INTRODUCTION Soybean (Glycine max (L.) Merrill) is one of the most important oilseed crop in the world which is grown since ancient times and is one of the oldest grown crop known to mankind (Peerzada et al., 2014). The North-eastern China region is believed to be the primary centre of origin of this crop. Introduction of soybean to the Indian subcontinent dates back to 1000 AD (Orf et al., 1980). Soybean (2n=40) is botanically placed under the family Fabaceae. Soybean plant favours an optimum temperature of 30°C for its normal growth and development while an optimum temperature for seedlings emergence is found to be 25–33°C. It also fixes the atmospheric nitrogen in soil through symbiotic relationship with biological nitrogen fixing bacteria namely Bradyrhizobium japonicum and Bradyrhizobium diazoefficiens (Siqueira et al., 2014). Soybean is an important multi-purpose leguminous crop known for its highly valued protein and oil, and its use in food, feed and industrial applications. Hence it is popularly known as the “Golden Bean” of the 20th century. Soybean has been typically used for preparation of traditional and as well as value added food products such as tofu, shoyu, okara, miso, natto, soy milk, soya sauce, tempeh, soy sprouts and soymilk yoghurt etc. which are being consumed on a regular basis in China, Japan, Korea and Southeast Asia (Muredzi, 2013). Soybean seeds have the highest protein content (30-45%) of all food crops and also contains a considerable oil content (15-24%) comprising high percent of unsaturated fatty acids (Akram et al., 2011). The oil contains 85% unsaturated fatty acids which is free from cholesterol, along with ample mineral elements which is very desirable for human diet (Antalina, 2000). A rapid increase in area under soybean has been observed which was just 0.03 million ha in 1970 to 12.05 million ha in 2021 (MOA and FW, GOI, 2022) and mean annual productivity has been increased from 430 kg per ha in 1970 to 1280 kg per ha with the production of 13.58 m tons in 2021 (SOPA, 2022). However, world productivity level has not yet been achieved in India. The major constraints for low productivity in India are low varietal stability and narrow genetic base. Therefore, soybean productivity can be enhanced through creating genetic variability and employing effective selection methods to recover superior transgressive segregation lines in advance generation. Hence, the present investigation was carried out to find out superior crosses and the best method for selection in advance breeding lines through per se performance and inter correlation analysis, respectively. MATERIALS AND METHODS The present investigation was conducted during kharif 2019 at N. E. Borlaug Crop Research Centre, Govind Ballabh Pant University of Agriculture & Technology, Pantnagar, Udham Singh Nagar, Uttarakhand. The experimental material in the present study consisted of nine F6 crosses which were planted in compact family block design with three replications in kharif 2019 (each cross containing 5 lines of high yielding, 5 lines of low yielding and 5 lines of bulk). Bulk lines of each cross were handled through bulk method since F3 generation. 15 rows (5 progeny rows of high yielding, 5 progeny rows of low yielding and 5 rows of bulk progenies) of 4 meter with 45 cm row to row spacing for each F6 cross was planted in Kharif 2019, while two rows of parents were planted alongside the crosses. Plant to plant distance was maintained at 5 to 7 cm after thinning. Data was taken from 5 randomly selected competitive plants from each row of high yielding, low yielding progeny and parent of each cross. But 5 randomly competitive plants from 5 bulk rows of each cross were selected for recording data. Data was recorded from each line of all crosses for the most promising and variable character “seed yield per plant”. The per se performance was calculated by taking mean seed yield per plant. The phenotypic inter correlations between all lines within same family as well as among the families were estimated according to the method given by Searle (1961). RESULTS AND DISCUSION Per se performance indicates the performance of plants by “itself”. Therefore, performances of various crosses were selected in advance generation based on their mean seed yield per plant which is presented in Table 4. Among high yielding lines, the highest per se performance was shown by C7 (18.04 g) followed by C4 (17.13 g), C3 (13.11 g), C5 (10.81 g), C1 (10.28 g), C8 (6.94 g), C2 (6.17 g), C6 (5.27 g) and C9 (1.84). Among low yielding lines, the highest per se performance was exhibited by C7(14.10 g) followed by C3(11.13 g), C5(10.54 g), C1(9.26 g), C4(9.01 g), C6(7.62 g), C8(6.83 g), C2(6.47 g) and C9(5.63g) whereas among bulk lines, the highest per se performance was shown by the C7(33.66 g) followed by C5(23.26 g), C6(19.24 g), C3(18.79 g), C4(18.75 g), C9(16.56 g), C2(11.07 g), C8(7.83 g) and C1(7.1 g). The mean seed yield of three lines of each cross was added and overall per se performance was estimated. The study of overall per se performance stated that C7 was the most promising cross followed by C4, C5, C3 and C6 whereas C9 was the poorest performing cross. Sharma et al. (2012) also reported the similar result. In the present investigation, inter correlation among high yielding lines, low yielding lines and bulk lines within same family at phenotypic level was estimated for the trait “seed yield per plant (g)” to screen out the best method of selection in above promising crosses. It was indicated that there were significant and positive correlation found among different lines in different crosses which revealed that method of selection through inter correlation study played an important role in soybean in advance generation. Inter correlation analysis within same family in F6 generation for seed yield per plant at phenotypic level exhibited that there was significant and positive correlation was present between high yielding lines and bulk lines viz. HB(0.77) in C1 (JS20-29 × JS20-55) which indicated that seed yield variation between these two line was really significant. The mean seed yield of high yielding lines was 10.28 g which was higher than bulk lines (7.1 g) which showed that pedigree method in high yielding lines was performing better than bulk method. In C2 (PS 1584 × JS 20-41), high yielding lines was significant and positively correlated with low yielding lines for the trait seed yield per plant (0.54). Similarly, there was negative and significant correlation present between low yielding-bulk lines (-0.84). As mean seed yield of bulk lines was higher than high and low yielding lines thus bulk method was performing better than pedigree method. There was positive and significant correlation present between high yielding-bulk line in C3 (JS 20-69 × JS 20-59) and mean seed yield of bulk lines was more than other lines; hence bulk method is better than pedigree method. In C4 (PS 1584 × JS 20-69), a negative and significant correlation was present between high yielding- bulk lines (-0.68) and bulk lines had more mean seed yield per plant which indicated the predominance of bulk method over pedigree method. Similarly a positive significant correlation was found in high-low yielding lines (0.67) in C5 (JS 20-64 × JS 20-54) and correlation between high yielding and bulk lines was negative. As bulk lines have high mean seed yield thus bulk method was better than pedigree method. The inter correlation study in C6 (PS 1042 × PS1347) showed that high yielding and low yielding lines were negatively and significantly correlated with each other whereas low yielding and bulk lines were positively and significantly correlated. As bulk lines showed a negative correlation with high yielding lines and exhibited higher mean seed yield than other two lines thus bulk method was performing better than pedigree method. There was positive and significant correlation 0.92 and 0.61 were exhibited by high yielding-bulk and low yielding-bulk line, respectively in C7(PS 1583 × Bragg). Bulk line had high mean seed yield than other two lines so bulk method was predominant than pedigree. All the lines in C8(PS 1583 × JS 20-29) didn’t show any significant correlation. Therefore, mean seed yield did not significantly differ among these lines of C8. In C9 (RVS 2000-1 × PS 1092), there was significant and negative correlation between low yielding and bulk lines (-0.56). Bulk lines also showed a negative correlation with high yielding lines. As bulk lines had more mean seed yield thus bulk method was performing superior than pedigree method. The above inter correlation analysis values within same family for the trait seed yield per plant is shown in Table 1. The above inter correlation analysis within same family and overall per se performance study exhibited that C 1 was better performing through pedigree method of selection whereas in C2, both pedigree and bulk lines were equally performing. The bulk method of selection was the best for C7 followed by C4, C5, C3 and C6. In the present investigation the inter correlation between high yielding- high yielding, low yielding-low yielding and bulk-bulk lines among different family at phenotypic level was carried out for the traits seed yield per plant. It was revealed that the significant correlation was present among different crosses in different lines which helped to find out the best cross among high yielding, low yielding and bulk lines through direct selection. In F6 generation, inter correlation analysis among different family at phenotypic level for the trait seed yield per plant revealed that among high yielding lines, C1 exhibited positive and significant correlation with C2 (0.49), C3 (0.45), C5 (0.51) and C6 (0.39) and also showed a negative and significant correlation with C4(-0.62). The positive and significant correlations exhibited by C2 with C5 (0.54), C6(0.83), C7 (0.41) and a negative significant correlation with C9 (-0.41). Similarly, C3 showed positive and significant correlation with C4 (0.81), C7 (0.5) and negative significant correlation with C8 (-0.9) and C9 (-0.66). C4 exhibited a positive significant correlation with C7 (0.49) and negative significant correlation with C5 (-0.71) and C8 (-0.74). There was a positive significant correlation between C5C6 (0.43) and C5C8 (0.53) and a negative significant correlation between C5C9(0.61). C7 showed a negative significant correlation with C8 (-0.81). From the above correlation analysis and mean seed yield per plant study, it was revealed that among all high yielding lines, C7 (18.04g) was the best cross followed by C4 (17.13g) and C3 (13.11g). Among low yielding lines, the positive and significant correlation at phenotypic level exhibited by crosses viz. C1C6(0.9), C4C8(0.79), C6C9(0.77), C2C5(0.74), C5C8 (0.7), C1C9(0.65), C3C4(0.46), C3C8(0.46), C7C9(0.44) and C2C8(0.4). Similarly negative and significant correlations at phenotypic level exhibited by the crosses viz. C5C7(-0.83), C1C3(-0.74), C2C7(-0.74), C3C6(-0.69), C4C6(-0.55), C2C9(-0.46) and C7C8(-0.4). Among all low yielding lines, C7 and C3 both were best performing crosses as they had high mean seed yield per plant (14.1g) and they were not significantly correlated with each other. Among bulk lines, the positive and significant correlation at phenotypic level exhibited by the crosses viz. C2C5 (0.99), C2C3 (0.97), C1C9 (0.95), C6C8 (0.95), C3C5 (0.94), C5C8 (0.93), C2C8 (0.9), C7C9 (0.89), C5C6 (0.77), C3C8 (0.75), C1C2 (0.74), C1C7 (0.71), C2C6 (0.71) and C3C6 (0.5). The negative and significant correlation at phenotypic level exhibited by the crosses viz. C1C6 (-0.99), C1C8 (-0.96), C6C9 (-0.96), C8C9 (-0.83), C1C5 (-0.8), C6C7 (-0.74), C4C7 (-0.71), C2C4 (-0.6), C4C5 (-0.6), C5C9 (-0.57), C1C2 (-0.56), C2C9 (-0.5) and C7C8 (-0.49). Among bulk lines, C7 and C5 exhibited high mean seed yield per plant (33.66g ) and did not significantly correlated hence these two cross emerged as best crosses among all others followed by C6 (19.24g). The above inter correlation study among different lines among different family is presented in Table 2. The inter correlation analysis and mean seed yield per plant study revealed that among all high yielding lines C7, C4 and C3 were best performing crosses whereas C7 and C3 were better performing crosses in low yielding lines. Similarly in bulk lines, C7 and C5 both were equally better performing crosses followed by C6 which are given in Table 3. Recent study showed the comparison between bulk method and pedigree method of handling of soybean lines. In advance generation (F6), the bulk lines which were handled since F3 exhibited superior performance for yield contributing characters especially seed yield per plant than those of pedigree lines. Hence, it not only gives us a great scope for natural selection of many traits including disease resistant but also enables us for screening large segregating population in advance generation.