Only wheel 2 is pointing along the line of the arc. The others are
pointing slightly away from the arc, so will be 'dragged' around the
arc.
So, for a wheel at an angle to travel in a straight line, some slippage
must occur, but wheels pointing in the same direction cannot follow an
arc without slippage.
Either way there is some slippage. Therefore the least slippage will
probably occur when the skate traces an arc somewhere between the
thoeretical one defined by method 1, and the straight line.
This does not prove that it is impossible to trace an arc on one skate,
just that some slippage must occur if more than 1 wheel is touching the
ground at any one time. Of course, whith a rockered setup, it is quite
possible that only 1 wheel is on the ground, making 'The Non-concentric
Effect' invalid.
Alternatively, other effects may help the skate to turn, including
Wheel/frame deformation - has 2 effects, the wheels can turn to point parallel
to the arc, and the contact point on the wheel can move up, putting it closer
to the arc.
Wheel slip - When the wheel is not on the arc, or is not pointing along the arc
(but still on the floor) it must slip to some extent to follow the arc.
Wheel lift - If one or more wheels is not touching the floor, then this wheel
will not restrict the turning of the skate.
Bearing tolerences - The movements on the bearings will have the same effects
as wheel and frame deformation.
Specially shaped wheels - If, when gliding the center wheels contact the
ground lower down than the front-back wheels. This creates an arc similar to
an ice-skate by bringing the contact points of the wheels closer to the arc
being traced.