Pooled analysis of variance revealed that environments and entries (parents, F1's and parents v/s F1's) were highly significant differences for all the characters. Environment x entries was also significant for all the characters. Environments x parent were also significant for all the characters except plant height, primary branches per plant and seed yield per plant, while environments x F1's were also significant for all the characters except plant height and oil content and environments x parents v/s F1's interaction significant for all the characters. Entries mean sum of squares were partitioned into varieties and overall heterosis. Varieties (vi) were significant for all the characters in all the environments except test weight in environment-IV.
Over all heterosis was partitioned into average (?), variety (hi) and specific (sii') heterosis. Average heterosis (?) was significant for all the characters in all the environments except days to maturity, siliquae per plant and oil content in environment-I, while days to maturity and oil content in the environment-II and days to maturity, primary branches per plant, number of seeds per siliqua, test weight and oil content in environment-III, while days to maturity and siliquae per plant in environment-IV. Varietal heterosis (hi) was significant for all the characters in all the environments. Specific heterosis (Sii') was also significant for all characters in all the environments except days to flowering in the environment I and days to flowering and days to maturity in environment-III. Specific heterosis (sii') component accounted for more than 50% to the overall heterosis (hii') sum of squares. This indicated the presence of additive, dominance and epistatic components in the genetic control of the characters in the present investigation. Among the parents (varieties) RTM-1359 and RTM-1375 in all the environments, RTM-314 in the environment-II and III were found to be superior based on vi effects and per se performance not only for seed yield per plant but for some of the other yield traits studied. The parents worth considered on the basis of hi values were RTM-1375 in environment-I, RTM-314 in environment-II, while RTM-1415 and RTM-1359 in the environment-III and RTM-314, RTM-1359 in environment-IV. It was therefore recommended to include these parents in hybridization programme to improve seed yield and seed quality. In the present investigation, highly heterotic crosses RTM-1351 x RTM-314, RTM-314 x RTM-1359, RTM-1415 X RTM-2002 and T-27 x RTM-2002 in all the four environments, were desirable as they had high seed yield per plant, high sii' values, with high heterobeltiosis and economic heterosis not only for seed yield per plant but for some than the other yield traits studied.