I never would have figured it out by myself, except that I knew I liked some and not others. But there is an excellent book by Tony Oliver (even though I don't agree with everything he says, especially about tandem frame configurations), called Touring Bikes.
He writes (quoting from memory): The tandem should not be affected by leaning. Thus, stoker movement won't affect it. This usually is achieved by a steep head angle (74 degrees or so) with longish fork offset, giving minimal trail. All the great tandems of old I have measured (see below) are built along those lines. According to Oliver, this wouldn't work on a single, but a tandem has enough mass to be fine. (Of course, many older Herse and other French bikes used a similar, very small trail geometry, but with shallower head angles and even more rake.)
Oliver reports how he built his first tandem with a nice touring bike geometry, and every time the stoker scratched their leg, it went for the ditch. He then replaced the fork with one with more rake, and now it behaves fine.
Modern tandems often adopt a similar single bike geometry, with a shallower head angle and more trail. When they come toward you, you can see how they usually move in slight zigzags because the pedaling forces affect the steering of the tandem. This requires constant corrections from the captain, which are quite tiring (on the Cannondale, my arms and shoulders hurt after only 100 miles), and like any turn, they must consume energy. When you get off one of these tandems and immediately onto a single (I used to ride friends' tandems, so I'd ride a single to and from their house), you weave from side to side, because you have become accustomed to countering the stoker's input. I used to think this was inherent in tandems, until I got my first Taylor. I get off that one, and immediately ride the single as if I never had been on a different machine.
Standing up on the pedals also is a lot easier if the bike doesn't steer due to leaning.
That is all I know. I don't know much about bike design (thus the quote from Oliver) - I just ride the things, and I know very well what works and what doesn't. With a French friend, we measured every tandem we could find, including four Herse, two Singers, an Hugonnier-Routens and a few others, and found a consistent thread. We wrote an article for the next VBQ on this, with drawings of the tandems' geometries and measurements. The exception are tandems so small that the "desired" geometry would produce toeclip overlap. Those have a shallower head angle and more trail. My friend, Hervé, reports that those tandems are harder to ride, and much harder to ride standing up.
How they manage to make the tandem turn with ease, yet resist input from the stoker, may be the true skill in designing the geometry. I know some people believe a machine (single or tandem) either will have good straight-line stability or excellent cornering, but not both. This is not true: All of the ones I like corner better than single bikes. We rarely drop people on the straight downhills, because they hang on for life in the slipstream, but come the twisty bits, they are far, far behind.
That said, some people train their stokers to sit perfectly still on the tandem, which should solve the stoker input problem. I once rode with Pamela Blalock from Boston, and I couldn't even tell whether she had got off the bike after we stopped. But this training isn't necessary on the older ones, and I ride with many different stokers, so it's enough effort trying to get in sync, rather than telling them to sit still and never move. In fact, I prefer stokers who concentrate on putting out power, rather than sitting still!
Maybe this thread should move to the tandem list? -- Jan Heine, Seattle Editor/Publisher Vintage Bicycle Quarterly http://www.mindspring.com/~heine/bikesite/bikesite/
Subject: Re: [CR]was early lightweight tandems..probably still is..
Jan apparently wrote:
"And about Taylor: I like them because they got the geometry right: The tandem isn't influenced by stoker movement."
Since you feel that Rodriguez, Mercian, Cannondale, Trek, and Co-Motion all have poor handling, what is it specifically in the Taylors front end geometry that is "right" and which the other makers you mention have got "wrong".
Hugh Enox La Honda (in the middle of a rain storm today)