Simply, understandable and repeatable by everybody.

I've got some more math that is "simple, understandable, and repeatable" using the data YOU provided us, and working from your flat earth assumption. Your North-South legs are measured from 66° north to 56° north latitude, right? So those legs are 10° long. Averaging them gives a figure of 111.221 km/degree. The lengths of the northern East-West leg and the southern East-West leg can easily be calculated then, since on a flat earth each should be 1/36th of the flat circle of 2*pi*radius (or maybe a little bit less, if the measurement cuts across the circle and takes the shorter 'polygon side' route). Radius at the northern line is (90°-66°) * 111.221 km/degree, or 2669.304 km, and the leg should be 465.88 km. Not 451.33 km. The calculated value is 3.2% larger than your measured value. Let's check the southern leg. (90°-55°) * 111.221 = a radius of 3892.735. Multiply that by 2 pi and divide by 36: 679.41 km, not 621.33, for an error of 58.08 km, an even larger 9.3% error.

Notice how the flat earth error grows the further south we get? This is because the radius of the latitudinal circle grows by the cosine of the angle times the earth's radius, not linearly with the angle. Let's do the math again, using my round earth assumption. Radius at the northern line is cos(66°) * 6371 km, or 2591.32 km, and the leg should be 452.27 km. Much closer to your measured value of 451.33 km, only 0.2% error! Let's check the southern leg. Cos(55°) * 6371 km = a radius of 3654.25. Multiply that by 2 pi and divide by 36: 637.79 km, again much closer to your measured value of 621.33 km, only a 2.6% error. The errors in the round earth method are there because the math was done for a perfectly circular cross section, but in fact the cross section is larger the further south you get (to a point), so the surface distances will measure longer than the calculations. Which is what you found for us.