. Residual Effect of Various Herbicides on Japanese Millet '

FIGURE 1 Echinochloa frumentacea control over three seasons. These trials were conducted under a wide variety of soil and climatic conditions. A randomised complete block designwith four replicationswas used. The trials werela.id out in such a way that each plot could be compared WIth an adjacent control strip. A plot size of 20 m' was used. ~er­ bicides were applied pre-emergence by means of specially designed small plot sprayers with a 4 m b~om. fitted with six 9502 or 9504 flat fan spray nozzles, dehvenng a spray mixture of 200 to 400 l/ha at a pressure of 200 kPa. The herbicides were applied at rates applicable to the particular soil type. The application rates of metolachlor and acetochlor were 1,5 kg ailha, alachlor 1,92 kg ai/ha, and metazachlor 0,56 kg ai/ha. In all the trials the application of the herbicides was followed by 15 mm irrigation. Two rows of Japanese millet were planted as the indicator plant, In each season the plantings were made at the following intervals: 1985/86 season, three, five and six weeks after application; 1986/87 and 1987/88 seasons, at the time of application and two four six and eight weeks later. In order to calculatethe me~n co~trol over all seasons, the mean of the first planting of the 1985/86 season (three weeks after application) was added to the mean of the second planting (two weeks after application) of the 1986/87 and 1987/88 seasons. The mean of the secondplanting(five weeks afterapplication) wasadded to the mean of the third planting (four weeks after application) of the 1986/87 and 1987/88 seasons. Each planting was followed by 15 mm irrigation. The percentage control was assessed visually by comparing the treated plots with the untreated control strips, two, four, six and ten weeks after planting. . A separate analysis of variance was carried out for. ~ach year. Localities, replicates wi~~in localities and herbicides were the main effects. In addition, all the tnals were combined and an overall analysis of variance carried out. Yearl locality combinations, replicateswithin these combinations and herbicides were the main effects. For the abovementioned, each planting was analysed separately. The Tukey test (P=0,05) was u~ed to com~ar~ treatment means (Steel and Terrie'). Coefficients of vanation were calculated ov~r localities for herbicides individually when locahty x herbicide interactions were significant.