The direct sale of emissions allowances by auction is an emerging characteristic of cap-and-trade programs. This study is motivated by the observation that all of the major implementations of cap-and-trade regulations for the control of air pollution have started with a generous allocation of allowances relative to recent emissions history, a situation we refer to as a "loose cap." Typically more stringent reductions are achieved in subsequent years of a program. We use an experimental setting to investigate the effects of a loose cap environment on a variety of auction types. We find all auction formats studied are efficient in allocating emissions allowances, but auction revenues tend to be lower relative to competitive benchmarks when the cap is loose. Regardless of whether the cap is tight or loose, the different auction formats tend to yield comparable revenues toward the end of a series of auctions. However, aggressive bidding behavior in initial discriminatory auctions yields higher revenues than in other auction formats, a difference that disappears as bidders learn to adjust their bids closer to the cutoff that separates winning and losing bids.