Plant populations of American pima cotton (Gossypium barbadenseL.) and upland cotton (G. hirsutumL.) giving maximum lint yields vary widely in reports in the literature. This study attempts to explain the wide variations in plant populations giving maximum lint yield for pima cotton by considering the plant population X plant height interactions with lint yield. Field tests were conducted for 5 yr at four locations on Anthropic Torrifluvents, Torrifluventic Haplostolls, and Typic Torrifluvents soils. Fifteen pima cotton tests with varied plant populations, and with 1.02‐m row spacing, had plant population, plant height, and lint yield measured on each plot. The relationships among yield, population, and height were analyzed by multiple linear regression in each test. In three of the 15 individual test analyses all regression components were significant. A combined analysis of the 15 tests showed a significant negative (P=0.001) plant population × plant height interaction for lint yield. Most of the plant population and plant height treatment means in these tests were within the range of 20 000 to 200 000 plants ha −1and 60 to 190 cm high. Plant population giving maximum yield decreased about 11 000 plants ha −1for each 10‐cm increase in plant height. Cultivars and irrigation treatments did not greatly alter the interaction. Year and geographical location effects were more varied. The variation in plant population producing maximum lint yield for different plant heights was great enough to explain the varied plant populations for maximum lint yield reported in the literature.

Plant populations giving the highest lint yield of American pima cotton (Gossypium burbdense L.) grown in 1‐m spaced rows change as plant heights change. Therefore, plant height should be considered in future plant population studies with pima cotton. This study was conducted to determine the number and type of plant height measurements needed to give a satisfactory estimate of mean plant height for a plot. Plant height uniformity data were obtained from 12 plots (four locations with three plots per location) in Arizona. Soils were Torrifluvents Haplostolls, Anthropic Torrifluvents, and Thermic Torrifluvents. Measurement of four to eight plants per plot, half of the number on each end, appeared to be adequate to characterize the plant height of a plot. Plant heights were normally distributd. A systematic selection produced lower variation (about onehalf as many plants needed to be measured) but over‐estimated mean plant height by an average of 12.5%.