Abstract
We review the methods used to study seabirds at sea from ships, discuss the problems posed in making reliable observations in relation to the design of research programs, and describe a method currently in use around the seas of Great Britain. We suggest a framework for future studies, incorporating features likely to stabilize bias. The key items in the recommendation are (1) the use of a brand transect in order to provide density estimates, and (2) a method to correct for movement of flying birds in the band transect in order to minimize bias caused by such movement.