That's typical, my guess since it's along the property boundary is that it ties into a monumented line that didn't quite match up with the math. So to "fit" found evidence the curve was adjusted to match monuments or an established line. That's my speculation. The plat is giving a tangent slightly off the lot line to facilitate the location of the curve.
So, not to let a sleeping dog lie I computed the rest of that block. The radius points for curve E and F are within 0.03' of each other so they check. I comped curve F along the road and tied in at the 79.12' line and came up with 79.14', the front lot arcs are off by 0.16' not big deal. It looked like the interior lines were radial so I computed all the front lot corners and let the 0.16' fall in the last lot. I then computed from those corners radially to the radius for curve E and intersected that curved line.
The arc on lot 226 computes as 63.83' i thinks thats the 5' bust . And the side line bearing on the plat is N61-29-04W and I computed it as N61-29-20W so not bad.
I guess you could distribute the 0.16' in the 7 lots and recompute rear lot corner intersections for any field comps you make to compare to whats on the actual ground. This would distribute any plat errors and make comparisons relative to each other.
FUN TIMES ! ! !
That plat is awesome, I wish 1960 plats I've worked with were half as good as that one.