The quote from "Surely You're Joking Mr. Feynman" explains it better. This Wikipedia article is pretty bad.
One mechanical engineer at Frankfort was always trying to
design things and could never get everything right. One
time he designed a box full of gears, one of which was a
big, eight-inch- diameter gear wheel that had six spokes.
The fella says excitedly "Well, boss, how is it? How is it?"
"Just fine!" the boss replies. "All you have to do is
specify a shaft passer on each of the spokes, so the gear
wheel can turn!" The guy had designed a shaft that went
right between the spokes!
The boss went on to tell us that there was such a thing as
a shaft passer (I thought he must have been joking). It was
invented by the Germans during the war to keep the British
minesweepers from catching the cables that held the German
mines floating under water at a certain depth. With these
shaft passers, the German cables could allow the British
cables to pass through as if they were going through a
revolving door. So it was possible to put shaft passers on
all the spokes, but the boss didn't mean that the machinists
should go to all that trouble; the guy should instead just
redesign it and put the shaft somewhere else.
So, a shaft passer really does exist, and really has been used. But for this example, of someone designing a machine that has one axle going through another wheel, the better solution is just to move that axle, rather than trying to actually build a wheel with a shaft passer on each spoke.
I'd like to see a reliable source, and ideally a photo/video of the device in action. I won't say it's impossible, but it seems like it would be hard to make sure the gear rotates smoothly instead of binding, and "Feynman says that a guy he worked with heard about it" isn't as convincing as it might be.
The shaft passer shown here has never been used in real life. The mechanism that's on each of the spokes of the cog has been used in real life to allow cables to pass each other (just imagine one of the spokes and the obstructive shaft are actually cables).
I considered it; I've done a good bit of Wikipedia editing in the past. The problem is, I have a hard time figuring out how to phrase it better without just including the Feynman quote. That's the only source, and it seems that the Wikipedia page is merely trying to re-phrase what Feynman said; there really isn't anything to the page other than Feynman's quote and the diagrams. As such, it's hard to do a good job rephrasing it rather than simply quoting Feynman.
If you can find other sources than Feynman on the shaft passer (and blog posts and forum discussion that are merely discussing Feynman don't count), I'll consider updating it with an improved description and reference to the other sources.
It looks like it should work, but the tolerances and tensions required make it unlikely to ever actually work. For the model in the article to work, the inner gears (in red) need to be A) Loose enough that the blue shaft can pass between them B) tight enough to pass power from the interior of the wheel to the exterior. This is improbable, and, while the solution seems sound, it's never been shown to work; It maintains mythical status.
Yeah, I think in practice, as drawn, the red wheels would simply jam up. And if they don't, the outer rim of the big wheel would not hold its place relative to the inner part. I dont see how the outer rim could be solid enough, relative to the inner hub, to do any useful job. The whole thing just breaks.
The mine application seems to be a subset version. If that is the best example any one has of this design in practice it suggests that its not possible to build a working version of the whole thing. A mechanically weak proof of concept might be possible, but a solid use-able version seems to me to be impossible.
Might work if the outer rim is supported or guided on some way. I still think loading is a problem though.
I don't know if tightness has anything to do with it. As long as the passer cannot escape its enclosure, it would pass power, because it is being dragged by the inner spoke. The real problem is the amount of force being placed on the passer and the probably impact of routinely passing the shaft through at high velocities.
This is a really good analysis – I thought the design was simple at first, but now even the cable-passing-cable version linked in the wikipedia page looks liable to jam as teeth pass in and out of the top channel.
For something like this to even have a chance, it would probably need many more, thinner teeth than these diagrams illustrate.
EDIT: @analogwintermut I think this addresses your question.
In that situation the cable needs to be strong along its length but allow passing through orthogonally. Whereas a gear would need to be both strong and weak on the same axis.
I agree the implementation isn't sound, but there's nothing wrong with the concept of a shaft passer.
As an example of why it is conceptually feasible, remember that you could computationally (or mechanically) determine when a shaft is closing in and open a gap in that particular support at the right time, maintaining ability to pass torque through all other supports.
It's certainly possible to build. It's a joke because it should never need to be built. There are so many simpler ways to route power around the wheel rather than through it. If it ever became apparent that you needed such a device, it would indicate some seriously poor choices early in the design process. The joke revolves around the wheel/shaft problem, not the cable/cable problem.