As Jack said, rotating a square tube 45° improves its bending resistance considerably (but a PITA to work with, all miter cuts). The off-axis (across the flats) is almost completely immune to bending stress except in a wreck.
IIRC the change is +41.4% to the side length (the stressed axis being the diagonal rather than 1 side).
Where S = length of 1 side, and 1.4142 is 1 ÷ SIN (45°), the new stiffness (assuming equal execution) is:
(S*1.4142)^4 ÷ S^4
Example only; if S = 1":
(1*1.4142)^4 ÷ 1^4
1.4142^4 ÷ 1 = 400% (+300%)
Nice improvement with no weight or material change, just more work!
Note: a smaller change will still occur if the tube is a rectangle, substitute the sine of the diagonal angle in the values.