Disc Playback Speed
Original Encyclopedic Discography of Victor Recordings (EDVR) data included playing speed estimates for hundreds of Victor classical vocal recordings, the collecting emphasis of EDVR co-founder William R. Moran. Moran personally determined what he believed to be the proper playing speeds for these recordings and added them to EDVR entries. These playing speeds are retained, unedited, in the EDVR database—now the Discography of American Historical Recordings database. Moran’s writing on Victor playing speeds, from the introduction to Vol. 2 of the print Encyclopedic Discography of Victor Recordings, is reproduced below.
The question of correct playing speeds for "78 rpm" recordings has been a vexatious one for many years. No written record appears to have been maintained in the early recording studios of the turntable speeds when a recording was made. A deviation in speed of 4 rpm (revolutions per minute) changes the pitch (and thus the key) by one half tone. Not only is the pitch incorrect when a record is played at a speed at which it was not recorded, but even more important, the tone quality of the sound is distorted. Thus it is extremely important to play these old recordings at a turntable speed as close as possible to that used when the original was made. Obviously this is more important when dealing with vocal records than instrumental. This speed problem was recognized from the earliest days. A few of the early recordings had a separate track whereon the note of a pitch pipe was recorded. What a shame this practice was not continued! The 1904 Melba recordings carried a note on the labels as issued both in Europe and America of the key in which the selection was sung. Clara Butt must have had it stated in her contract with the Gramophone Company that not only were the keys to be printed on the labels of all her recordings, but in every catalog where her recordings were to be listed, the following statement was to be printed: "It is important that these records be played in the keys indicated." Emilio de Gogorza, writing in 1937 about recording Marcella Sembrich for Victor in 1904, notes that the pitch could be varied on playback:
I can visualize Sembrich’s look of astonishment when she heard her first recordings, and I can hear her rapid exchanges with William Stengel, her husband, in Polish. I devined the gist of the discussion and promised her that it could be remedied by playing the records at a lower revolution. It gave her a turn to have La Sembrich being reproduced in a different key. Most singers wouldn’t have known the difference, or cared, but the peerless "Traviata" and "Rosina," though she would stand for much without flinching, would not sing in an uncertain pitch. On this occasion, she was emphatic in her displeasure.
Yet Sembrich’s recordings never carried a notation of the key or speed on their labels except for the Lucia Sextette in which she sang. This stated for years on its label that it should be played at 82 rpm, yet this record plays in pitch at 78.26! The first recordings by Emma Eames carried the keys on their labels, but this practice was dropped on later pressings.
While the problem was well understood by the companies involved, the great mystery is why a standard speed was not established by the two major and associated recording companies, and why incorrect information on the matter of speeds was served up to the record-buying public on both sides of the Atlantic. By February 1912, the Gramophone Company general catalog gave the speed of every recording, and there was a notation at the bottom of each and every page, "The figures in brackets at the end of the selection indicate the speed at which the records should be played," but one wonders where they obtained the figures they printed in their catalogs. Seldom is a correct speed given: as an example, the customer is told to play certain Caruso recordings at 81 rpm, when they are in score pitch at 75! In the Americas, where these identical recordings were offered by Victor, the May 1912 Victor catalog carried the statement:
All records should be played at a speed of 78. Every record is recorded at this speed and requires a speed of seventy-eight to reproduce it properly. Set the regulator so that the turntable of your Victor revolves at 78 times per minute and never change it…. You will of course meet the man who insists on turning the regulator of his Victor up and down, thus changing the speed of each record he plays. Don’t imitate him—He is wrong. Only at a speed of 78 can you hear the actual tones of the singer just as they were recorded.
Victor carried some variation of this statement in every catalog. In 1913 they added: "Never change it unless for some special purposes as when using the records for dancing to suit the exact taste of the dancers." In November 1915 they added: "taste of the dancers (or when the maid has dusted your Victor in the morning)." In November 1916 they added: "When using the Victrola and piano together it is occasionally necessary to change the speed to get the two instruments exactly in tune and sometimes the dancers wish a little variation to suit the steps. (And the maid sometimes insists on polishing the regulator screw!)" In May 1917, we have the same old paragraph except that we were told to play all Victor records at 76.00! This is not merely a typographical error: it is used twice in the paragraph heading, spelled out in the text, and again the paragraph concludes: "Only at a speed of seventy-six can you hear the actual tones of the singer or player just as they were recorded." The July 1917 issue of "the Voice of the Victor" ran a correction, saying this statement was in error, and in November 1917 we go back to the old words, but again all records must be played at 78. It stayed at 78 (although as a sign of the times we lose the reference to the maid) through 1925. Even after the introduction of electrical recordings (which initially appeared to be standardized at 75 rpm) the public is told to play all Victor records at 78, although the language is not so strong. Why this insistence that the public play all recordings at speeds which the company knew were not consistent?
Attempts have been made to group speeds used by Victor into certain years, and tables have been published to guide the record collector. A glance through the speeds given in this discography will confirm that from time to time certain brief trends seem to confirm some orderly approach, but this always breaks down. Recordings made in New York the same day as other records made in Camden are often different speeds, and it even appears that different recording machines operating in Camden at the same time ran at different speeds. There is concrete evidence that recording technicians were not always as careful as they might have been in attempting to adjust their machines. Some specific recording sessions were disastrous! Five records made by soprano Roxy King on 4 and 5 May 1908, play at 72.73 rpm, an extremely low speed for Victor at this time. The effect of playing these at 78 can be imagined. Is there any corollary between this obvious error in recording speeds and the fact that today these records are among the rarest ever made by Victor? The same machine must have been used for the sessions of 18 and 19 May 1908, when the artists were Emma Eames and Emilio de Gogorza. A total of ten sides were recorded; of these only 5 were released, and all play at 72.73 rpm. All are exceedingly rare in their original form, as they were quickly withdrawn from circulation. One of the unpublished recordings was issued many years later by one of the collectors clubs: there is nothing wrong with that recording except the speed, so one is led to believe the other unpublished records from this session were destroyed only because of their speed. Could the fact that the Abott-Ancona duets made on 6 January 1908 were recorded at the unusually fast speed of 81.82 rpm account for their short catalog life and consequent rarity today? Ancona’s aria from Faust, recorded the same day and at the same speed, never received commercial publication, perhaps for the same reason. An unpublished Mardones recording, made in 1924 has been found that plays (wonderfully!) at 74.23 rpm—there could be only this reason for its non-publication.
Almost invariably records made at the same recording session by the same artist are recorded at the same speed (a fortunate circumstance for the bedeviled researcher). An occasional variance to this rule can be observed when an artist recorded with both orchestra and piano on the same day. A reasonable explanation would seem to be that a different studio, and thus a different recording machine, was used. Those interested in working with pitch problems will find the chronological tables of recording sessions most useful, provided the issued takes made on any given day are confirmed in the matrix section.
Since there is no historical record of the recording speeds, the user today is thrown on his or her own initiative and must resort to experimentation with the recordings themselves to determine a satisfactory play-back speed. Over the years I have developed certain techniques for “pitching” old recordings. These have been discussed elsewhere, and there is no need to detail them here. Suffice it to say that all the suggested speeds given in this discography have been determined with considerable care with the use of a reed organ (trumpet stop A’ = 440 Hz.) and a multispeed stroboscope. A variation in pitch caused by a change of one rpm can be detected by the trained ear; the fact that the playing speeds are given to two decimal points is not intended to suggest an accuracy to that degree. Rather, the figures given correspond to the bands so named on a stroboscope designed for observation under a 60 Hz. lamp, or they are an arithmetical average of speeds falling between two named bands that appear to be revolving in opposite directions at the same rate. If the listener, using whatever device, will attempt to adjust the turntable speed as closely as possible to the turntable speeds suggested, the records should be reproduced at very close to the original recording speeds. Since each and every record must be individually pitched I have reported only on recordings in my own collection, and I apologize for the fact that speeds have not been given for all listings. The opportunity to make contributions of playing speeds in various fields of expertise (such as band records) was publicly offered, but no volunteers came forward.
The determination of the speed at which a recording machine turntable was revolving on a particular day perhaps eighty years ago is far from an exact science. Determinations set forth in these pages are indeed subjective, and the reader will note that they are cautiously labeled "suggested playing speeds." They have been arrived at after much diligent effort and should be accepted in the spirit in which they are offered!