Populations of spaced plants of each of the four American upland cotton varieties Deltapine Smooth Leaf, Rex Smooth Leaf, Pope, and Rex and one African upland variety, Bar 7/8, were grown in the 1965–6 wet season at Kimberley Research Station, in the Ord River valley, Western Australia. The number of bolls, boll weight, lint percentage, seed cotton yield, lint yield, micronaire value, mature height, and morphological type were recorded for each plant.
For all varieties the range of each attribute was wide; for example, for a typicalvariety, boll weight ranged from 3·0 to 7·8 g, lint percentage from 30·5 to 43·5%, micronaire value from 3·3 to 5·9 units, and mature height from 78 to 195 cm.
The most variable varieties judged on coefficients of variation were those maintained at the station for the longest period without selection while the most recent accession was the least variable.
Frequency distributions for all attributes in each population departed significantly from normality: the majority showed both significant skewness and kurtosis. Positive skewness shown for the yield parameters seed-cotton yield, lint yield, boll number and boll weight in all varieties suggested there was scope for improvement by selection. Negative skewness for lint percentage was found in three varieties, suggesting previous intense selection for this trait. The coefficient of variation of lint percentage with one variety (Deltapine Smooth Leaf) was considerably less than that recorded 30 years earlier for its predecessor D x PL 11 A.
Covariance adjustments to seed cotton yield of individual plants, based on various hypotheses of the relationship between a plant's yield and that of its neighbours failed to give reductions of consequence in the variability of yield. It was concluded that inter-plant influences were far more complex than the models assumed.