“…It seems that the estimates obtained by the path analysis varied in function of genotypes, locations, years and sowing seasons, which would explain the variations in magnitude, sign and order of importance found in the literature (Santos and Vencovsky, 1986;Nienhuis and Singh, 1986;Peternelli et al, 1994;Coimbra et al, 1998aCoimbra et al, , 1998bCoimbra et al, , 1998cCoimbra et al, , 1999bCoimbra et al, , 1999cCoimbra et al, , 2000Kurek et al, 2001;Ribeiro et al, 2001aRibeiro et al, , 2001bCoelho et al, 2002;Furtado et al, 2002).…”