The ever-alert Gerard Van der Leun turned up this stirring account of something that seems impossible: the recreation of the sound from a recording from a picture of the recording.
Actually, it doesn’t seem impossible to me. While a record (or cylinder) groove is clearly a three-dimensional object, and a photo contains only two, vertical modulation in the early days was mostly unintentional: the cutting head wiggles from side to side, and if you can duplicate that wiggle, you can duplicate the sound. It’s more difficult today, what with stereo and all: the Westrex 45/45 system to cut a two-channel groove requires that the stylus track both horizontally and vertically. (This is, incidentally, why you weren’t supposed to play stereo records on your old mono clunker phonograph: the stylus didn’t have enough vertical compliance and could damage the groove.) For a brief period in the 1970s, there were even four-channel grooves, either with complicated phase-encoding schemes (SQ, mostly) or multiplexing patterned after FM stereo (CD-4). God forbid you should try to read one of those from a photo.
Although Arthur Lintgen, now 70, might be able to give it the old college try. From The New York Times, 11/19/81:
Before an audience in the auditorium of Abington Hospital, near Philadelphia, two weeks ago, Stimson Carrow, professor of music theory at Temple University, handed Dr. Lintgen a succession of 20 long-playing records chosen by Mr. Carrow and 10 of his graduate students. All identifying labels and matrix numbers were covered over, but Dr. Lintgen, simply by taking the records in his hands and examining their groove patterns in a normal light, identified the piece and the composer in 20 cases out of 20.
The event was arranged and filmed by the ABC-TV program That’s Incredible, which plans to air the segment early next month. Mr. Carrow had “never heard of Dr. Lintgen” before ABC called and asked him to administer the test. “We chose mainstream music the Beethoven Fourth and Fifth Symphonies, ‘Also Sprach Zarathustra’ by Strauss, the Tchaikovsky ‘Nutcracker’ things the audience could relate to. Not only could he do it, he could recognize some of them 15 feet across the room.”
This was astonishing enough that Time sent James “The Amazing” Randi to check it out. Randi constructed an experiment, and was apparently satisfied with Lintgen’s explanation. He wasn’t actually reading the groove so much as he was engaging in a complex form of pattern-matching:
The trick, Lintgen explained, is to examine the physical construction of the recording and look at the relative playing time of each one of the movements or separations on the recording.
According to Lintgen, a Beethoven symphony will have a slightly longer first movement relative to its second movement, while Mozart and Schubert would compose in such a fashion that each movement in many cases would have the same number of bars. Beethoven, however, had set out in a new direction and that changed the dynamics of the recording. In addition, if there was a sonorous slow beginning, one could look at the recording at that point and see a long undulating groove that would not contain the sharp spikes that would identify sharp percussion.
On an impulse, I went to the record shelf and pulled out Also Sprach Zarathustra. The opening “Sunrise” movement is pretty distinctive-looking, what with that long contrabass/organ/bassoon drone, followed by brass and then tympani. I couldn’t see anything in the later movements, though Lintgen probably could:
All phonograph grooves vary minutely in their spacing and contour, depending on the dynamics and frequency of the music on them. Lintgen says that grooves containing soft passages look black or dark gray. As the music gets louder or more complicated, the grooves turn silvery. Percussive accents are marked by tiny “jagged tooth marks.” The doctor correlates what he sees with what he knows about music, matching the patterns of the grooves with compositional forms. In a way, it is like reading a graph of a given work’s structure.
He probably wouldn’t have been able to read that 1890 gramophone record you saw in the video, but he might have been able to determine whether it was music or speech. Which is still pretty darn remarkable.
(Title from the original 1982 Time article. All the Lintgen links go to Snopes, just because.)