544 ANNUAL REPORT SMITHSONIAN INSTITUTION, 1941 



the phonodeik, an instrument of Dr. Miller's invention, and an 

 analytical study of the result was made by Dr. Miller, with a com- 

 parison of the photographs and the transcriptions of the same songs 

 by hearing. Dr. Miller stated "the close agreement of the two 

 methods hardly justifies the great amount of labor involved in the 

 photographic method. This study was undertaken principally to 

 learn what could be done if it were desirable." ^ 



In determining the meter of the songs I use an ordinary Maelzel 

 metronome which was tested at the Bureau of Standards. The metro- 

 nome is not a precision instrument. The marks on its scale are not 

 near together, and the "bob" is some distance from the scale, but this 

 metronome was found to be reasonably accurate with the bob set at 

 120. on a level with the eye. This indication is about midlength of 

 the scale. For very slow or rapid songs the instrument is placed in 

 this position and the tempo indicated by the position of the bob. The 

 exact tempo of Indian singing is not important, and this mode of 

 measurement is suflScient, the same metronome and method being used 

 with all the songs. 



Having determined the meter of the song, it is necessary to note the 

 accented tones by which the transcription is divided into measures. 

 The use of measures does not imply that the Indian has any knowl- 

 edge of our musical customs, but it is a convenient form for showing 

 the rhythm of his musical performance. Each accented tone is tran- 

 scribed as the begining of a measure, regardless of the time interven- 

 ing between the accents. In some songs the accents are evenly spaced ; 

 in others they seem erratic, but on further study they often combine 

 to form a rhythmic pattern. Such a pattern usually comprises several 

 measures and is designated as a rhythmic unit. Sometimes a 5-8 

 measure is followed by a 3-8 measure. The note values may suggest 

 two measures in 2-4 time, but the accent divides the series as indicated. 

 A measure transcribed in 7-8 time cannot be divided, as there is no 

 secondary accent. Quadruple time rarely occurs, but 2-A: time is com- 

 mon in the songs. The accents in a song do not always correspond to 

 the accents in the words of the song when spoken. The rhythm of the 

 song is the rhythm of the melody in the mind of the singer. 



The tempo sometimes changes during a song. Such a change may 

 be either abrupt or gradual, and in the latter instance the new time 

 indication is shown when the new tempo is established, preceded by 

 "ritard" or "accelerando." A question to be determined is whether 

 the change is intentional. The several renditions are compared, and, 

 as a general rule, the change is found in all the renditions, showing 

 it to be part of the song. Old Indian singers have a remarkable sense 

 of both pitch and tempo. Thus, Mrs. Holding Eagle, a Mandan, re- 



' Densmore, Frances, Northern Ute Music. Bur. Amer. Ethnol. Bull. 61, Appendix, pp. 

 206-210, pis. 12-15, figs. 20, 21, 1918. 



