Cotton square production was determined during 1988-90 on three cotton cultivars [Deltapine 90 (DPL), McNair 220 (McN) and Stoneville 506 (STV)L subjected to four levels of square removal (0, 5, 10 and 20%). Squares were counted weekly and a designated percent removed each of six weeks during the beginning of the fruiting period. During 1988, significantly more squares were counted on DPL on weeks five and six for all four treatments than on McN and STV. The total number of squares counted during the six weeks was higher on DPL compared to McN and STV on treatment one only. During week four, significantly more squares were counted on DPL and McN than on STV in treatment two. Significantly more squares were counted on STV than on DPL or McN in treatment one on week two.During 1989, no significant square count differences were noted among cultivars within treatments. During 1990, significantly more squares were counted on DPL and McN than on STV on week 5 in treatment two. In treatment three, more squares were counted on DPL and STV than on McN in week three.

In the analysis-of-variance model, the linear effect of week was significant for all four treatments in 1988 and 1989, but only for treatment four in 1990. The quadratic effect of week was significant for all four treatments for all three years. The cubic effect of week was significant for all four treatments in 1989, but none in 1988 or 1990. The effect of cultivar was significant for treatments one and two in 1988, treatment four in 1989 and treatment three in 1990. The linear week by cultivar interaction was significant for all four treatments in 1988 and for treatment two in 1990. The quadratic week by cultivar interaction was significant only for treatment four in 1989.