Methodology and Definitions

Period of Study

The study is based on a random sample of 1,500 recordings commercially released in the United States between 1890 and 1964. We chose 1890 as a starting point because that year approximates the beginning of the commercial recording industry in the United States (Brooks 1978). It is the earliest period from which reissuable commercial recordings survive and the earliest year from which recordings are still under the exclusive control of a present-day rights holder (i.e., the first full year of recording by a predecessor company of a rights holder that is still in existence).1

The end year of 1964 was based on three factors.

  1. Scope. The study covers the first 75 years of commercial recording in the United States.
  2. Industry changes. A cutoff of 1964 makes it possible to include the cylinder era, the 78-rpm era, and the first decade of widespread acceptance of 45-rpm and LP formats. All these are formats now challenged by the lack of generally accessible reproduction capability.
  3. Feasibility. Because of the explosion in the number of recordings issued in recent years, as well as of the proliferation of reissues of those recordings, the project would be much more difficult to execute for more-recent periods. As will be seen, rights-holder reissues are in any event more frequent in more-recent periods.

The 1890–1964 time span was then broken into 15 five-year blocks, with a quota of approximately 100 recordings drawn per block. This permitted a more granular analysis of changes over time than decade-long blocks would allow, yet kept the size of the sample needed manageable (a minimum sample of 100 is generally considered necessary for statistical analyses). Five-year periods also allowed us to map changes coinciding with major changes in the industry that began mid-decade, e.g., the shutting down of independent producers by the patent-holding major companies around 1905–1908, the introduction of electrical recording in 1925, the post–World War II record boom that began around 1945–1946, and the inception of the rock-and-roll/microgroove era in 1955.

Scope of Study

We initially considered basing our analysis on a random sample of all recordings released in the United States during the 75-year period in question. However, most of such a sample would have consisted of recordings that are of little interest to scholars, students, or the general public. Thus, we decided to restrict the study to recordings for which there is documented historic interest. We drew the study sample from approximately 20 modern discographical sources, representing seven major fields of study:

  1. ragtime and jazz
  2. blues and gospel music
  3. country and folk music
  4. music of U.S. ethnic groups
  5. popular, rock, rhythm and blues (R&B) music
  6. classical music
  7. other (including spoken-word recordings and show music)

The discographies chosen list more than 400,000 recordings from the period 1890–1964. These are recordings in which modern scholars, students, and collectors have shown special interest, as documented by the widespread use of the source publications. Thus, the sample used for this study is not of recordings in general, but rather of recordings in which there is documented interest. Indeed, many of them could be considered “historic.” This is a sample of the recordings most in need of continued availability today.

In addition, we drew 10 pre-1965 selections from the National Recording Registry (NRR) list for 2002 and an equal number of such selections from the 2003 list.

We used the following criteria in choosing the discographic sources for this study (see Appendix A):

  • Each is an acknowledged standard reference in its field.
  • Each is a genre discography covering all labels relevant to its musical field, as opposed to a discography of specific labels or artists. Label and artist discographies would have skewed the sample toward specific labels, and the protected or nonprotected status they represent.
  • Each covers some part of the period 1890–1964. In most cases, no single discography covered the entire period; consequently, more than one discography was required to cover the entire time span.
  • The discographies are nonduplicative to the extent possible. This required some difficult choices: It meant, for example, that Brian Rust’s well-known Jazz Records (1897–1942) was not used because it is a subset of the much larger The Jazz Discography (1896–2001).

Statistical Methodology

Once we had chosen the source discographies, we established quotas for each musical genre within each period. Not all genres of music were recorded in every period (for example, the first country records date from the early 1920s), so we divided the quota of approximately 100 recordings for a period equally among the genres that were represented in that period. We gave each genre equal weight. If five genres were recorded in a period, we allocated each genre a quota of 20 recordings; if all seven genres were represented, each genre was allocated 14. See Appendix B for a table of specific quotas by genre.

We used a random-sampling methodology to choose specific recordings within each genre. We drew a random number and used it to point to a specific page in a discography. We then chose the first recording on that page that met our definition of commercial recording (see box, below). If we found no qualifying recording on that page, we examined subsequent pages until a qualifying recording was located. The goal, in accordance with sampling theory, was to ensure that each qualifying recording in the discography had an equal and known chance of being chosen.

We conducted two rounds of sampling. The purpose of the first round was to estimate the proportion of all recordings in a period that is protected. Once 1,500 recordings had been chosen, we researched their status in order to develop an estimate of the proportion of all recordings in each period that is protected and nonprotected. In the second round, we used the same methodology to draw a sample of 1,500 protected recordings. (Protected recordings already identified in the first round were used toward the quota.) These were researched to determine the proportion of protected recordings that is currently available in reissue, and the sources of those reissues.

DEFINITIONS

Brief definitions follow for terms used in this report. More-detailed definitions are provided in Appendix C.

Commercial recording. A single recording of a selection or selections by an artist, issued for sale in the United States to the general public during the period specified.

Protected/nonprotected status. Whether or not a recording is currently protected (i.e., controlled by a rights holder) is in the judgment of the compilers. The approach was to replicate the determination that a reasonable person would make, after a reasonable amount of diligent research, if that person, or his or her institution or association, wished to reissue the recording legally. We used three tests to determine whether a recording is probably protected: corporate lineage, marketplace evidence, and consultation with experts.

  1. Corporate lineage. Is the entity that originally issued the recording, or a known legal successor, still in existence today?
  2. Marketplace evidence. Has a person or company asserted ownership of the recording in the years since the recording was made, either through legal claims or “authorized” reissues?
  3. Consultation with experts. The project director and contractor for this report are both recording industry historians and could trace the ownership of most recordings; in some especially difficult cases a number of experts with years of experience in the field of reissues were also consulted.

Reissue availability. Reasonable availability of a new copy to an ordinary person, through normal commercial channels.


FOOTNOTE

1The original recording company was the Columbia Phonograph Company of Washington, D.C. The successor company, and present rights holder, is Sony BMG.