Since the number of patients to be sampled in each hospital was known in advance, it was theoretically a simple matter to select these from the hospital register. In practice there were many difficulties, in view of the diversity of methods of record keeping employed in the institutions. Some main units had a list of the names of all patients in all ancillary units, in others each unit kept its own list and there was no central register. Some hospitals kept the information in alphabetical order, others by date of admission. Sometimes male and female patients were kept in separate registers, sometimes jointly. By no means were all registers up to date, and discharges and deaths had not been recorded (possibly admissions too, but we had no way of checking this). The facility with which it was possible to locate the relevant ward for each patient also varied greatly; in some hospitals a good deal of time was wasted in this kind of administrative congestion.