VIDEODISCS FROM THE 1920s

UNTIL RECENTLY, REsearchers into preWorld War II television had only contemporary descriptions and blurry still photos of glowing screens to rely on. Now, however, a Scotsman named Donald F. McLean has managed to extract moving images from television signals that were recorded onto shellac phonograph disks as early as 1927—when most people were still getting used to radio.

The existence of the disks has been known for decades among television historians and engineers. Whenever someone tried to play one, however, the result was unrecognizably distorted. Then McLean, who originally trained as an electrical engineer, analyzed the disks’ content with a computer, which allowed him to store and analyze the signal. The result, which he describes in his recently published book Restoring Baird’s Image (London: Institution of Electrical Engineers), provides a unique and thoroughly unexpected glimpse at how television looked in its paleolithic era.

The disks McLean has restored are of two types. The first set was recorded between September 1927 and March 1928 by John Logic Baird, the Scottish-born television pioneer, who was experimenting at his London studios with what he called Phonovision. Earlier in the decade, Baird had developed a system that divided images into 30 vertical bands (known as lines) and updated them 12Vz times per second. (Modern American television, by contrast, has 525 horizontal lines and 30 frames per second.) Images were scanned mechanically with a Nipkow disc, which had a spiral of lenses around the outer edge of a large spinning wheel. The resultant television signal was broadcast over ordinary radio frequencies.

After Baird had worked out the bugs and formed a company, he investigated the possibilities of recording his television signals. He and his staff made a number of silent test discs, which showed such things as a hand moving in front of a puppet head and a woman talking and laughing. Recorded at a rate of about four frames per second, each disc held about three minutes’ worth of images.

McLean’s second set of disks consists of amateur home recordings, made on aluminum discs, of BBC Television Service broadcasts from the early 1930s. While retaining the 30-line standard, these broadcasts did not use Baird’s mechanical system but instead scanned electronically, as in modern television.

McLean’s book gives extensive detail on how he reconstructed the disks’ content. First he played them on conventional phonographic apparatus and recorded the signal, which he digitized. Next he had to decide where each 30-line frame began and ended, a tricky matter because the signals lack any sort of timing information. He figured that out by laboriously comparing the contents of each frame with the ones before and after and repeatedly adjusting his work to get as close a match as possible. This method was far from foolproof because the contents of a frame can change considerably in a quarter of a second, especially with only 30 lines to go on.

Having aligned the frames, McLean used clever algorithms to strip out extraneous noise (caused by resonances between different parts of the apparatus, for example), restore missing information resulting from degradation of the recording medium, and otherwise clean up the images to give some idea of what a 1920s or 1930s viewer would have seen. To make things even harder, he had to write all the necessary programs himself. Off-theshelf signal-processing software was useless, since something that has the electronic profile of a click or pop on a sound recording may be a legitimate part of a video signal.

Along the way, McLean dealt with such arcana as stylus profiles (an incorrectly designed stylus would distort the playback) and equalization characteristics (instead of the now-universal RIAA standard, which boosts certain frequencies in audio recordings, the recordings use the older Blumlein characteristic, which treats all frequencies the same). Another essential concern was the seemingly simple matter of centering the disk on the turntable: With precise time relations so vital, a small deviation that yields a minor “wow” on audio can virtually destroy a video image. He also learned the importance of using a tangential tracking arm, in which the stylus travels along an exact radius, instead of the pivoting arm found on most audio turntables.

To examine extracts of the videos, visit McLean’s Web site, www.dfm.dircon.co.uk. His book can be purchased through the Institution of Electrical Engineers at www.iee.org.uk.