musical notation for Ex. 309. De - vā-di De - va sa-dā - śi - va . . De - vā - di De . . . va sa-dā - śi - va . . dī-na-nā - tha su - dhā &c.

c13.s3.32

Musical notation for (2) Che-ra - rā va - de - mi-rā . . Rā-ma - yya che -

c13.s3.33

2nd half.

c13.s3.35

- - ra - rā va - de - mi - rā Me - ra - gā - du - rā i - ka ma [Figure: ] &c. hā - - Me - ra - gā du - ra

c13.s3.36

Or it may begin on the second unit of the four, and is then called *Madhyādi* (‘middle’ *Ādi*). As the *Āvard* of *Deśādi* beginning on the fourth unit ended on the third, so here beginning on the second it will end on the first:

c13.s3.37

Madhyādi Tālam. (1)

c13.s3.38

Ex. 310.

c13.s3.39

Vi-du-la-ku mrok-ke-ḍa . . vi-du-la-ku mrok - ke-ḍa .

c13.s3.41

. . vi-du-la-ku mrok - ke-ḍa . . .

c13.s3.43

Ka - li - na - ru - la - ku ma-lu . . ma-lu de - li - [Figure: ] &c. pe - mi . . Ka-li - na-ru

c13.s3.45

En - na - ḍu zū - tu - no . . in a - ku - la [Figure: ] &c. ti - la - ka . . . . En - na - ḍu

c13.s3.46

These are South Indian names; the things are common enough in Hindostani song and probably have specific names, though I was unable to make sure of them.

c13.s3.47

A different interest attaches to the *tṛtīyaḥ*, ♪♪♪┐. This is also common time. But this peculiar form has left its mark, not in any special metrical form, but in the way a bar of common time is counted, in whatever metre. A short explanation is necessary first.

c13.s3.48

When three notes of neighbouring pitch and no more are sung without any particular emphasis on any one of them, the voice tends of itself to emphasize and the musical consciousness to give pre-eminence to the middle one. Thus in the scale of the Andaman islands, which (in some parts at least) consists only of three notes separated by some very small interval, it is recorded that the middle note is made the tonic. Again, in the snake-charmer’s pipe (see Ex. 274) the middle note of three is made the drone. Similarly, if three equal units of time are repeated and an accent is put on one of them, it will gradually determine to the middle one of the three: ◡ ú ◡, ◡ ú ◡ (or, if the commas between them represent rests: 2/4 ♩ ♩ ♩ ⁑ ♩ ♩ ♩ &c.).

c13.s3.49

It is difficult to prove this, because it is so hard to make the experiment fairly. But, for some reason best known to the builders, the wheels of railway carriages commonly beat out a tripleted rhythm, and the reader may have noticed that if he is not particularly attending to them (and so making the experiment a fair one) these take the form ◡ ú ◡, and not ú ◡ ◡ nor ◡ ◡ ú. However that may be, here is Indian verse on the one hand employing three equal syllables in succession followed by a rest ♩ ♩ ♩ ⁑, and Indian music on the other counting a bar of common time with three equal beats (tāl) and a blank, or empty (khālī) beat; and the curious fact that this blank beat comes invariably on the third of the bar which gives the exact effect of prosodical feet: | ◡ ú ◡ | ◡ ú ◡ |¹ Common time is, accordingly, called in the North invariably ‘three-beat’ (tintāl), the fourth beat being ‘empty’. The first of the bar, which bears the musical accent, is called sam ² (‘complete’, or ‘total’), because at that point the two independent rhythms of singer and drummer from time to time coincide. Time is ordinarily beaten with the hand upon the thigh (the singer is invariably seated), emphasizing the *sam* slightly, and beating in the air with upturned palm for the *khālī*. This method of beating is also applied to other times besides common time in a way which will appear presently.

c13.s3.50

¹ Cp. the first four bars of Tristan, of which the main accent is on the second bar and the fourth is blank.

c13.s3.51

² Pronounced like, and accidentally having the same meaning as, our word ‘sum’. It is important to be clear about this conception. The sam has no such stress as we place on the first of the bar. It gets its pre-eminence over the other tāls (beats) owing to the cross-rhythm (either with some other instrument or in the melody itself) being adjusted at that point. Or if there is no cross-rhythm, still the sam is pointed out by its distance from some typical or recurrent phrase elsewhere in the āvard (or vibhāg); as the quaver triplet of Tschaikowsky’s Scherzo tells us how far we have got each time in the series of crotchets. It is true that when singer and drummer coincide at that point their united triumph at having got it right after all does result in a little more tone, i. e. in stress, but this in no way necessary to the rhythm of the music.

c13.s3.52

These other times, various forms of three, five, and seven rhythm for the most part, are often regarded as intricate and unaccountable; but in the light of prosody they have a simple explanation. They are best summarized in the Carnatic system, which enumerates thirty-five different kinds as typical; but the principle which underlies them is of indefinite application, and as a fact I saw a book at Bhavnagar which enumerated some hundreds of them. It will be unnecessary to give the names of these thirty-five rhythms, because their names are only a piece of *memoria technica*,¹ or the picturesque ruins of titles whose meaning is long forgotten, and would only confuse us. It will be enough to arrange them in seven ranks and five files, lettering the ranks and numbering the files.

c13.s3.53

¹ See Mudaliar's Oriental Music, p. 25.

c13.s3.54

Counting by mātrā (Carnatic).

c13.s3.55

| | I. | II. | III. | IV. | V. |
| A. | 3 | Eka. 4 | 5 | 7 | 9 |
| B. | 3 2 | Rūpaka. 4 2 | 5 2 | 7 2 | 9 2 |
| C. | 3 1 2 | 4 1 2 | 5 1 2 | Jhampa. 7 1 2 | 9 1 2 |
| D. | Tripuṭa. 3 2 2 | (Ādi). 4 2 2 | 5 2 2 | 7 2 2 | 9 2 2 |
| E. | 3 2 3 | Mātya. 4 2 4 | 5 2 5 | 7 2 7 | 9 2 9 |
| F. | 3 2 3 3 | Dhruva. 4 2 4 4 | 5 2 5 5 | 7 2 7 7 | 9 2 9 9 |
| G. | 3 3 2 2 | 4 4 2 2 | Āṭa. 5 5 2 2 | 7 7 2 2 | 9 9 2 2 |

c13.s3.56

N.B.—Rank A is usually given in the order 4, 3, 7, 5, 9; the point being that 4 and 3 make 7, and 4 and 5 make 9 (9 is never treated as 3 × 3). 6, 8, and 10 are obviously omitted as being multiples of three numbers which are already there, and 9 is, as we saw, not considered to be a multiple. Those in heavy type are in common use. The names refer to the whole rank (Eka, for instance, to the whole of rank A, Tripuṭa to D, &c.), but are confined in practice to those files to which they are here assigned.

c13.s3.57

In rank A are the different values (in *mātrās*) given to the bar (*vibhāg*) when it is counted as a complete whole, that is, when it is not composed of various beats (*tāl*).¹ This merry little tune is an instance of the five-unit bar. Four bars make a section, as usual:

c13.s3.58

¹ In this chapter *Tāl* is used for ‘time’ and *tāl* for ‘beat’.

c13.s3.59

Khaṇḍajāti Laghu. Rāgam Mohāna (= Rāg Bhūpkaliān). ♩= 96.

c13.s3.60

Ex. 311.

c13.s3.61

(Musical notation)

c13.s3.62

(Musical notation)

c13.s3.63

[Figure: (Musical notation)] &c.

c13.s3.64

A common way of counting the uneven numbers is by continuous syncopation:

c13.s3.65

3 ♩ ♩ | ♩ ♩ | &c. 5 ♩ ♩ ♩ | ♩ ♩ ♩ | &c. 7 ♩ ♩ ♩ ♩ | &c.

c13.s3.66

The name of this is *Chapu Tāla*. File V is rarely used.

c13.s3.67

Rank B has two beats to the bar, C, D, and E three beats, and F and G four. It is not difficult to see in these the two, three, and four-syllable feet of poetry. B gives various forms of the trochee (— ◡) and iamb (◡ —), according to the beat with which the rhythm begins. The five files allow for different proportions between the long and the short beat. We may arrange the ten possibilities of iambs and trochees in a graduated proportion to a scale of 100, and it will be seen that a large number of ways of dividing a bar into two beats has been reached by very simple means:

c13.s3.69

We see from this diagram that just as the scales took account of *śrutis* of different sizes, so here the rhythms take account of different sized units (*mātrā*) and embody them in a system; and it is characteristic of Hindu practice to tabulate ten species of a thing and to make practical use only of one of them. The other ranks may be taken in the same way, and these thirty-five are seen to include by implication one hundred varieties of *Tāl*:

c13.s3.70

A B C D E F G 5 + 10 + 15 + 15 + 15 + 20 + 20 = 100

c13.s3.71

To continue: C and D give five varieties of dactyl, amphibrach, and anapaest (— ◡ ◡, ◡ — ◡, ◡ ◡ —). D. II is the normal dactyl, &c.; the rest of D lessens or increases the proportion of the long to the two shorts. C is our old friend the ‘cyclical’ dactyl¹ which modern theory has criticized out of existence, for Greece at any rate, but which made a considerable stir in its day. Here at any rate it has five separate values of its own. E gives the cretic (— ◡ —), bacchic (◡ — —), and antibacchic (— — ◡).

c13.s3.72

¹ The cyclical dactyl is one in which the middle syllable is shorter than the last.

c13.s3.73

Of the four-syllable feet (four beats in the bar—F and G) there are sixteen possibilities (for two things taken four together). Only eight appear here. Thus: F. I gives 3 2 3 3, and implies 2 3 3 3, 3 3 3 2, and 3 3 2 3. G. I, similarly, four more.

c13.s3.74

The other possibilities are: . 2 & 2 & 2 & 3 / 2 & 2 & 3 & 2 / 2 & 3 & 2 & 2 / 3 & 2 & 2 & 2 rare in the *śloka*. . 2 & 3 & 2 & 3 / 3 & 2 & 3 & 2 C. IV is used if wanted. . 3 & 3 & 3 & 3 / 2 & 2 & 2 & 2 already accounted for by *āditālaḥ*.

c13.s3.75

So that ranks F and G fairly represent the chief possibilities of the *śloka*.

c13.s3.76

But it will be more interesting to see how the most characteristic of these work out in practice. We will take: B. II. 2 + 4 *Rūpaka* (iamb). D. I. 3 + 2 + 2 *Tripuṭa* (dactyl). C. IV. 7 + 1 + 2 *Jhampa* (cyclical dactyl). and as a curiosity G. III. 5 + 5 + 2 + 2 *Āṭa* (ionicus a majore).

c13.s3.77

Rūpaka Tālam. (2 + 4.)

c13.s3.78

Ex. 312. Vi-na-ve O ma-na-sā vi-na-ve . . O ma-na-sā vi-va-ram-bu-ga ne . . del - pe da vi-va-ram-bu-ga &c.

c13.s3.79

A - la-ka-lā - lla lā - ḍa ga-ga-ni A - la - ka - lā - lla lā - ḍa-ga-ga-ni A-la-ka-lā - lla lā - &c.

c13.s3.80

Su-gu - na-mu-le - - jep-pu koṇ - ṭi sun - da-ra ra-ghu Rā - ma . - - Su-gu - &c.

c13.s3.81

Rūpaka is in idea Saraband rhythm. It is quite one of the commonest Tāls, and presents no difficulty, as, in practice, it is a simple 3/4.

c13.s3.82

Tripuṭa is harder. The secret of all these Tāls is that the units are taken as sums, not multiples. And though seven is not a multiple of anything, yet the European having counted three, finds it very difficult not to expect another three. The only way is to take a fresh lease of counting with each beat. It will be found quite easy to realize all these Tāls if one does as a native does—count out each beat 1, 2, 3, &c., and tap at every ‘1’.

c13.s3.83

Tripuṭa Tālam. (3 + 2 + 2.) (Telugu words.)

c13.s3.84

Adagio. ♩ = 100. Ex. 313. (Musical notation with lyrics: En - du - ku - da - ya - rā - - du - rā . . śrī rā - ma - chan - dra nī - ken - du . . ku - da - ya rā - - du - rā - - śrī . . rā - ma - chan - dra nī - ken - du - ku - - da - ya - rā - - du - rā . .)

c13.s3.85

Jhampa Tālam. The ‘bar’ (Āvard) is from one double bar line to the next. The single bar lines show the beats. Though it is counted 7 + 1 + 2, it is thought as 2 + 3 + 2 + 3 quavers.

c13.s3.86

Andante. ♩ = 144. Ex. 314. (Musical notation with lyrics: Am - ma rā - vam - ma Am - ma rā - vam - ma Am - ma rā - vam - ma Am - ma rā - vam - ma tu - la - sam - ma na - nu pā - lim - pa - ram - ma sa - ta - ta - mu pa - da - mu lē - - - - nam - mi - nā nam - ma rā - vam - ma.)

c13.s3.87

Āṭa Tāla. 5 + 5 + 2 + 2. (The Rāg omits E and A in ascent). Molto vivace. ♩ = 160.

c13.s3.88

Ex. 315.

c13.s3.89

Āṭa is one of the more difficult rhythms. This particular example seems to us almost impossible. We lose our bearings in the course of fourteen crotchets divided into semiquavers, especially at such a tempo as this, and with all sorts of cross-rhythms in the ‘bowing’. The slurs really represent the syllables of words; and this tune is actually sung, and sounds very puzzling but very interesting. The Āvard is divided into four groups (marked I, II, III, IV), and the phrase, which is the same length as the Āvard but does not coincide with it, begins and ends at the third beat of the first group. (Mind the flat!)

c13.s3.90

Later on in the Svaras of this song another sort of rhythm is induced upon this. Every second crotchet of the melody ends on the note Sa, which occurs in the words at the same moment. This would have the effect of combining

c13.s3.92

But the last two Sa’s come one unit late. The original tune began, we saw, on the third beat, and the Svaras begin on the first beat; as there is a Da Capo at the end of them back to the tune, there will be two beats to make up; and this is the ingenious and effective way in which it is done.

c13.s3.94

N.B.—The slurs represent the syllables of the words. The *āvard* finishes at the end of each line. The phrase begins on the third of the fourteen beats.

c13.s3.95

Svaras (patter words).

c13.s3.96

Ex. 316.

c13.s3.97

The Hindostani *Tāls* are very much the same in practice as those of the Carnatic. The usage in different parts of Hindostan varies a good deal, as it does with the *Rāgs*, and it is difficult to be at all sure of more than the commonest. But as these are remarkably like those described as the commonest in the Carnatic, there is little doubt that the underlying idea is the same in both—the poetical metre. The following table gives them as corresponding to the Carnatic list:

c13.s3.98

Counting by mātrā (Hindostan).

c13.s3.99

| | I. | II. | III. | IV. | V. |
| A. | | 4 Ekka Ektāl } | | | 2 4 3 [Madhyamāvati] 2 3 4 [Matta] 4 2 3 [Khaṭṭ tāl] |
| B. | 2 3 [Rūpak] | 2 4 Rūpak | | | |
| C. | | | | 2 3 2 3 Jhampa or Jāp | |
| D. | 3 2 2 Tevra | 4 2 2 Titāla Tinṭāl Tritāla } | | | |
| E. | | 2 4 4 4 2 4 Surphakta Sulaphakata Manṭ } | | | |
| F. | | 4 2 4 4 2 4 4 Āḍa-chautāla } | | | |
| G. | 3 3 2 2 [Jhampa] | 4 4 2 2 Chautāla 2 2 4 4 [Farodast] | 5 5 2 2 Dhamār | | |

c13.s3.100

Amongst these the only Tāl which is commoner than in the Carnatic is G. II Chautāla (i.e. Char-tāla, ‘four-beat’). It is used especially for Dhrūpads (see Chapter XI). The three names of Titāla (D. II) all mean ‘three-beat’. Some of the other names are descriptive—Jhampa, ‘jumping’; Surphakta, ‘zigzag’; Āḍa-chautāla, ‘crooked four-beat’; Dhamār (for dha-mātrā), ‘having a high number of time-units’.

c13.s3.101

But there is one Tāl peculiar to Hindostan which entered it apparently with the Mohammedans—Dādra, 6/8. It is the only rhythm which is treated as a multiple (and therefore cannot be Hindu), and it is in very common use.

c13.s3.102

Most of these Tāls which give the rhythm of one bar (vibhāg) can be taken in pairs of vibhāgs, and are then counted as one bar of common time (Tinṭāl), thus:

c13.s3.103

3 + 2 + 2 3 + 2 + 2 two bars of *Tevra* | d. d d | d. d d | counted 1 2 0 3 like *Tintāl*.

c13.s3.104

Or even a single *Tāl* is so counted: 2 + 3 + 2 + 3 *Jhampa* | d d. d d. | counted 1 2 0 3

c13.s3.105

*Surphakta*, however, has two blank beats (*khālī*): 4 + 2 + 4 *Surphakta* d d d d counted 1 0 2 3 0

c13.s3.106

and *Chautāla* has four beats (*tāl*) and two blanks (*khālī*): 4 + 4 + 2 + 2 *Chautāla* d d d d d counted 1 0 2 0 3 4

c13.s3.107

*Aḍa-chautāla*, again, four beats and three blanks: 2 + 4 + 4 + 4 *Aḍa-chautāla* d d d d counted 1 2 0 3 0 4 0

c13.s3.108

Indian rhythm, then, moves in *āvards* broken up into *vibhāgs*, each of which contains one or more *tāls*.¹ We can equally say of ours that it moves in sections broken up into bars each of which contains one or more beats. In what does the difference between the two systems consist?

c13.s3.109

¹ It is unfortunate that the word is also used in the sense of *vibhāg* to mean the ‘time’ of the music.

c13.s3.110

It may be answered that theirs is derived from song, ours from the dance or the march. That both are based on the numbers 2 and 3, but that they add and we multiply in order to form combinations of these. But the answer which goes deepest is that their music is in modes of time (as we saw also that it was in modes of tune), and that ours changes that mode at will by means principally of the harmony.

c13.s3.111

In order that rhythm, an articulation of the infinite variety of sounds, may be upon some regular plan, the plan must have some recognizable unit of measurement. India takes the short note and gives it for a particular rhythm a certain value as opposed to the long; Europe takes the stressed note and gives it in a particular rhythm a certain frequency as against the unstressed, and graduates its force. We find the unity of the rhythm in the recurrent bar (which is always in duple or triple time, just as our two melodic modes are either major or minor), and have to look elsewhere for the variety; they find variety in the *vibhāg* whose constitution is, as we have seen, extremely various, and must look elsewhere for unity. Both of us find what we want in the larger spaces of time; they find unity in the *āvard*, we find variety in the sections.

c13.s3.112

The *āvard* is a constant. In slow time there may be only two *vibhāgs* in it, in quick time there will certainly be four. And this series of four is unbroken, except at the end of the verse or the song, that is, where the singer breaks off temporarily or altogether. We can see why this must be so. Suppose the song is in *Rūpak* (♩ ♩ | ♩ ♩ |, &c.). Unless there were some accepted number of *vibhāgs* to the *āvard* it would be impossible to know that the music was not in the larger *āvard* (i. e. in fewer *vibhāgs*) of *Tevra* (♩. ♩ ♩ | ♩. ♩ ♩ |, &c.) or of *Jhampa* (♩ ♩. ♩ ♩. | ♩ ♩. ♩ ♩. |, &c.).

c13.s3.113

*Rūpak* (one *āvard*).

c13.s3.114

Ex. 317.

c13.s3.115

*Jhampa* (⅔ *āvard*).

c13.s3.117

*Tevra* (⁶/₇ *āvard*).

c13.s3.119

If the length of the *āvard* were not certain, the rhythmical ‘mode’, which has a connotation of its own just as the melodic ‘mode’ has, would not be clear, and all cross-rhythm, a corner-stone of this music, would become impossible; for this presupposes a constant against which something else is inconstant.

c13.s3.120

The ‘section’ in Europe is often, too often, also of four bars; it is this fact that makes much of our church services musically unattractive. But the section can have, and in the best music it has, a varying number of bars. Its length is altered, either really or apparently; it is either actually shortened or lengthened, or else two equal sections are made to overlap so that one seems shorter than the other. Any work on large lines will supply examples of this; here, for instance, in the passage for flute and bassoon out of *Leonora*, No. 3:

c13.s3.121

Ex. 318. Fl. Passage begins.

c13.s3.124

Bn.

c13.s3.125

Fl. 4 &c. Overlap; bas- 3-bar section. Fl.

c13.s3.126

soon scale suppressed.

c13.s3.127

4 3-bar section.

c13.s3.128

4 Overlap; last bar of previous flute phrase suppressed.

c13.s3.131

Passage ends.

c13.s3.132

in which the sections are twice shortened from four bars to two, and the phrases twice overlap so as to produce the effect of two three-bar sections. It would be clearer if our music were printed as Handel wrote (though not consistently) his *Messiah*, with barlines running all through the staves at each new section, and between these places only through the single staves; it would have the effect which punctuation has upon the written word (and would be as difficult to get right). An instance of the judgement he exercised in this matter may be seen in the two versions, given in the autograph score, of the opening symphony of ‘But who may abide’.

c13.s3.133

There is one case, however, which puts the thing in a nutshell. In 1893, when 5/4 rhythm was less familiar to Europe than it now is, Tschaikowsky felt it necessary in his Sixth Symphony to do two things. He kept the metre free from complications; the Scherzo is almost entirely based on the crotchet, and the Trio on the minim. He also kept the sections uniform; they are in four-bar rhythm throughout except in two places—at the return from the Trio to the Scherzo, and at the Coda—the very places in which, as we saw, liberty was allowed in the *āvard*. It was only there, in fact, that he felt his rhythm so securely established that he could play with it. But now that it no longer piques our curiosity to see how 5/4 is ‘beaten’, and that the problem of counting it accurately no longer provides us with a painful pleasure, the monotony of the sections comes back upon us with fatal force. But the four-bar section would not have this effect if, like the Hindus, we had a large assortment of variously characterized rhythms to fill it.

c13.s3.134

It is harmony which puts it in our power to prevent this frequently changing length of the phrase or section from upsetting the unity of the rhythm. In this Adagio from Haydn’s First Quartet:

c13.s3.135

Ex. 319.

c13.s3.136

the phrases are made to stand out from each other by means of the harmony : *a* and *d* taper towards the end, *c* steals in unawares, and *b* contrasts firmly with its neighbours *a* and *c*; and that such insistence upon the rhythm was necessary is shown by the base form into which the melody has relapsed by the constant use which has ignored it.

c13.s3.137

In the Menuetto which follows it :

c13.s3.138

Ex. 320.

c13.s3.139

the harmonies by their postponement to the second bar clearly mark the beginning of a section there, and tell us that the phrase of the first bar is to be regarded as *extra sententiam*. Then by entering at the normal place in the sixth bar they proclaim that that phrase is now taken into the sentence. They are not responsible for the assertion that bars 6, 7 and 8, 9 each make a section ; that is made clear by the balance of the phrases they contain—the phrase of 8, 9 is an inversion of that in 6, 7. But the harmonies again compel us to take bar 10 as a section all to itself, for they exactly repeat those of bar 9, only a place higher. This somewhat intricate arrangement of sections would be confusing but for the fact that they make up in all a period of twelve bars—that is, the time length of three normal sections—and so their excessive variety is unified.

c13.s3.140

Very similarly in Chopin’s Sixth Nocturne a twelve-bar unity is reached by a different distribution:

c13.s3.141

Ex. 321.

c13.s3.142

and this is the work of the harmony, more especially of the tenor part:

c13.s3.143

Ex. 322. Preliminary. overlap with new phrase.

c13.s3.144

But these two examples are the opening phrases of two several compositions where a certain clearness of balance is necessary and some general unity (here twelve bars) is indispensable: in the course of a composition (as we may conclude from Ex. 318) no such higher unity need be or is aimed at; the sections may be as various as the composer pleases.

c13.s3.145

It is needless to give instances where the harmony graduates the stress of the rhythm, as every page of music is full of them. This is a distinction which Indian music is also fully alive to, as we shall see in the next chapter.

c13.s3.146

It is sometimes thought that these uneven times—5, 7, 10, 14, and so on—are full of suggestion for European composers. This on the whole may be doubted, because duration is not the same thing as stress. All these Indian rhythms have their *raison d’être* in the contrast of long and short duration, and to identify these with much and little stress is to vulgarize the rhythms. The only genuine instance we have of their method is Saraband rhythm, and we see instantly how in

c13.s3.147

Ex. 323.

c13.s3.148

La - scia ch'io pian - ga mia cru - da sor - te

c13.s3.149

the effect is lost if the second minim in addition to being lengthened is also emphasized. Stress pulses, and demands regularity; duration is complementary, and revels in irregularity. In order to get the true sense of duration we have to get rid of stress, and this would mean that we must find some other means (as the Hindus do) of marking the beginning of the bar than by accenting it. Tchaikowsky’s Scherzo and Trio do this. The Scherzo is in elongated trochees (7 + 3 instead of 6 + 3):

c13.s3.150

Ex. 324.

c13.s3.151

instead of

c13.s3.152

Ex. 325.

c13.s3.153

Both these phrases are characterized; the first by crotchets with an incidental triplet, the second by the dotted minim. This rhythm pervades the whole, although the units are afterwards slightly varied. The Trio is a truncated Saraband, and by its figuration appears as a contracted iamb (2 + 3 instead of 2 + 4):

c13.s3.154

Ex. 326.

c13.s3.155

instead of

c13.s3.156

Ex. 327.

c13.s3.157

The way the composer has defeated the stress is by a variety of devices; in the Scherzo by the cross-bowing, running quavers, and syncopation of the accompaniment, in the Trio by the crotchet drone bass. And if the conductor would now lay down his stick for this passage there would be nothing to distinguish it from Indian rhythm.

c13.s3.158

But the fact remains that we feel these things as elongations and truncations of rhythms which are familiar to us, and they could hardly be naturalized. Perhaps this is why in his variations on a Hungarian air in 7/4 (Tevra) Brahms drops in the fourth variation into common time beginning on an off-beat, in the fifth and sixth into 3/4 beginning on an off-beat, in the seventh into overlapping phrases, and continues from the ninth onwards frankly in 2/4 and 6/8. Perhaps he felt that he could best explain the nature of this rhythm by showing that it was like all these things which we have, but was not the same as any one of them.

c13.s3.159

¹ The two movements (Exs. 324 and 326) correspond practically to C. IV (7 + 1 + 2) and B. I (2 + 3).

c13.s3.160

But considered as forms of our rhythms gone mad they have a definite use, as, for instance, in Tristan’s delirium (Act III, sc. ii), where fragments of melodies, his whole past in fact, come thronging disjointedly into his mind. Only these are not Indian rhythms. They are European rhythms distorted to suit the state of his bodily pulse; it is precisely the pulsation of them that is insisted upon.

c13.s3.161

The modern use of these unusual rhythms is again different. The constant changes of time—some half dozen perhaps, 7/8, 5/8, 2/4, 7/8, 1/8, 3/4, &c., occurring in not more than twice six bars—are part of the striving of the last twenty years after realism and away from formalism. Composers aim at substituting the rhetorical accent of prose for the metrical accent of verse, at imitating in instrumental music the broken accents of the speaking voice. As such their music has nothing in common with the systems we have been discussing where metrical law prevails. They appear to defy law; and they are justified, but only by success in the particular instance. Thus, of two such instances, we may say that Cyril Scott’s Sonata, Op. 66, is less convincing rhythmically than Stravinsky’s Petrouchka. For the latter has large tracts of uniform time which enable the vagaries to be felt as contrast, whereas in the former there is no such contrast—because it is all contrast.