Well, I have done some reading and it appears cylinders
were molded from 1901 on, but lost out to discs because of the shorter playing time of the cylinder (2 min vs 4) and the fact that discs were still easier to mold.
Up to that point, recording sessions consisted of as large a number of recording machines as the studio could afford, each recording one cylinder. There
was no duplication, so the musicians had to repeat the performance for a new batch of cylinders.
As a result, there was no such thing as two identical cylinder recordings from this era.
Even if it was the same performance of the tune, the position of the recording horn had to be slightly different for each cylinder. In the mid 1980s, in the earliest days of digital recording, there was a project afoot to find pairs of cylinders from the same session (many had inscribed codes with unique numbers for each session and each machine) so that the recordings could both be played back and time-aligned with the aid of digital processing and one sent to each channel for a form of stereo reproduction!
Here is an interesting site, with a number of cylinders for the hearing, including the first known recording, from 1878:
http://www.tinfoil.com/In that photo you posted, Buddy, of the recording session, notice the player nearest the camera - he is playing a recording violin (not sure if that is the correct term). I have seen one of these, and they are pretty amazing. There is no body, but the bridge the strings go over is on a diaphragm like the one on the old phonographs which is funneled to two horns - a larger one pointed at the recording horn and a shorter one pointing at the player's ears! This is all very clearly visible in the foreground.
