Relativity and the low-energy<mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" display="inline"><mml:mrow><mml:mi mathvariant="italic">nd</mml:mi></mml:mrow></mml:math><mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" display="inline"><mml:mrow><mml:msub><mml:mi>A</mml:mi><mml:mrow><mml:mi>y</mml:mi></mml:mrow></mml:msub></mml:mrow></mml:math>puzzle

We solve the Faddeev equation in an exactly Poincar\'e invariant formulation of the three-nucleon problem. The dynamical input is a relativistic nucleon-nucleon ($\mathit{NN}$) interaction that is exactly on-shell equivalent to the high-precision CD Bonn $\mathit{NN}$ interaction. $S$-matrix cluster properties dictate how the two-body dynamics is embedded in the three-nucleon mass operator (rest Hamiltonian). We find that for neutron laboratory energies above $\ensuremath{\approx}20$ MeV relativistic effects on ${A}_{y}$ are negligible. For energies below $\ensuremath{\approx}20$ MeV dynamical effects lower the nucleon analyzing power maximum slightly by $\ensuremath{\approx}2%$ and Wigner rotations lower it further up to $\ensuremath{\approx}10%$, thereby increasing disagreement between data and theory. This indicates that three-nucleon forces (3NF) must provide an even larger increase of the ${A}_{y}$ maximum than expected up to now.