Speed Variations in the Frances Densmore Recordings

By David Swenson

Currently the American Folklife Center, Library of Congress houses the Frances Densmore collection, a collection of more than 2,500 American Indian songs recorded by Frances Theresa Densmore between 1907 and 1941. Of these catalogued recordings approximately 260 were made in Fort Yates, North Dakota between 1911 and 1914. Through investigation it was discovered that this collection of the Fort Yates recordings has been catalogued and transferred to digital tape at a speed that is approximately 10% slower than the original recordings. This paper includes historical perspective on Densmore and her work along with the method used for establishing the correct playback speed.

Contemporary Lakota culture represents not only the values of today’s American Indian society; it is also a very real connection to the past. Theirs is a culture of strength and family, of power and gratefulness. They acknowledge the past and expect the future. At their best, they value and honor the land; they treat it as a living member of their own family, and the Lakota sing about all these things. The music of the Lakota people is proof of a culture that has survived in spite of all that it has suffered.

The collections of wax cylinder recordings made by Frances Densmore in Fort Yates during the early years of the 20th century are a voice from the past that speak to the Lakota today. These songs reaffirm a reverence and strength of a proud people. Correcting these recordings is a small return that can be made to a living society. It is cleaning the glass to see the past clearly, and to learn from it.

Ethnomusicologist Frances Theresa Densmore (1867-1957) was born in the small town of Red Wing, Minnesota. Her first experience with Native music was as a child when she heard the singing of tribes camped across the Mississippi River. From 1884 to 1887, Densmore studied at Minnesota’s Oberlin College Conservatory of Music. After leaving Oberlin, she moved to St. Paul, Minnesota where she taught piano and gave lectures on Wagnerian Operas. Inspired by Alice Fletcher’s studies of Omaha music, Densmore eventually added the subject of American Indian music to her lectures. At the time she stated that, “I availed myself of every opportunity to hear Indians singing at fairs and other exhibitions and began a systematic course of reading on the history and customs of the American Indians.” She began to write accounts of her experiences transcribing Native music. One story tells of meeting Geronimo at the 1904 Louisiana Purchase Exposition. She stood behind him listening to the songs he sang and wrote down the melody of each, watching as he “printed his name in careful letters on cards to sell to passersby.”

In 1905 Densmore visited the annual Chippewa summer celebration in White Earth, Minnesota and became “more and more impressed with the idea that I must record Chippewa songs.” In 1907 she petitioned the Smithsonian Institution's Bureau of American Ethnology for financial assistance. Thus began her 50 year association with the Bureau. With an initial grant of $150.00 allotted by Prof. William H. Holmes, Densmore purchased an Edison Home Phonograph which, according to her, “was noted as the best recording equipment available at the time.” She returned to White Earth, recorded the Minnesota tribe and later published her observations in The American Anthropologist (April-June 1907).

Columbia Gramophone

In 1908 home recording was at the height of its popularity. The Bureau replaced her Edison system with a Columbia gramophone; “a machine made to meet the demand.”
Certain characteristics of this unit suited Densmore’s tastes better than any other and she continued to use it in some capacity until 1940. All of the cylinders made at
Standing Rock were recorded on this machine.

Densmore's office in Fort Yates on the Standing Rock Reservation

In a 1941 Smithsonian bulletin she described her working conditions at Fort Yates:

“I remember with queer affection an office at Fort Yates, N. Dak., that had been part of the kitchen of the old fort. Subsequently it had been used as a coal shed, and it had neither door nor windows when I took over. The agent let a prisoner from the guardhouse help me fix it up and he suggested boring holes in the floor to let the water run through, when the floor was cleaned. He made steps, rehung the door, and nailed window sash over the openings, and I pasted paper over the broken plaster and used packing boxes as tables. For many weeks I used that office, and the Indians felt at home there, which is important... It is a rare combination if I have a comfortable place to stay, an interpreter, singers and a place to record all at the same time.”

This rare combination that existed at Fort Yates led to the publication of Densmore’s noted work, Teton Sioux Music. Her work there produced over 260 catalogued recordings of Lakota songs. Teton Sioux Music is commonly regarded as the benchmark of Sioux ethnology studies. It has remained the standard reference since its publication in 1918.

Densmore's office in McLaughlin, SD, 1913

Densmore’s writings contain detailed melodic and rhythmic analysis of the music she recorded. For example, 140 songs were of a minor tonality, 94 of the songs were in 3/4 time, 35% were comprised of 8 tones, etc. While this detailed analysis is very informative, the limitations of western notation become evident when one matches notes to voices that don’t always fit a perfect western 12-tone scale. Densmore herself noted that while the song could be tested for intervals on a tuned instrument “a native melody... can best be reproduced vocally.” Proof of this statement is the wax cylinder recordings themselves.

Kevin Locke, the well-known Lakota dancer, musician and performer introduced me to Densmore’s work. His interest was in learning the old songs she had recorded and he was having trouble understanding the singer’s words. He impressed upon me the cultural importance of preserving these old songs.
In Teton Sioux Music Densmore writes, “The songs comprised in the first group, almost without exception, were recorded by men 65 to 80 years of age... This group contains 147 songs believed to be from 50 to 150 years old... The oldest songs, which are considered the best songs, ‘were composed in dreams.” Locke referred to her work as a preservation of the “old songs of the Lakota people," and described the recordings as “hard to understand” and “very old sounding”. He expressed his desire to get a clearer, cleaner copy and wondered if I could help. I listened to his personal cassette copies. They were, in fact, very poor quality audio, barely discernible. At that time an attempt was made to improve the sound using the available studio equipment. He was kind in his assessment of my effort, but I felt there was much more that could and should be done. I was working from Kevin’s analogue cassette copy that was several generations removed from the master. In order to make any notable improvement I needed a source master closer to the original.

In October 1994 I contacted Chris Dill, Director of the Museum Division of the State Historical Society of North Dakota and Gerald Newborg, Director of the State Archives and Historical Research Library Division of the State Historical Society. I asked for permission to review the North Dakota State Archives’ copy of the Densmore recordings. While this copy was far better than Kevin’s cassette dubs, I was still unhappy with the sound of the tapes. These were analogue copies made in 1981 and they too were of marginal quality. Again, something just didn’t seem right.

I then began a correspondence with Judith Gray at the American Folklife Center about purchasing a set of digital masters. I was granted permission by Standing Rock Tribal Council Chairman Jesse Taken Alive to purchase a copy from the Smithsonian and received the digital tapes in November 1995. My plan was to make an indexed CD set and make them available to the library of Sitting Bull College at Standing Rock. Initial work proved frustrating; the sound of the tapes was still very bad. Something just didn’t seem right.

I worked with the Sonic Solutions digital system in an attempt to improve the sound of the tapes and spoke to their technical staff regarding the recordings. It was determined that processing the tapes with the Sonic system could aid in the restoration, but I was still unhappy with the results. Some cataloging, restoration and copy work was done, and then the project was shelved during a business transition.

In September 1996 while reviewing the Densmore recordings with some traditional Lakota singers, the actual transfer speed of the tapes became suspect. I’m a firm believer that oral interpretation by a member of the culture is an essential element in the archival process. I investigated further and eventually came to the reel marked AFS 10,582B. I hadn’t recognized it before, but that reel contained several of the few recordings Densmore had made of women singers at Standing Rock. It was difficult to recognize the singers as women because the recordings didn’t sound like women at all. They sounded like men. When I increased the tape speed by approximately 10 to 11 percent the tapes sounded correct. After checking other masters in the collection it became immediately apparent that the entire collection had been transferred to the current version of digital masters at an incorrect speed.

In October 1996 I began a correspondence with Judith Gray at the American Folklife Center and Tom Vennum, Senior Ethnomusicologist, Office of Folklife Programs, Washington, D.C. regarding my concerns. In a letter dated 11/7/96 Judith Gray responded... “Densmore herself supervised the transfer of the cylinders to discs in the late 1940s and early 1950s... Those discs were the basis of the preservation tapes made in the 70’s or 80’s, from which your copies were made.”
She goes on to discuss the tone that Densmore placed at the beginning of her recordings for pitch reference... ”it strikes me that another unknown is the exact pitch Densmore used. I remember seeing her pitch-pipe in the collections at the National Anthropological Archive... would be surprised, however, if its internal mechanism still operates well enough to know with any certainty what “A” she sounded on the cylinders. I doubt she was using anything as high as A-440.”

On November 22, 1996, I received an email from Tom Vennum...”Densmore for transcription accuracy often blew into a pitch-pipe - I would guess concert A, although that has changed since 1910 - so that she could later adjust the playback speed.
CAVEAT: she seems to have been using a chromatic pitch-pipe, so what she actually blew is anyone's guess... ALSO one must be careful that today's Native singers aren't imposing their own aesthetic judgment when they say the song from 1910 is "too low, too slow (which from the standpoint of a recording are clearly the same thing). Most Ojibwa singing today is considerably "higher and faster" than it was earlier in the century - partly due to Siouan influence already at work in late 19th century. A ticklish problem for you: scientific accuracy vs. current community taste. Good luck with your worthy project.” Our correspondence continued with me asking for specific information on how the transfer speeds were determined. The reply was...”transfers were made at the Library of Congress. Seriously doubt anyone would remember or have notes.” Frustrated with the lack of documentation regarding this, I began to search local resources and found the answer at my local library in Densmore’s own writings. The following is from her article, “The Study of Indian Music”, which appeared in the Smithsonian Annual Report of 1941 (publication 3651), and was found in Vol. XXIII - Contributions from the Museum of the American Indian Heye Foundation, New York, 1968.

“Frances Densmore and American Indian Music”
A Memorial Volume compiled and edited by Charles Hofman

TRANSCRIPTION OF RECORDS

“The transcribing of records is seldom done in the field, as time is so valuable and facilities are limited. The speed screw of the phonograph is removed when the instrument is shipped and it is necessary to adjust the speed of the instrument when songs are transcribed. Without this adjustment the pitch would not be the same in recording and transcribing, and the two performances would not be uniform. The desired speed is 160 revolutions per minute and this could be attained by counting the revolutions of the mandrel, but I devised a different method. The tone of C of a pitch pipe was recorded on a wax cylinder. This is placed on a phonograph and the speed screw adjusted until the tone produced by this record is the same as that of the pitch pipe. The piano used when transcribing is tuned to the pitch (A-440). Thus the pitch of the singer’s voice and the original tempo are preserved.”

Her article answers both the question of what tone she used on her pitch pipe and her acknowledgment of A-440 as a reference pitch. While it is true that a concert “A” has changed (the worldwide standard was established as A-440 in the 1930s) it is clear that Densmore used the American standard A-440 as her reference. It is also clear that she recognized an accurate playback speed was extremely important. When matching the pitch played on the current Smithsonian masters to contemporary pitches, one finds that her pitchpipe “C” has been slowed down to match an A-440 making the playback speed of the tapes a minor third off. When listened to at the original playback speed, the difference is quite dramatic. It’s as if a culture has reawakened in song. The singers' voices become real, the melodies vibrant. One can even hear the sonic artifacts of that old room in Fort Yates.

An additional method to establish the correct speed involved using the defects found in the recordings themselves and analyzing them with the Sonic Solutions digital audio system. One of the reasons Densmore had the recordings transferred to reference disks was that she recognized the impending deterioration of the original wax cylinders. In fact, some of the cylinders were cracked but still playable when she oversaw the transfers. When viewed on the Sonic Solutions system (See figure A) these cracks appear as “spikes” that appear in absolute equal intervals. If the cylinder is rotating at the correct speed one should see 160 revolutions (or spikes) per minute. I established the speed by taking a 10 second section of the recording so as to marginalize any error for drift in the position of the crack and then counting these “spikes”. The current speed of the Smithsonian transfers show up as 24 revolutions per ten second segment, or 144 revolutions per minute; 10% slower than the correct speed.

Figure A

In a letter dated 12/16/96, Judith Gray acknowledged the transfer speed as incorrect, but she used a different method to arrive at her conclusions. She used Densmore’s notations of the songs themselves and then matched the pitches to a piano. While one reaches the same conclusion using this method (that the current tapes are too slow), the method of establishing correct cylinder rotations per minute and matching pitches to her pitch pipe tones seems preferable.

None of the methods used to establish a correct playback speed could be considered as accurate as the transfers overseen by Densmore herself. Her first-hand knowledge of the source materials and the methods she developed to insure correct reproduction of the recordings make her work the best reference. The reference disks (referred to by Judith Gray) made under Densmore’s direct supervision should be the key to establishing a playback speed closest to the original. One must suspect that the transfers made from the reference disks to tape under Vennum’s direction in 1981 were all transferred incorrectly; that is the reference disks were simply played back at the wrong speed. Meanwhile, as the years have passed, studies have been made and papers written based on these incorrect transfers. Native singers have struggled with the tapes in an effort to understand them and to learn the songs, some of which are now over two hundred years old. The situation not only begs correction, it calls into question the methods used to transfer and catalogue the approximately 2,300 other songs of the Densmore Collection. Most importantly, it stresses the importance of the need to involve Native peoples in the process of archival recordings and their restoration.

©1999 David R. Swenson
208 North 4th Street
Bismarck, ND 58501
701-223-7316
mac@makoche.com

photos courtesy of the State Historical Society of North Dakota

Return to the Thomas Jefferson Hour Page

Visit the Makoche Vision Page