Evidence: You can start with the very basic empirical fact that, as previously mentioned, a time trial bike is both heavier and faster than a standard diamond-frame bike. An über-light racing road bike can be 12-14 pounds, and an über-light TT is more like 16 or 17 and will leave the 12 lb bike in the dust. Similarly, as I previously mentioned, aero and disc wheels are much heavier than standard wheels but are faster (unless there's a stiff cross wind). Aerodynamics is far more important factor than weight.
You can continue by checking out the world's fastest HPV, the
Varna Diablo II. Max recorded speed on the flats: 82.3mph. Weight: 60 lbs.
And keep going by playing around with the Kreuzotter calculator and examining their equations:
http://www.kreuzotter.de/english/espeed.htm
I have heard of a study where a researcher plunked a 5lb weight in a downtube, and a group of pro racers couldn't distinguish between a loaded and unloaded bike. I'll see if I can find the reference.
If my own experiences are of any use to you: My Dahon Mariner weighs about the same as my steel road bike (26-28 lbs) but is substantially slower. This is almost certainly due to a very upright position that, on that particular Dahon model, cannot be adjusted. My Swift was also 4-6 lbs lighter than my road bike, but performance was too close to reliably say which is faster.
Theory, as best I understand it: Weight affects acceleration, and that's it. Once in motion, the primary force slowing down the bike is friction, specifically aerodynamic drag, which increases at roughly the square of velocity. Once you go above 12mph or so, even tiny increases in your drag coefficient will slow you down much more than adding weight. Above 18mph, you need huge increases in power output to gain tiny increases in speed.
Ergo, any recumbent rider who can't beat a diamond frame on the flats should be blaming the engine, not the bike.
When climbing, you're fighting gravity; most (non-pro) riders won't ride fast enough up a hill for drag to be a major factor. Even here, though, weight is a smaller factor than you think;
Chris Carmichael, Lance Armstrong's coach, describes the impact of adding 2.5kg (~5 lbs) as adding 38 seconds off of a 20-minute long, 5km, 1200 foot climb -- huge for a pro, noticeable for an amateur racer, inconsequential for everyone else.
Subjective rider perceptions about weight are basically BS and highly inaccurate; the human body simply isn't that well calibrated, and has absolutely no reason to develop that ability, and cognitively it would be a very "expensive" one to develop. Perception of performance isn't much better, especially when a new or expensive bike is involved.
From what I've gathered, the only truly objective way to determine the efficiency / performance of a bike is with a power meter. Even an HRM might not do it since your biological efficiency will vary from day to day, so you'd need a large pool of samples.
Anyone willing to kit out a folding bike and a similarly equipped 700c bike with PowerTaps?
