Riding a bicycle fast is largely about one's ability to produce power to overcome aerodynamic drag and gravity. Power is typically measured in terms of watts, and cyclists serious about improving their power have recently been turning to on-bike wattage meters to guide their training, gauge their improvement and, occasionally, to pace their races.
There are four commonly available on-bike wattage meters on the market: the SRM crank system, the Power Tap rear hub, the Polar S710 heart rate monitor with optional power unit which measures power via chain speed and tension, and the Ergomo bottom bracket. In addition, there are a few indoor power-measuring trainers onto which one can mount a bike (such as products by Computrainer, Tacx, and Velodyne), though the main focus here is on-bike power meters.
I've had the chance to analyze data collected from each of the on-bicycle power meters. Several (but certainly not all) of these data were provided by members of Topica's Wattage List (see link at bottom of page). Included here are some data files collected during races; some that reveal oddities and anomalies in the power meters; and a few files for rides on which more than one power meter was mounted (including the first publicly-available "Rosetta Stone" file where three on-bike systems were mounted on one bike).
Anyone who is thinking about buying a power meter wants to know how accurate they are. Here are some examinations of data collected during the same rides. Note that these are not reviews of power meters from the perspective of features, usability, reliability, ease of installation, long-term cost, or how cool they look while on your bike: those are available elsewhere. These are analyses of the quality of data they produce. There are very few published analyses of such paired data.
If you've ever wondered about the power that riders produce during the course of race, here are a series of data files collected during road races, time trials, and hillclimbs, with analyses of some of those races. Included here is a pointer to Dede Demet's Montreal World Cup winning ride, Bjarne Riis' 1997 Amstel Gold World Cup winning ride, and some races ridden by Bike.com columnist Adam Hodges Myerson. In contrast to the previous section, which included files collected from two or more devices by the same rider, this section includes data files collected by two different riders in the same race, and two races by the same rider on the same course (though one year apart). I know of no other place where such contrasts and comparisons are available. All of these data files have been converted to comma-separated-value ASCII data which can be read by most spreadsheets.