Statistics on ornithological (bird) data collected from Heard Island in 1950-1951

The download file contains a scanned copy of correspondence, the report, and statistics from data collected on birds from Heard Island in approximately 1951 (the report is dated May 1951, so data collection must have occurred before this). The original documents were not in the best condition, so unfortunately some of the scans are difficult to read. Taken from part of the report: For Heard Island birds, measurements on females, fresh and dry, and males, fresh and dry, were generally available. The data were therefore submitted to an analysis of variance (2x2 non-orthagonal treatment, interaction non-significant) to test for differences between the classifications. When the difference between members of a classification was found to be non-significant, the distinction was disregarded and new means and variances were worked out, combining the results from both sub-classifications. For example, if condition (wet or dry) was found to be non-significant, means for male and female were computed as if no sub-division into fresh and dry existed, and a pooled common variance was obtained from the variances within sex classifications. Similarly, if neither male-female nor fresh-dry differences were significant, a grand mean for the island was computed, and the variance of teh whole set of data was taken to apply. This mean has been quoted as appropriate to all four categories, the coefficient of variation and the standard error of the mean being similarly quoted as the same for each sub-classification. Note has been made of the number of observations, (n), on which the mean is based, and the number of degrees of freedom, (v), appropriate to the estimate of the standard deviation. One further point may be mentioned here. Frequently, the dimensions of individual birds have been given without classification into male and female. It has been possible to utilise these measurements in the general analysis only when the sex-difference was not significant.