The collector need not be musically gifted; on the contrary if he consults his own musical taste, he is likely to choose songs which show traces of European influence. The machine can reproduce music immediately after registering it. The reproduction never fails to delight native musicians, and so the phonograph is often used by travellers as a means of forming friendly relations with the population; curious persons come from neighbouring districts and crowd round to see the miraculous instrument.1

As material for study, phonograms are immensely superior to notations of melodies taken down from direct hearing; and it is inconceivable why again and again the inferior method should be used. To begin with, only by means of the phonograph can we get the 'real thing'. It is generally supposed that the substance of a song can be written down in staff notation, with the help, may be, of diacritical marks and an explanatory text. But this is a typically European prejudice brought about by the evolution of our music and by our general way of thinking. The singers themselves attach as much importance to the timbre of the voice and the mode of recitation as to anything else, and very often more. In fact, timbre and recitation appear to be racial characteristics deeply rooted in physiological functions and give therefore valuable evidence of anthropological relations and differences. Peoples and their music, then, are not so much distinguished by what they sing as by the way in which they sing. What should we think of a student of languages who disregarded phonetics? And how could anybody study a language phonetically without having heard it spoken? But even confining ourselves to what can be expressed by musical notation, we are apt to go wrong unless we proceed with the utmost care. This is true even when we transcribe phonograms, although in this case there are a great many facilities at hand: difficult passages, even intervals and notes can be singled out and repeated over and over again; sounds can be intensified or softened down; pitches and intervals can be fixed precisely by means of a tonometer, time and rhythmical pro

1 Cf. R. Pöch, Sitzungsber. d. Wiener Akad. d. Wiss., Mathem.-naturw. Kl., Abt. II, 126, 3, 1917.

2 Melos; 1921, Heft 9 ('Musikal. Exotismus').

portions by a metronome; and, last but not least, one can begin by transcribing the simpler passages, and gradually get accustomed to the foreign musical diction, compare one part of the record with any other, interrupt the work when tired and come back to it with renewed strength-and all this in the peaceful atmosphere of a study. Even so, a great amount of practice and constant revision are required to avoid mistaking what has really been sung or played for intervals and rhythms prevailing in our own music. Talented and conscientious hearers can, indeed, succeed in noting down music without the help of the phonograph (we have examples of this from older periods), but what difficulties do they not have to surmount? A native singer who is intelligent enough to understand what one is aiming at may be willing to repeat a tune over and over again. But although he may have the best of intentions he will, in most cases, be unable to repeat it without varying it in some way or other; and even less will he be able to recite separate passages without losing the thread which connects them. A good way of becoming familiar with foreign musical expression is to learn the melodies oneself vocally or on an instrument. But it is never possible to be certain of having reproduced them correctly, even when a native instructor has declared himself satisfied, for either he may be very polite or merely tired of teaching. To grasp a dance tune during the turmoil of a performance will be still more difficult and generally impossible. The reconstruction of music from memory, with the help perhaps of stray notes, has also been undertaken; we have glaring examples of the degree to which melodies have been transformed and disfigured by this process.'

Phonograms alone therefore offer reliable material. No musical talent, no special instruction, no technical skill, and but very little effort is required to obtain the records. They are required by the person who works them out (who, of course, need not be the collector himself). The transcriptions can be revised at any time as it is possible to save the records permanently from destruction by an electrotype process and to obtain any number of copies desired. 1 C. Stumpf, Anfänge der Musik, Leipzig, 1911, S. 69 ff.

2 'Phonogr. Methoden', Abderhaldens Handb. d. biolog. Arbeitsmethoden; 1923.

D

II. GENERAL CHARACTER OF NON-EUROPEAN MUSIC. MELODY

VERSUS HARMONY

The opportunities of hearing the music of different nations side by side and of comparing them, provided by the existing collections of phonograms, have suddenly immensely widened and deepened our knowledge. Many prejudices had to be overcome. In the first place, of course, the general distinctions between native and European music became evident. The main difference is this: our music (since about A.D. 1600) is built on harmony, all other music on pure melody. In fact it is non-European music which has made us remember what pure melody really is. The consonance of thirds and triads, the stereotyped cadence, the contrast between major and minor-mostly held to be identical with the contrast between cheerful and mournful -the exact intonation of certain intervals called for by harmony and necessitated by instruments with fixed tuning-all this has grown upon us so much that we are inclined to think of it as the natural basis of all music. But pure melody—unlike ours—is not conceived harmonically; and as long as we cannot divest ourselves of the idea— so natural to our way of thinking—that it is based on harmony and interpret it in this sense, we arbitrarily change its meaning and cannot arrive at a proper understanding of it.

In purely melodic songs certain natural traits have maintained themselves which in our harmonic music have been superimposed upon or supplanted by other traits. They are 'natural', i.e. rooted in the psycho-physical constitution of man, and can therefore be found all over the world. The natural motion of melody is downward; like breathing or striking, from tension to rest. Typically melodic steps are comparatively small, amounting, at most, to a major third or so. But within this group of small intervals no distinctions are made: a 'third' is not different in its function from a 'second', and the width of an interval can vary time after time within wide margins. What we call 'out of tune' entirely loses its sense here, for there is no norm and consequently no deviation from a norm. Nor can we, with regard to vocal melody, speak of 'scale', this term implying a suc

cession of fixed degrees. The singer does not possess in his mind a store of notes or intervals from which he selects and combines those which may please his ear. What he really aims at more or less consciously is melody forming an undivided unity which he performs at one stroke as an athlete does an exercise. This melodic unity determines the parts which we afterwards dissect in the course of our analysis. Recent experimental investigations have shown how, and in what way, the size of intervals in unaccompanied song depends, even with our singers on the 'form' (Gestalt) of a melody: 2 owing to natural (psycho-physical) agents a wide leap overshoots its mark, the step to the centre of gravity from the leading note is narrowed down; intervals grow wider when a melody descends from high to low notes; steps following each other in one direction tend to become equal in size, an inserted note divides intervals of any size in certain simple (interval) proportions. It should be noticed that all these principles pay no attention to the consonance, but to the distance between notes (the size of steps)3 which is the more primitive and more important element in melody rather in the same way that in pictures the relative degrees of light are more important than those of colour. This point—which cannot be too much emphasized— constitutes a chief distinction between purely melodic and harmonic music. We cannot comprehend pure melody really and adequately unless we keep this in mind, and disregard the harmonic relations which we are disposed to hear (just as a painter disregards colour in favour of values).

On the other hand, pure melody does contain elements related probably to those which, in our music, have contributed to form harmony. Certain tones can be said to 'fit together' more than others. This mutual relation makes them sound 'consonant' when they are struck simultaneously; and it is likely that the same kind of relation is also felt in succession. But as the relation in this case is certainly 1 B. F. Gilman, Journ. of Amer. Archaeol., v, 1908; Jahrb. d. Musikbibl. Peters, 19,

1913.

O. Abraham, Psycholog. Forschung, 4, 1, 1923; Handb. d. Physik, hrsg. v. Geiger u. Scheel, Bd. 8, Kap. 9.

3 O. Abraham u. E. M. v. Hornbostel, Ztschr. f. Psychol., 98, 233, 1925.

4 Handbuch d. Physiol., hrsg. v. Bethe u. a., Bd. II, 714, 1926.

much slighter, it only affects intervals of perfect consonance (octaves, fifths, fourths, but not thirds); and even these need not be 'in tune'. As the compass of melody, in most cases, is small, fourths and fifths are more important and more frequent than octaves. But even they are too wide to be used in melodic sequence; one rarely leaps from a note immediately to its fourth or fifth. They are rather used as constructive elements determining the distance between predominant notes, the compass of melodic phrases, and the rise or fall of a melody when its position is changed (transposed). In this way they mark out the field where melody is at play. We can say, therefore, that pure melody, too, has 'tonality', its tonality being established by the function and the mutual relations of the notes. There are predominant notes of greater or smaller weight ('tonic', 'dominant'), satellites gravitating towards them ('leading notes' in ascent or descent), servants or assistants presenting themselves only when needed (e.g. passing notes), in short, a complete hierarchy. The position of the different functionaries both within the melodic unity and towards each other characterizes the mode of the melody. Mode-in the sense in which we speak of the ecclesiastical modes—is vitally different from scale. A scale fixes the relative pitch of the notes, and therefore, the intervals without taking account of their melodic function; a mode, on the contrary, only determines the melodic function of the notes, whereas their relative pitch and the size of the intervals are merely an outcome of that function. Entirely different modes may have a common scale, e.g. EDCBAG: one mode may have G and D as predominant notes (tonic and dominant), A and E as their leading notes in descent, C and B as passing notes or rather as one passing note which, being but slightly touched, varies, or is unsettled as regards intonation; in another mode A and E may be predominant, B and G may lead up to the tonic A, and D and C may be passing notes used alternately. On the other hand, one mode may yield several scales; so when the leading notes of the above-mentioned modes are moved towards the predominant notes: EDCBA,G; EDCB AG, &c.

The prevalence of the fourth as a frame for melodic phrases together with the necessity of dividing it in order to obtain melodic