bottleneck.move_mean get different result for the same ndarray

Question

import bottleneck as bn

import numpy as np

test = np.full((5,8), 0)

test[1, :]=1

the different result:

result = bn.move_std(test, 3, axis=0)[-1, :]

array([8.60318943e-09, 8.60318943e-09, 8.60318943e-09, 8.60318943e-09,
   8.60318943e-09, 8.60318943e-09, 8.60318943e-09, 8.60318943e-09])