I think the consensus definition of 'major' is a piece that, together with its own royal, can in general force a win against a bare royal of the opponent. Pieces that cannot do that are 'minors'. In orthodox Chess the situation is clear: R and Q are majors, B and N are minors. Accidentally this correlates perfectly with piece values, but this doesn't have to be the case in general. The lowest-valued major I could find on 8x8 is a leaper with only 5 moves (the Deva, fFbrFfWlW, from Maka Dai Dai Shogi), which in a FIDE context probably would not be worth more than 2 Pawns. Minors can be more valuable than a Queen. (Think of a color-bound universal leaper, which can teleport to any dark square.)
It of course depends on how the royal moves: in Knightmate a Rook is a minor. Board size matters too: a non-royal King is a major up to 14x14, but a minor on 15x15. And on a cylinder board the Rook is a minor. The stalemate result also matters: in Xiangqi even a Pawn is a major.
It can also be a bit obscure what 'generally won' means. A piece that moves as Rook in one dimension, and as Dababba in the perpendicular one (e.g. vRsD) has mating potential on any size board, except that on even-sized boards there is a fortress draw when the bare King is on the edge that the piece cannot reach. So it depends on whether you can cut off the King from reaching that edge. With the perpendicular move a Dababbarider (vRsDD) you have even better chances to achieve that, and a relatively small fractions of the possible start positions is draw. (But many more than the 0.5% or so that you typically see in a generally won end-game.) With the weaker majors, such as the range-2 Rook, there are also drawn positions where the R2 is cut off from its King, and 'dynamically trapped' in a corner area by the bare King on the same diagonal. There is then no way to outflank the King; it can always get on the diagonal where the R2 moves to, and will approach and eventually attack it when it doesn't move. So I guess in general there still is something like a continuous spectrum between minor and major.
I think the consensus definition of 'major' is a piece that, together with its own royal, can in general force a win against a bare royal of the opponent. Pieces that cannot do that are 'minors'. In orthodox Chess the situation is clear: R and Q are majors, B and N are minors. Accidentally this correlates perfectly with piece values, but this doesn't have to be the case in general. The lowest-valued major I could find on 8x8 is a leaper with only 5 moves (the Deva, fFbrFfWlW, from Maka Dai Dai Shogi), which in a FIDE context probably would not be worth more than 2 Pawns. Minors can be more valuable than a Queen. (Think of a color-bound universal leaper, which can teleport to any dark square.)
It of course depends on how the royal moves: in Knightmate a Rook is a minor. Board size matters too: a non-royal King is a major up to 14x14, but a minor on 15x15. And on a cylinder board the Rook is a minor. The stalemate result also matters: in Xiangqi even a Pawn is a major.
It can also be a bit obscure what 'generally won' means. A piece that moves as Rook in one dimension, and as Dababba in the perpendicular one (e.g. vRsD) has mating potential on any size board, except that on even-sized boards there is a fortress draw when the bare King is on the edge that the piece cannot reach. So it depends on whether you can cut off the King from reaching that edge. With the perpendicular move a Dababbarider (vRsDD) you have even better chances to achieve that, and a relatively small fractions of the possible start positions is draw. (But many more than the 0.5% or so that you typically see in a generally won end-game.) With the weaker majors, such as the range-2 Rook, there are also drawn positions where the R2 is cut off from its King, and 'dynamically trapped' in a corner area by the bare King on the same diagonal. There is then no way to outflank the King; it can always get on the diagonal where the R2 moves to, and will approach and eventually attack it when it doesn't move. So I guess in general there still is something like a continuous spectrum between minor and major.