On a strange observation in the theory of the dimer problem (Q1567277)

Let \(A_n\) denote the number of different ways in which a \(2n \times 2n\) chessboard can be covered with \(2n^2\) dominoes such that each domino covers two adjacent squares. Set \(A_n=2^nB_n^2\) where \(B_n > 0\). The following theorem is proved. The statement \(\mathcal{S}_r\): \[ B_n = \begin{cases} n+1 & \text{if \(n\) is even}\\ (-1)^{(n-1)/2}n & \text{if \(n\) is odd} \end{cases} \pmod{2^r} \] is true for \(r=5\) (thus also for \(r=1, 2, 3, 4\)) and false for \(r=6\) (thus also for \(r >6\)). The authors also obtain that (1) \(B_{n+32}\equiv B_n+32 \pmod{64}\) and (2) \(B_n \pmod{64}\) is periodic in \(n\) with primitive period 64.