A large-scale survey of the process for submitting research papers to scientific journals has revealed a surprising pattern: manuscripts that were turned down by one journal and published in another received significantly more citations than those that were published by the first journal to receive them.