and third methods described on pp. 594 and 595 but, of course, averages fromr
many more would be necessary to indicate this correctly.* The observations
are grouped, those in each group having been made ...