steps produces, all over the world, melodic forms which are generally classed under the common heading of 'pentatonic music'. The use of this term shows the readiness of musical theorists to take scale instead of melodic structure for the primary element in music. When two fourths, each of them divided by an intermediate note, are linked with one another, and supplemented by the octave of the starting note-which often enough does not actually occur in the melodythe result is indeed a pentatonic scale: GXDYA(G). Within this pattern the arrangement may vary according to the intonation of the intermediary or passing notes (X and Y), which again depends on the melodic 'form' and its demands. If the two fourths delimit identical or equivalent melodic phrases, they will both be divided similarly, either into two halves, hereby giving existence to a sequence of five-quarter tones: GEDCBA; or-in agreement with a tendency formerly mentioned-having the larger interval below, e.g. GFDC-A; or, again, having the smaller interval below, e.g. G-EDBA, this being explained by the attraction which predominant notes exercise on leading notes. All these arrangements which occur frequently in non-European songs (and in old European folk-songs) avoid the semitone. This does not imply that it is altogether unfit to do duty as a melodic step. In primitive song (Wedda) there are even narrower steps; it is only sequences of semitones ('chromatic' sequences) that are lacking in non-harmonic music, with rare exceptions. But there is another way to account for the avoidance of the semitone. A division of the fourth into major third+semitone would be too remote from the principle of equal division; further, in pentatonic melodies with their notes arranged as above, one intermediate note and one of the predominants are more intimately connected, being in a fifth relation (the intermediate notes among themselves are in fourth relation like the predominants):

GF-DC-A

G-ED-BA

(There are, however, also pentatonic scales with semitones, as G-ED-BA [Japan].)

Like tonality (the arrangements of notes) rhythm and harmony (as far as it exists) in non-European music show characteristics which are the natural outcome of pure melody. These characteristics will be discussed in the following section. As they are universal, and even in our music are only checked by the development of harmony, they cannot be typical of a primitive state. There are forms of civilization which we have reason to consider as typically primitive, e.g. that of the Weddas or the Fuegians. Their music shows as clearly as anything else characteristics which, being due to a narrow range of consciousness, really represent an early stage of development. Their melodies consist of a single short phrase made up of only two to three notes of a very small compass (smaller than a fourth) repeated indefinitely. Primitive melody in this sense is no longer found in Africa except in rare cases, but is perhaps typical of Pygmy music. As evolution continues, however, the number of notes is increased, their compass is enlarged and the tunes lengthened; only the perseverance in repeating the same phrase over and over again persists-probably for the sake of the ecstasy which it rouses.

III. THE CHARACTER OF AFRICAN NEGRO MUSIC

Antiphony, Part-Singing, Rhythm

As we know our own music so well, it is comparatively easy for us to name those features which distinguish it from all other music. But it is much more difficult to specify the idiosyncrasies of any one kind of foreign music, even of so large a group as the African. This does not mean that we do not notice the peculiarities, e.g. of American Indian, Melanesian, and African music when we hear it sung. On the contrary, even without close study, we very easily arrive at distinguishing and recognizing them. But what makes us distinguish between such categories as American Indian, Melanesian, African, is just the unanalysed total impression. This impression cannot be described in words, and it is necessarily destroyed by analysis. But even in staff notation, although it may be done very carefully, many essentials besides the timbre get lost; for notation, in order to be readable, must reduce facts to formulas. Still, it retains enough of the

original to convey an idea of the character of any particular kind of music, and is an indispensable basis for a description or an analysis.

Proceeding then to characterize African native music with the help of a few musical illustrations, we must begin by making some reservations. In the first place the music of the Islamic North will be excluded; for, although it shows traces of negro influence, it belongs to the Arabic-Persian civilization. The music of the Pygmies will also be left out of account because we are not in possession of sufficient facts to come to any conclusions about it. Lastly, the study of the subject has not yet been pursued far enough to bring out typical distinctions between different tribes, even between such large groups as the Bantu, Sudanese, and Hamites. Such distinctions certainly exist, and survive even where different tribes living in close contact with each other tend towards assimilation, as in Ruanda.' But it will not be possible firmly to establish musical dictions and, later on, dialects until a great deal of individual research has been carried out; nor can we decide meanwhile whether, and to what extent, these groups coincide with those marked by language and culture. Comparative musicology is thus, for external reasons, forced to proceed downward, as it were, and by differentiation; and has, therefore, in certain respects, an advantage as compared with linguistics and ethnology. These began by having a bewildering number of individual facts at their disposal; starting from these facts, they had to search for resemblances, to prove connexions, and are now on their way upwards from dispersion to increasingly large units. This synthetic process is directly opposed to the natural process of differentiating a unit, and therefore has difficulties and pitfalls of its own.

In African music, three features stand out above all others, and have been noticed and stressed accordingly by all those who have heard Negroes sing: antiphony (here understood to be the alternate singing of solo and chorus), part-singing, and highly developed rhythm. None of these features is confined to Africa, antiphony being also in use, e.g. among North American Indians, part-singing among the South-Sea tribes, highly developed forms of rhythm in

1 Czekanowski, Wiss. Ergebnisse d. D. Zentr.-Afrika-Exped., 1907–8, VII, 379 ff., Leipzig, 1918.

Indonesia. Still their sway over, and the fullness and variety of their forms in African native music let them appear as characteristic of it.

In almost all African song a soloist alternates with a chorus. Solo performances like those of the Sudanese bards in praise of chieftains rank far behind this form. The predominance of antiphony may well be connected with the Negroes' talent for extempore recitation. The soloist gives himself up to his spontaneous inspiration and the chorus repeats the newly-formed stanza, or at least its refrain-this, too, attesting the high musical faculties of the race.

Songs produced in this way are often remembered, especially those which, from the first, are associated with worship or other ceremonies. On the other hand, the ease of production counteracts the longevity of individual songs. What endures is the style of the music rather than individual examples of it. Tradition and creative power in alliance result in a variety of forms unequalled, perhaps, in any other race, simple and complicated forms coexisting within the same tribe. The complicated forms may be considered as having sprung from the simple ones; but we have no means of deciding if these hypothetical stages of development represent the true course of events (and it is difficult, therefore, to foresee to what results further undisturbed development will lead).

We can, nevertheless, see clearly that the typically African forms of polyphony arise from antiphony. In many cases, the soloist begins a new stanza while the chorus part of the preceding one is not yet finished. This overlap is produced quite naturally and, no doubt, was originally unintended. Everywhere singers sustain the final note of a tune regardless of the metre or the uniformity of the bars, and accordingly the soloist is unable to know how long the sustained final note of the chorus will last. The simplest case, then, is that the final and the starting note of a tune meet. The result is a dichord; but as it is accounted for by a melodic (and not by any harmonic) principle, it may be an interval of any kind, possibly a dissonance, and may vary from one stanza to another according to the frequent variations in the solo part. But frequently also it is a fourth or a fifth, owing to the importance of these intervals as constructive elements of melody. This happens particularly when the final and the starting note are,

at the same time, the extreme notes of the melodic compass, i.e. when the melody simply descends from the upper end of a fourth or a fifth to the lower one. The richness of simultaneous sounds differing in pitch as compared with unison may strike singers even when the divergence is accidental, and may then be kept in reserve and used only as a final effect. As a matter of fact final fourths and fifths often occur, e.g. in Wanixa songs, not as a result of overlapping, but by the chorus dividing into two to execute the dichord. Further, this manner of increasing the volume of sound is not always confined to the last note, but extends over several notes at the end of the phrase, and occasionally over the whole of the last line of the stanza (e.g. in Wanixa and Wanyamwezi songs).1

The musical form arrived at in this way is well known from the sacred music of the early Middle Ages; it is the organum in parallel motion which represents a primitive stage of polyphony. In the tenth and eleventh centuries freer forms, in which the voices deviated from strict parallelism, grew out of the organum; and the same thing occasionally happens in East Africa (Wasukuma). Here, too, only the three most consonant combinations of sound, octave, fifth, and fourth, are used. This kind of polyphony also is based on pure melody, and has nothing to do with harmony as we understand it. This is clear from the fact that the accompanying voice, which generally lies above the principal one, pursues its parallel motion regardless of tonality; it only increases the volume of the sounds in about the same way as a powerful partial tone would, and gives them a different timbre. It does not, however, change the notes of the melody into 'chords', with their special functions which differ from those of purely melodic notes. Thus the melodic structure of the occupation song given as Example 1 rests on the relation of the fourth and fifth which exist between the principal notes (A, E, D) and between the extreme notes (C-G, B-E, G-D) of the three phrases. In the formula appended to Example 1 as an illustration of the melodic structure these relations are indicated by square brackets, the connexion between predominants and the notes leading up or down to them, by slurs, 1 Anthropos, iv. 781 ff., 1909, Ex. 1.

I

2 Bull. de l'Acad. d. Sciences de Cracovie, Sc. nat., 1910, 711 ff.