In experimental design, especially in evaluating the effectiveness of drugs or other medical interventions, there is the possibility that beneficial effects accrue because the patient psychologically expects to feel better, which will interfere with identifying any actual benefits of the treatment. In a single-blind experiment the patient does not know which treatment they have been assigned, and in a double-blind experiment neither the patient nor the doctors or others dealing with the patient know which treatment they have been assigned.

Where the nature of the treatments means they cannot be interchanged without the subject being aware, a double-dummy may be used where patients are apparently given both treatments, but one or other will be a placebo, and the control group will be given both as placebos.

http://www.bmj.com/cgi/content/full/321/7259/504 A full discussion of the ethics and principles of blinding in clinical trials (subscription).