Reason for the large scale differences in the results for the neutron-antineutron oscillation problem

It is shown that there is a double counting in the standard model of $n\bar {n}$ mixing in the medium, resulting in full cancellation of leading terms. The direct calculation of $n\bar{n}$ transition followed by annihilation is performed. The lower limit for the free-space $n\bar{n}$ oscillation time is $\tau_{n\bar{n}} \sim T_{n\bar{n}}>10^{16}y$, where $T_{n\bar{n}}$ is the lifetime of neutron bound in a nucleus. This limit exceeds the previous one by 16 orders of magnitude.