(1) Cloud seeding, either experimental or operational.
(2) Experiments with the injection of aluminum nanoparticles into the upper atmosphere to increase the Earth's albedo as a potential geoengineering solution to climate change. There's some documentation for this if you Google it, and it might look pretty funky to a ground observer.
(3) Other experiments with spraying nanoparticles, such as for radar reflection or to attempt to simulate difficult radar conditions to test stealth-busting radars.
Folklore and wild rumor often starts around a seed crystal of truth. Note that many of the early chemtrails rumors came from the American Southwest, which is where the black project kids play with their toys.
Yeah, I'm perfectly aware that most "chemtrails" are just contrails, but I have a feeling now that it might have started around a quite solid kernel of truth.
You know, I used to post this picture as a joke: http://i.imgur.com/MM9YsTM.jpg. Now I realize it's exactly what the US military did in San Francisco.
Medical experiments? Sure, those were "harmless" bacteria, but maybe a few people died as a result of the test.
But would you believe that indigent cancer patients were given lethal whole-body x-ray exposures, and then observed (without treatment) while they died?
Obviously you can trot out edge cases. The claim that extraterrestrials run the US government is quite extraordinary. But edge cases aren't that interesting. Most claims fall somewhere in the middle, and in that case can be called extraordinary or not based on the personal biases of the claimant. Thus "extraordinary claims require extraordinary evidence" becomes a goal-post-moving fallacy in practice.
Personally I don't consider claims about governments wanting to surveil people, economic systems tending to concentrate wealth, or the dystopian potential of supra-national corporate governance to be particularly extraordinary. YMMV.